National Academies Press: OpenBook
« Previous: 4 Dust
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 36
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 37
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 38
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 39
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 40
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 41
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 42
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 43
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 44
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 45
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 46
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 47
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 48
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 49
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 50
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 51
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 52
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 53
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 54
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 55
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 56
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 57
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 58
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 59
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 60
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 61
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 62
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 63
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 64
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 65
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 66
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 67
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 68
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 69
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 70
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 71
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 72
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 73
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 74
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 75
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 76
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 77
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 78
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 79
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 80
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 81
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 82
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 83
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 84
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 85
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 86
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 87
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 88
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 89
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 90
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 91
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 92
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 93
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 94
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 95
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 96
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 97
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 98
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 99
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 100
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 101
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 102
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 103
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 104
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 105
Suggested Citation:"5 Fires." National Research Council. 1985. The Effects on the Atmosphere of a Major Nuclear Exchange. Washington, DC: The National Academies Press. doi: 10.17226/540.
×
Page 106

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

S. . fores OVERVIEW* It is clear that nuclear explosions can ignite large-scale fires (Broido, 1960. In addition, it has been estimated that the smoke emissions from nuclear-initiated fires could produce major atmospheric perturbations (Lewis, 1979; Crutzen and Birks, 1982; Turco et al., 1983a,b). Only two nuclear explosions have ever occurred over populated areas (Hiroshima. Auoust 6. 1945. and Nagasaki. Auoust 9 . _ _ _ _ _ ~ ~ , , _ , _ _ , ~ , _ _ , , 1945); In each case, a clty-slzea conflagration resulted. At Hiroshima, a ~12-kt weapon caused a mass fire over an area of ~13 km2, essentially the entire central city (Ishikawa and Swain, 1981~. At Nagasaki, where high terrain shadowed large regions of the city from direct irradiation by bomb light, a ~20-kt device burned ~7 km2 (Ishikawa and Swain, 1981~. It is difficult to extrapolate the effects of these two isolated events, which involved <40-kt total yield, to the possible effects of a global nuclear exchange involving 6500 Mt. Nevertheless, a logical sequence of steps can be taken to obtain estimates of the areal extent and particulate emissions of fires initiated in a full-scale nuclear war: 1. Review historical fire experience to assess the probability of ignition and spread of large fires. 2. Define the effectiveness of nuclear explosions for initiating fires in urban and forest settings. *In the text, the following symbols are used: a, approximately equal to; ~, of the order of; < , less than or of the order of; > , greater than or of the order of. tFor the purposes of this report, large-scale fires can be classified as "mass fires, n in which many individual fires burn simultaneously over a large area, "conflagrations, n in which the fire is most intense along a line of propagation, "firestorms, n in which the entire area of the fire burns intensely and strong winds blow inward from all directions, and "fire whirls, n in which a firestorm plume develops an unusually strong vorticity. 36

37 3. Determine the burdens and distributions of combustible materials around potential nuclear targets. 4. Evaluate data on the quantity and physical properties of smoke generated by common fuels. 5. Consider mass fire dynamics to determine the likely heights and rates of injection of the smoke. 6. Describe a scenario for the locations, yields, and heights of nuclear detonations (See Chapter 3, The Baseline Nuclear Exchangers. 7. Combine the foregoing information to estimate the total quantity and optical characteristics of nuclear war smoke emissions. These topics are discussed in subsequent sections of this chapter. On the basis of such an analysis, an approximate equation can be written that emphasizes the important factors that enter into the estimation process, E = YfA0m0fbe x 101°, where E is the total smoke emission (in grams), Yf is the total explosion yield (in megatons) in air bursts that effectively ignite fires, AD is the average area ignited by each megaton of yield (in square kilometers per megaton), my is the average loading of flammable materials (in grams per square centimeter), fb is the fraction of my burned, and £ is the mean smoke emission factor (grams of smoke per gram of material burned). The factor of 101° converts square kilometers (Ao) to square centimeters. The key parameter values that apply to the baseline nuclear war scenario are given in Table 5.1. The total smoke emission calculated for the baseline case is ~180 Tg (1 Tg = 1012 g ~ 106 metric tons), or ~0.7 g/m2 averaged over the northern hemisphere. Since the specific extinction (scattering plus absorption) coefficient of many smokes at visible wavelengths is ~5.5 m2/g, the hemispherical average optical depth* in this case is ~4. Of course, if the smoke were confined to the northern mid-latitude zone, the optical depth would be ~2 to 3 times larger, or ~8 to 12. A more detailed discussion of these estimates follows. The optical and climatic effects of the smoke are discussed in Chapter 7. PRESENT-DAY SMOKE EMISSION AND REMOVAL It is estimated that the current global smoke emission to the atmosphere is ~200 Tg/yr (Seller and Crutzen, 1980; Turco et al., 1983a,c). The graphitic carbon fraction is about 5 to 10 percent by *The optical depth is a dimensionless quantity that determines the light transmission properties of a layer of gas or aerosols. If the layer has an optical depth T. e ~ is the fraction of a beam of light perpendicularly incident on the layer that suffers no scattering or absorption in passing through the layer. The total light transmitted consists of the direct light plus a scattered (diffuse) component.

38 TABLE 5.1 Baseline Nuclear War Fire and Smoke Parametersa Parameterb Urban Fires Forest Fires Yf (Mt) 10001000 AD tkm2/Mt)250250 me (g/cm2)42 fb g/g)C 0.75 0.20 0.02 0.03 aExcursions from the baseline case, and uncertainties in the baseline parameters, are discussed in the text. bYf is the effective ignition yield in megatons, An is the average ignition area per megaton, my is the burden of combustibles per unit area, fb is the fraction of the combustibles burned, and ~ is the net smoke emission factor per unit of fuel, assuming in the case of urban fires that 50 percent of the smoke is promptly scavenged and removed from the plumes mainly as black rain. n CThe smoke consists of 20 percent graphitic carbon (soot) by mass, and 80 percent transparent oily compounds. NOTE: Urban fire smoke emission: Eu = 150 x loll g Forest fire smoke emission: Ef = 30 x 1012 g Total smoke emission: Et = 180 x 1012 g mass. The primary sources of smoke are agricultural burning, fossil fuel combustion, and wildfires. The important characteristics of background smoke emissions that distinguish them from "nuclear" fire emissions are as follows: 1. The smoke emission factors are low in relation to the quantity of fuel burned, because most of the burning takes place under controlled conditions. 2. The overall graphitic carbon component is low, because most of the smoke is generated during the prescribed combustion of natural cellulosic materials. 3. Almost all of the smoke is injected into the lowest 1 km of the atmosphere, because the sources are small in horizontal scale and/or total power. 4. The smoke emissions occur in diverse locations throughout the course of a year, which prevents significant concentrations from building up. 5. The average atmospheric lifetime of the smoke is < 10 days (Ogren, 19821.

39 As a result of these factors, the average background concentration of airborne graphitic carbon is typically only ~ 0.1 ug/m3, and its integrated vertical absorption optical depth is <0.01 (Charlson and Ogr en, 1982; Turco et al., 1983c). Over a period of about 1 month, background smoke emissions would be negligible in comparison with the estimated smoke emissions of a nuclear war (Turco et al., 1983a,b; Crutzen et al., 1984~. Removal of smoke and soot from the atmosphere occurs mainly through precipitation scavenging. Smoke particles have sizes of about 0.1- to 0.5-pm radius, at which sedimentation is negligible and dry deposition is very inefficient (Slinn, 1977; Sehmel, 1980~. In the background atmosphere, soot is usually found as a minor component of hydroscopic sulfate aerosols. This suggests removal by efficient scavenging of the hydroscopic aerosols in and below clouds (Radke et al., 1980; Ogren, 1982; Turco et al., 1983c). The arctic haze that forms in winter and spring is known to contain soot (Rosen and Novakov, 1983~. The haze is (relatively) highly absorbing because of the soot (Patterson et al., 19821. The seasonal conditions that lead to the formation of the winter polar vortex create a stable air mass with low precipitation in which carbon emissions produced by combustion can remain suspended for several months. This demonstrates that under some meteorological conditions, particularly with the suppression of precipitation, smoke and soot can have an extended atmospheric lifetime. Generally speaking, it is expected that smoke from nuclear-initiated fires would have a longer atmospheric lifetime than background smoke (notwithstanding prompt scavenging in the fire plumes), because of its greater heights of injection. This point is expanded in subsequent sections. HISTORICAL FIRE EXPERIENCE Human experience with mass fires and firestorms includes urban conflagrations triggered by natural disasters (e.g., earthquakes), wartime city fires initiated by incendiary and nuclear bombing, massive wildfires and forest fires, and field experiments with large-scale fuel beds (Carrier et al., 1982~. Although few of these experiences are directly applicable to the nuclear war problem, all contribute to a general understanding of the properties and behavior of large-scale fires. Earthquakes Earthquakes have started urban conflagrations by breaking gas lines, exposing stored fuels, shorting electrical circuits, breaching open fires, and hampering effective firefighting. Particularly striking examples of large fires induced by earthquakes occurred in San Francisco in 1906 and Tokyo in 1923. A nuclear blast wave would have

40 similar impact and, in combination with the thermal light pulse, would represent a much greater fire threat than an earthquake. World War II The World War II saturation bombing of German and Japanese cities provided ample evidence that mass fires can be readily ignited in urban settings. The nuclear explosions over Hiroshima and Nagasaki are discussed later. The conventional bombing of cities such as Hamburg, Dresden, Darmstadt, and Tokyo produced intense fires over many square kilometers and, in some instances, triggered firestorms. From anecdotal evidence, it is known that thick, dark plumes rose from these fires to altitudes of 6 to 12 km. Within the fire zones, almost all the buildings were gutted and all combustible materials consumed. Such experiences show that, when many simultaneous fire ignitions occur among closely spaced structures and firefighting capability is suppressed, mass fires are likely to develop. Occasionally, massive urban conflagrations, such as the Great Chicago Fire of 1871, are touched off by single ignitions (Kerr, 1971~. Although such fires are not typical, they are symptomatic of the hazardous fire conditions that exist in many crowded urban centers. Forest Fires Plummer (1912), Ayers (1965), and F.E. Fendell (in Appendix 5-1), among others, have reviewed the largest forest fires of the past 160 years in which areas up to 20,000 km2 were blackened. The conditions under which these catastrophic fires developed included long drought, low humidity, and high winds (e.g., Plummer, 1912~. Clearly, such conditions are not common over large areas of the northern hemisphere during most of the year (Chandler et al., 19631. However, for the analysis of nuclear-induced fires, three general types of fire danger conditions should be distinguished: (I) fires are difficult to ignite and do not spread if ignited; (II) fires are readily ignited, but their spread is limited by factors such as humidity, moisture, topography, winds, and firebreaks; and (III) fires readily ignite and spread uncontrollably over large areas. Historical catastrophic forest fires are exclusively of type III. By contrast, most nuclear forest fires would probably be of type II. Historical fires are characterized by a limited number of ignition points, perhaps one ignition for each 50 to 500 km2 burned (Ayers, 1965~. Nuclear explosions, by contrast, can ignite forest debris instantly over a large area, with numerous ignition points developing into moderate size fires (although the probability of extensive fire spread outside of the original burning zone would be much lower--see below). The great Tunguska meteor, which fell over Siberia on June 30, 1908, provides a very rough indication of the effects that might be produced by a high-yield nuclear explosion over a forest. The Tunguska

41 event was equivalent, in terms of the blast wave, to a ~10-Mt detonation at 8-km altitude (Krinov, 1966~. (As noted below, high-yield nuclear bursts have smaller incendiary efficiencies than low-yield bursts.) Roughly 16QO km2 of Siberian forest was flattened. Eyewitness accounts describe burning falling trees" and widespread fires. A series of Russian scientific expeditions to the fall site concluded that several major fires had broken out in the central zone of devastation and burned for 5 days. From the description of the charred remains, it appears that bark and many small branches were stripped from the trees and burned, to an extent not usually observed in natural fires of that area. Experimental Fires Experimental large-scale fires have been used to study fire development and plume dynamics. Among these experiments are the Flambeau series (Martin, 1974; Palmer, 1981), the Euroka fires (Williams et al., 1970), and the Meteotron events (Desserts, 1962; Church et al., 19803. However, because the extent of these fires was only about 103 to 105 m2, extrapolation of the results to city-size fires is difficult. Of particular interest here is the height of the smoke plume in a large fire. In the experiments noted above, the plume aspect ratio (i.e., the plume height divided by the fire diameter) was always >>1, and the plumes often formed vortices penetrating to heights >1 km. (The plume aspect ratio cannot be simply scaled to larger fires. The dependence of plume height on fire size and intensity, and extrapolations to city-sized fires are discussed in later section.) The Flambeau experiments also led to the definition of a set of conditions for firestorm genesis that has been widely accepted (FEMA, 19821. The conditions include a fuel loading of >4 g/cm2, a building density of >20 to 30 percent, a fire area of >3 km2, initial fires in >20 percent of the buildings, and ambient winds of <10 km/in (Baldwin, 1968; Martin, 1974~. However, these conditions are still controversial, as they have never been tested on an appropriate scale. Moreover, in view of the atmospheric effects being considered here, it is not clear that firestorms and very intense mass fires need to be differentiated, except perhaps to refine the estimation of smoke injection altitudes (see below}. IGNITION OF NUCLEAR FIRES Thermal Phenomena In a nuclear air burst at low altitude (<10 km), about 30 to 40 percent of the energy is released as an intense pulse of visible light; about 45 to 55 percent of the energy is converted to blast pressure waves; and about 15 percent is contained in prompt and delayed nuclear radiation (Glasstone and Dolan, 1977, hereafter GD77~. Most of the

100 N - C~ - LL ~ 10 CD o X LL Ad U] N 0. \ ~ ~ \ 2 ~2 _ 1 1 1 1 , t1 1 \1 1 i 1 ~1 1\ 5 10 20 50 100 HORIZONTAL DISTANCE (km) FIGURE 5.1 Maximum radiant exposures versus ground range from a 1-Mt air burst (detonated below several kilometers altitude) as a function of the ground level visibility. The radiant exposures scale roughly with the yield in megatons. (From Kerr et al., 1971) bomb light is emitted within a few seconds for megaton yield explosions, and in less than a second for kiloton-size bursts (GD777. For a 1-Mt low air burst, Figure 5.1 shows the thermal fluences (in calories per square centimeter incident on a surface normal to the line-of-sight through the burst point) as a function of distance from ground zero, and for various atmospheric Risibilities. With a 1-Mt explosion and normal Risibilities (>10 km), the 20-cal/cm2 thermal fluence contour lies about 7 km from the explosion hypocenter, versus 9 km in a perfectly transparent atmosphere. With a 100-kt explosion, atmospheric transmission, for Risibilities of >5 km, has little effect on radiant exposures where fluences exceed 20 cal/cm2. Lower Risibilities restrict the range at which nuclear thermal effects are important. Oblique incidence of the bomb light on exposed surfaces also reduces the effective fluence. On the other hand, cloud and surface reflections enhance the radiant fluxes in localized regions.

43 As sunlight, focused by a lens, can ignite flammable materials, so can the thermal emissions of a nuclear explosion (Glasstone, 1957; Miller, 19621. Ignition data obtained during atmospheric nuclear test detonations and by laboratory experimentation are summarized in Table 5.2. At a specific thermal fluence, small nuclear explosions are generally more efficient at igniting fires than large explosions because the thermal pulse has a shorter duration and larger peak intensity (in addition, there is a lower probability of significant atmospheric attenuation over the shorter ranges involved). Newspaper, brown paper, cotton cloth, and dried plant material can be ignited by 10 cal/cm2 from a <1-Mt explosion. The perimeter of the Hiroshima fire zone roughly coincided with the 10-cal/cm2 contour. At Nagasaki, in directions unobscured by hills, the conflagration zone also extended roughly to the 10-cal/cm2 limit. In the application of nuclear weapons against ~soft. targets (e.g., urban and industrial targets), peak overpressures* of >5 psi (pounds per square inch) are often used to define the zone of assured destruction (GD77~. The 5-psi contour circumscribes an area of ~1.4 km2/kt for a 1-kt explosion (at the optimum height of burst), ~0.30 km2/kt for a 100-kt explosion, and ~0.14 km2/kt for a 1-Mt explosion. The corresponding areas enclosed within the 20-cal/cm2 thermal irradiance contours (GD77) are ~0.30 km2/kt, ~0.30 km2/kt, and ~0.25 km2/kt, respectively (in the 1-Mt case, the atmospheric visibility is assumed to be 20 km). In estimating the potential fire areas for nuclear air bursts, the committee has chosen an average ignition area of 0.25 km2/kt (250 km2/Mt) for individual explosions, which is roughly consistent with 5-psi overpressures and 20 cal/cm2 thermal fluences at the limits of the ignition region, under normal conditions of atmospheric transmission. These areas are quite conservative in relation to the areas burned at Hiroshima (~1 km2/kt) and Nagasaki (~0.35 km2/kt). The question of overlap of ignition zones for closely spaced detonations, and the total potential fire area in a full exchange, are discussed in a separate section of this chapter. Close to the hypocenter of a nuclear explosion, the thermal energies are much larder than 20 cal/cm2. Within the 30-cal/cm2 contour (about 150 km for a 1-Mt explosion), substantial quantities of natural and synthetic organic and cellulosic materials would be instantly pyrolized, and the combustible vapors ignited in a massive "flashover" fire. The rising fireball would then draw the flames and smoke toward the stem of the nuclear cloud, establishing the conditions for accelerated burning and, in some cases, the core of an incipient firestorm. For surface and subsurface nuclear detonations, the potential thermal effects are greatly reduced (although the dust and prompt radioactive fallout effects are increased). The bomb light from a *The term "overpressure" refers to the incremental static pressure above ambient atmospheric pressure (about 14.7 pounds per square inch at sea level) caused by the passage of the explosion wave.

44 TABLE 5.2 Approximate Radiant Exposures for Ignition of Various Flammable Materials for Low Air Bursts Radiation Exposurea (cal/cm2) Effect on35 1.4 20 Material Color Materialkt Mt Mt Household Tinder Materials Newspaper, shredded Ignites4 6 11 Newspaper, dark Ignites5 7 12 picture area Newspaper, printed Ignites6 ~15 text area Crepe paper Green Ignites6 9 16 Kraft paper Tan Ignites10 13 20 Bristol board, 3 ply Dark Ignites16 20 40 Kraft paper carton, Brown Ignites16 20 40 used (flat side) New bond typing paper White Ignites24b 30b 50b Cotton rags Black Ignites10 15 20 Rayon rags Black Ignites9 14 21 Cotton str ing Gray IgnitesLob 15b 21 b scrubbing mop (used) Cotton string Cream Igniteslob lob 26b scrubbing mop (weathered) Paper book matches, Ignites11b 14b 2ob blue head exposed Excelsior, ponderosa Light Ignites__c 23b 23b pine yellow Outdoor Tinder Materialsd Dry rotted wood Ignites4b 6b 8b punk (fir) Deciduous leaves Ignites4 6 8 (beech) Fine grass (cheat) Ignites5 8 10 Coarse grass (sedge) Ignites6 9 11 Pine needles, brown Ignites10 16 21 (ponderosa)

45 TABLE 5.2 (continued) Radiation Explosurea (cal/cm2 ~ Effect on 35 1.4 20 Material Color Material kt Mt Mt Construction Materials Roll roofing, mineral surface Ignites c >34 >116 Roll roofing, smooth Ignites _ c 3077 surface Plywood, Douglas fir Flaming 9 1620 during exposure Rubber, pale latex Ignites 50 80110 Rubber, black Ignites 10 2025 Other Materials ~- Aluminum aircraft Blisters 15 3040 skin (0.020 in. thick) coated with 0.002 in. of standard white aircraft paint Cotton canvas sandbags, dry filled Coral sand Siliceous sand Failure 10 18 32 Explodes (popcorning) Explodes (popcorning) 15 27 47 11 19 35 aRadiant exposures for the indicated responses (except values marked with a superscript b, see footnote b) are estimated to be valid to +25 percent under standard laboratory conditions. Under typical field conditions, the values are estimated to be valid within +50 percent with a greater likelihood of higher rather than lower values. bIgnition levels are estimated to be valid within +50 percent under laboratory conditions and within +100 percent under field conditions. CData not available or appropriate scaling not known. dRadiant exposures for ignition of these substances are highly dependent on the moisture content.

46 surface detonation is more effectively shadowed by buildings, terrain, and other obstructions than is the light from an air burst (Miller 19621. The crater ejecta may also cover nearby fuel and smother incipient fires. In a subsurface explosion (where an armored penetrating warhead is used) the thermal pulse is substantially attenuated (GD77~. Moreover, the base surge (caused by ejected material falling back upon the crater) could snuff out small fires and cover the fuel near the explosion site. Nevertheless, in a surface burst, it is still likely that primary thermal (and in cities, secondary blast-induced) fires would occur out to the -2-psi overpressure contour (i.e., over an area of about 150 km2 for a 1-Mt detonation; GD77~. In buried explosions the situation is more complicated because both ground shock and air blast could contribute to secondary fire ignitions in cities. In any case, the present baseline scenario specifies air bursts against all urban and industrial targets, with only 30 percent (1500 Mt) of the remaining bursts detonated on the surface. The fire effects of multiple nuclear detonations over cities and forests are complex and undetermined. Smoke from the fires of initial bursts could block subsequent thermal flash effects in some cases. Delayed bursts would probably spread existing fires, however, particularly by generating strong surface winds and convective plume activity. Closely spaced explosions over forests could greatly enhance the probability of fire ignition and spread. The problem of multiburst phenomena has not yet been adequately treated in the nuclear effects literature. Urban Ignition Some evidence that nuclear explosions are unique in their ability to ignite mass fires is offered by the Hiroshima and Nagasaki experiences. One crude estimate of the average energy release rate places the Hiroshima fire among the least intense of the mass fires of World War II (Martin, 1974~. Nevertheless, centripetal winds characteristic of a firestorm apparently developed, and the fuel consumption within the fire zone was nearly complete (GD77; Ishikawa and Swain, 1981~. Some of the factors that affect nuclear fire genesis in cities are summarized in Table 5.3. Even though the blast wave that follows the thermal pulse could extinguish many of the primary thermal radiation fires, a substantial number of these ignitions would continue to burn. Idealized field tests to determine the efficiency of fire extinction by pressure waves are contradictory, and often little or no effect is observed (Wiersma and Martin, 1973; OTA, 1979; Backovsky et al., 19821. In fact, in one study, the blast dispersal of burning curtain fragments through a room was a major factor in fire development (Goodale, 19711. In addition, the blast ignites many secondary fires and creates conditions (Table 5.3) that strongly favor the growth and spread of the surviving fires. Overall, blast would appear to encourage mass fire development. The evidence from Hiroshima and

TABLE 5.3 Nuclear Mass Fires Factors Contributing to Mass Fires Thermal irradiation ignites materials at fluences of 10 to 20 cal/cm2 over a large area. Blast starts secondary fires out to -2 psi overpressure. Fires are started simultaneously over a large area. Fires are ignited on both sides of major firebreaks. Blast scatters solid fuels, ruptures gas and liquid fuel lines, opens windows, and breaches firebreaks. Blast breaks water mains, blocks streets, and causes injuries that prevent effective firefighting. Nuclear fireball rise establishes central convective motion of a mass fire. Natural wind vorticity contributes to the formation of fire whirls and firestorms. Factors Inhibiting Mass Fires Blast wave extinguishes many fires started by thermal radiation. Blast covers flammable materials with nonflammable debris. Reduced visibility, due to natural or nuclear causes, limits thermal radiation effects. Meteorological factors such as winds, high humidity, and precipitation retard fire spread and coalescence. Nagasaki suggests that both primary and secondary fires eventually contributed to the conflagrations. Detailed models of nuclear fire initiation and spread in urban and suburban settings have been constructed (Miller et al., 1970; Martin, 1974; FEMA, 1982), although their fidelity is in some doubt (Miller et al., 1970~. The models suggest that, within the 20-cal/cm2 irradiation perimeter, >20 percent of the buildings could have one or more initial fires. This assumes that the blast wave extinguishes almost all of the primary fires and, overall, inhibits fire growth and spread {FEMA, 1982~. However, even if the initial fires are sparsely distributed after a nuclear explosion, nearly all blocks of houses or

48 buildings are likely to have at least one fire (Martin, 1974~. By implication, few effective fir ebreaks would exist in the initial fire zone. Observations of everyday urban fires indicate that fire spread between buildings (mainly by heat radiation and firebrands) is very efficient (-50 percent probability) at separations of about 7 m or less, and can occur over distances of 15 to 30 m (Chandler et al., 1963; Ayers, 1965; FEMA, 1982~. Rows of residential homes, and certainly buildings in city blocks, are generally separated by less than 10 m. Accordingly, there is a high probability that 50 percent or more of these buildings would eventually burn out (Martin, 1974; FEMA, 1982~. Owing to the dispersal of fuel by the blast into the gaps between the buildings, and the strong winds generated by the explosions and conflagrations, fire spread could be even more efficient in the nuclear case. Large isolated (industrial) buildings would also have a high probability of burning because of their large total area of exposure and therefore high likelihood of having at least one initial fire (Martin, 1974~. At blast overpressures of >15 psi, concrete and steel buildings suffer severe damage and break apart to produce rubble. The area of such damage is about 25 km2/Mt (GD77~. In densely built up areas, the rubble could be several meters deep. Fires can burn in rubble, but generally at a slower rate. Obviously, civil defense and firefighting efforts would be futile under such conditions, and fire spread would be uninhibited by gaps and open areas. The buried fuels would tend to smolder and pyrolize in the heated air that filtered through the rubble, thus smoking copiously. It is expected that a large fraction of the combustibles in the rubblized zone would eventually burn, possibly with an exaggerated smoke emission confined to lower altitudes. If an effective firefighting effort could be mounted, many of the initial urban fires might be extinguished and fire spread substantially limited in the lower-overpressure regions (Kanury, 1976; FEMA, 1982~. Such an expectation is probably optimistic. In Hiroshima and Nagasaki, even under wartime preparedness, firefighting efforts were largely futile (Ishikawa and Swain, 19811. Once the initial fires had grown to even moderate size, attempts at containment were hopeless without sufficient water, tools, and manpower. It follows that, within 1 to 2 h after a nuclear explosion over a city, major fires would be burning throughout the original fire ignition zone. Forest Ignition Little information is available on forest, brush, and grass fires initiated by nuclear explosions (Jaycor, 19801. Some factors that would influence the extent of nuclear wildfires are as follows: 1. The number of low air bursts over areas of forest, brush, and grass. 2. Meteorological conditions, such as cloudiness, precipitation, winds, humidity, and snow cover.

49 3. The accounting 4. The 5. The development probability of igniting persistent fires in the fuel bed, for the shading of dry fuels by the live canopy. probability of fire spread in the fuel bed. effects of blast on the distribution of fuels and the of fires. 6. Other factors, such as terrain, existence of firebreaks, and nearby nuclear explosions. Rough estimates for some of these factors, based on past wildfire experience and theoretical analyses of nuclear effects, are discussed below. Although Ayers (1965) had pointed out that many fires are likely to occur in a nuclear exchange, Crutzen and Birks (1982) made the first quantitative estimate of forest fire smoke and gas emissions in a nuclear war, and proposed that large quantities might be generated. As in cities, the nuclear bomb light is likely to ignite numerous small fires over a large area, most of which would be extinguished by the blast wave (Jaycor, 1980~. The area initially subject to ignition could be as large as 500 km2/Mt (Ayers, 1965), which corresponds to thermal fluences of >10 cal/cm2. It is possible that the number of individual fires surviving the blast wave and developing into major conflagrations could well exceed one per 10,000 m2 (i.e., 100 ignitions per square kilometer}. The rise of the nuclear fireball would establish strong afterwinds to fan the fires. It is unlikely that organized firefighting crews with sophisticated equipment would be available to extinguish the flames. Nuclear forest fires would not resemble most forest fires of the past. It is conceivable, although uncertain, that, because of the simultaneous ignition over a large area and the fanning action of the afterwinds, some of the nuclear forest fires could develop into intense firestorms with towering smoke plumes. The distribution and consumption of fuel in nuclear forest fires could also be significantly modified. For one thing, much of the forest canopy and some heavy timbers would be shattered and blown down into the burning zone. If the nuclear fire were very intense, even large standing timbers could be substantially charred. Thus nuclear forest fires might consume a larger fraction of the forest fuels than typical natural wildfires (see below). Huschke (1966) surveyed wildland fuel patterns and flammabilities in the United States as a function of region and time of the year. Some of his data are summarized in Figure 5.2. For summer conditions, up to 50 percent of all fuels (grass, brush, and timber) can have medium to high flammability at any given time (and presumably would be capable of sustaining type IT nuclear fires). The total area involved is about 2 x 106 km2, or about 30 percent of the continental united States. In winter, only about 10 to 20 percent of the fuels would exist simultaneously in such a flammable condition. Huschke's analysis is consistent with data on the geographical and temporal occurrence of "actionable. and "critically fire conditions developed by Schroeder and Chandler (1966~.

50 100 of - m ~ I ~ X I ~ m 111 _ ~ I ,` or O u. I ~ O Z UJ CL \\ ~ 50 Jan.\ O _ 1 1 o \ \ \ \\ \ N~ \\ \\ \ ug. AV9. ~ l l l 0.5 1.0 All Timber --- All Fuels FRACTION OF TOTAL FUEL FIGURE 5.2 Flammability statistics for wildland fuels in the United States. The curves show the percentage of the time that various cumulative fractions of the fuel are in a state of moderate to high flammability. (From Huschke, 1966) Accurate flammability statistics are not available for the Soviet Union, but similar fire conditions are likely to prevail at the same times of the year (e.g., Shostakovitch, 1925~. The potential for fire spread in wildlands is illustrated by U.S. Forest Service estimates of burnout areas for 1-Mt explosions, which range from 500 to 20,000 km2 (Hill, 1961~. Typical burnout areas would, of course, lie near the lower end of the range.

51 BURDENS AND DISTRIBUTIONS OF COMBUSTIBLE MATERIALS By using published information, the quantity and distribution of combustible materials in cities and forests may be determined with reasonable accuracy for the purposes of this assessment. Urban Combustibles Surveys of combustible materials in urban settings are available U.S. Department of Defense, 1973; Culver, 1976; Issen, 1980; FEMA, Flammables in urban zones include leaves and brush, wood framing and siding, paper and cardboard, stored gasoline and oil, tar and asphalt, rubber, natural and synthetic fabrics, plastics, paints, solvents, and a variety of other household and industrial chemicals. Worldwide anthropogenic production of combustible materials includes about 2000 Tg/yr of timber (roughly half of which is burned as fuel), about 2000 Tg/yr of liquid fossil fuels, about 100 Tg/yr of tar and asphalt, about 150 Tg/yr of oil-based lubricants and solvents, about 50 Tg/yr of plastics and resins, about 30 Tg/yr of natural and synthetic fibers, and about 10 Tg/yr of rubber (U.N., 1981~. These materials have been accumulating in cities for up to 50 years. At any time, perhaps 1000 to 2000 Tg of liquid and gaseous fossil fuels and related compounds are stored in urban areas. In city centers, combustible burdens can exceed 100 g/cm2. However, over the great expanses of modern urban and suburban complexes, and in smaller cities and towns, average combustible loads of 1 to 5 g/cm2 are more likely (FEMA, 1982; Turco et al., 1983b; Crutzen et al., 1984~. An average urban combustible burden of 4 g/cm2 is assumed in the present baseline case to represent the mean distribution between the city centers and the suburbs in metropolitan complexes (although excursions to lower burdens are considered below). Studies of urban and suburban areas worldwide indicate about 2300 cities with populations exceeding 100,000. The corresponding urbanized area is roughly 1.5 x 106 km2. There are about 180 cities with populations exceeding 1,000,000 {Turco et al., 1983b, and references therein). About 85 percent of the world's urban areas are located in the northern hemisphere. In the NATO and Warsaw Pact countries (which includes the United States, USSR, and most of Europe), there are ~1100 cities with Copulations Greater than 100 000 and ~80 major 19821. Some of the data are summarized in Table 5.4. _ , _ urban areas with populations exceeding 1,000,000 (the major urban agglomerations include some of the smaller cities). The total urban and suburban area involved is about 500,000 km2, and the co-located population is about 500,000,000. In the United States, cities tend to sprawl over large areas, while in the USSR and Europe, the cities are more compact. The total area of the ~core. cities in the NATO and Warsaw Pact countries is roughly 10 percent of the total urbanized area, or about 50,000 km2. These central urban zones, in which industrial/economic targets are concentrated, and near which significant military targets are often located, are likely to be hit by (e.g.,

52 TABLE 5.4 Typical Combustible Burdens in Urban Areas Average Combustible Overall BurdenBuilding Combustible Building per StoryaNumber of Densitya Load Type (g/cm2) Stories (percent) (g/cm2) Residential: brick or frame 5-10 1-2 10-25 0.5-5 Office and commercial 5-20 2-10 20-40 2-80 Industrial 0-15 1-3 20-40 0-18 Storage 10-40 1-2 20-40 2-32 aData taken from FEMA (1982~. nuclear explosions, are heavily loaded with combustibles, and are expected to burn vigorously. Considering the typical design of cities, with most facilities and activities concentrated at the center, it appears that less than 10 percent of the total urban area may hold 50 percent or more of the total urban combustible material. The urban centers of the world may hold a to-tar of 10,000 Tg or more of flammables, while the surrounding suburban areas may contain an equal amount. It is estimated that up to three-quarters of all the combustible material consists of wood and wood products, with perhaps 5 to 10 percent in plastics, resins, and rubbers, and the remainder in oil, tar, asphalt, gasoline, solvents, and other fuels and organochemicals. At the present time, no comprehensive inventory of urban combustible materials is available. Crutzen and Birks {1982) pointed out that oil and natural gas production and storage fields might be targets of nuclear explosions. Fires in some uncapped wells, once ignited, could burn uncontrolled for months or longer. Crutzen and Birks estimated that the rate of fuel release and combustion might equal the current world production rate, or about 300 Tg per month. Although such fires are potentially important if ignited, they are not included in the baseline analysis. In some very intense urban fires, materials not usually thought of as "flammable" could ignite and burn, or decompose into vapors that later nucleate into particles. Of particular interest is aluminum, which is used in large quantities in modern construction, which burns readily, and which forms about 2 g of smoke per gram of aluminum burned. Other compounds containing magnesium, calcium, and sodium

53 could generate additional inorganic smoke under extreme conditions. These possible inorganic sources of smoke are omitted from the baseline case because of the difficulty in estimating the quantities involved. Forest and Wildland Combustibles Forests cover about 40 percent of the land surface of earth, or ~4 107 km2, and brush, grass, and pasture lands account for perhaps another ~4 x 107 km2. In the United States, Canada, Europe, and the USSR, the total forested area is ~1.8 x 107 km2, or about 40 percent of the total land area. The average concentration of combustible biomass in temperate and boreal forests is ~2 g/cm2 (Seller and Crutzen, 1980; USDA, 1981~. In natural wildfires, up to 25 percent of this fuel is burned, consisting of organic compost, ground litter, understory growth, leaf canopy, and small branches and twigs (Wright and Bailey, 1982~. Because of the potentially unique character of nuclear forest fires, a larger quantity of fuel might be consumed, as noted earlier. Nevertheless, it is assumed here that 0.4 g/cm2 of fuel would be burned in nuclear forest fires, which is typical of many natural forest fires (Wright and Bailey, 1982; Crutzen et al., 1984~. Brush and grasslands have much lower fuel burdens than forests (~1 g/cm2), but more of the fuel (~50 percent) is likely to burn. Accordingly, fuel consumption in nuclear brush fires might be estimated as about 0.1 to 0.2 g/cm2, and in nuclear grass fires, about 0.02 to 0.05 g/cm2 (USDA, 1972; Wright and Bailey, 1982~. Smoldering fires in peat soils have been known to burn out accumulated organic layers several feet thick (Wein and MacLean, 1983~. Nevertheless, brush, grass, and peat fires are ignored here. Urban Combustibles Consumed In a full-scale nuclear exchange, attacks against military, industrial, and economic targets in and around cities might involve multiple detonations to achieve complete destruction. This could limit the average effective area and combustible load ignited per megaton of explosive yield. A number of other complex factors could further limit the quantity of combustible materials exposed to nuclear fire, while some factors could increase the quantity. Factors that could limit the potential urban fire area and/or the burden of flammables consumed include the following: 1. Direct overlap of thermal irradiation zones when detonations are closely spaced. 2. Multiple targeting (two or more consecutive explosions on one target) in order to assure destruction. 3. Obscuration of bomb light by smoke and dust generated by previous explosions and fires. 4. Regions of high overpressure (8 to 15 psi), which produce more rubble and bury more flammable material.

54 5. Emphasis on "courter for ce" (military) targeting in many strategies. 6. Geographical factors (bodies of water) and topographical factors (hills) that delimit or shield possible ignition areas. 7. Civil defense efforts. Factors that could increase the potential fire damage include the following: 1. Confinement of the heaviest combustible loadings within the areas most likely to suffer multiple detonations, i.e., co-location of flammable materials and key urban and industrial targets. 2. Attack strategies that employ damage assessment--particularly in the case of large, immobile, and highly vulnerable urban and industrial targets--to conserve and optimize the use of nuclear forces. 3. Enhancement of fire ignition and spread probabilities for multiple bursts over targets, due to heating by additional thermal irradiation, fanning by surface and convective winds, and spreading by firebrands. 4. Fire ignition and spread well beyond the 4- to 5-psi region, as was observed during World War II. 5. Collateral damage to urbanized areas from attacks against military targets, for which there is (a) less inherent overlap of the fire zones and (b) ignition of the perimeter of cities with increased likelihood of spread into the city centers (Larson and Small, 1982a) 6. Localized enhancements in nuclear thermal irradiation due to reflection from clouds. 7. Fire damage from tactical nuclear and conventional weaponry. Obviously, a detailed calculation of the urban ignition area in a nuclear conflict would require an enormous amount of information, much of which is not easily obtainable. For the present assessment, certain simplifying assumptions have therefore been made. Of the 1500 Mt detonated over urban areas in the baseline scenario (see Chapter 3), 500 Mt is assumed to be ineffective due to overlap of thermal irradiation zones. In accordance with previous discussions, the average fire area per megaton of yield is taken to be 250 km2/Mt for the other 1000 Mt. The industrial, economic, and co-located military targets in this 1000-Mt attack are assumed to be distributed among the approximately 1000 largest cities of the NATO and Warsaw Pact countries (Chapter 3~; these cities collectively have an urban/suburban area of about 500,000 km2. Thus, by implication, fire ignition would occur over 50 percent (250,000 km2) of the developed area of the warring nations. While military facilities in China, Japan, and other countries might also be targeted, nearby urban areas are not assumed to be affected. The average combustible burden is assigned a value of 4 g/cm2 for urban/suburban construction. That is, regions of high combustible loading For example, central cities, where flammable burdens can reach 100 g/cm or more) are averaged together with regions of low loading to account for the likely fire damage to the vast suburban and

55 residential areas of the developed nations. It follows that 10,000 Tg of combustible material would be subject to nuclear ignition in the baseline case. The question of overlap in urban fire zones is probably not critical to this analysis because (a) anthropogenic flammable materials are highly concentrated in relatively small areas (in cities), and (b) urban nuclear attacks would be likely to trigger conflagrations that would spread outside of the original ignition zones. For example, if the present 1500-Mt urban attack were assumed to be concentrated over the city centers, with an areal ignition overlap factor of 10 (i.e., with an average ignition area of only 25 km2/Mt), 37,500 km2 of the central cities with the densest combustible loading could still burn. For a central city fuel burden of 20 g/cm2, nearly the same total quantity of combustible material would be impacted as in the baseline case. The inevitable spread of urban conflagrations would ensure even greater fuel consumption. Given that up to 1000 cities could be affected in the baseline exchange, such extensive overlap (a factor of 10) is very unlikely in the first place. The low sensitivity of the quantity of urban fuel impacted to the total yield of an exchange is also demonstrated by the 100-Mt excursion scenario of Turco et al. (1983a). In this case, a 100-Mt attack with 100-kt warheads directed exclusively against the largest built-up city centers (and spaced to ignite about 25 km2 per 100 kt) was found to consume roughly the same amount of combustible material as a generalized 1000-Mt urban/suburban attack. Other approaches could be taken to estimate the areas burned and flammables consumed in urban nuclear fires. For example, a comprehensive target analysis might be carried out. This would require detailed information about the precise locations of key industrial and military facilities, the flammable materials at these locations and over large surrounding areas, and the plan of attack to disrupt production and military operations. Clearly, unless all potential targets and a number of attack strategies are considered, the estimation of impacted urban flammables might not be significantly improved. Hence, the simplified approach adopted in this report seems reasonable at this time, and is in quantitative agreement with estimates of material consumption obtained by a number of other schemes (Turco et al., 1983a,b; R. Tur co, private communication, 1984; Broyles, 1984; Crutzen et al., 1984~. Forest and Wildland Fuels Consumed As in the case of urban fire ignition, a number of complex factors could affect the area and the quantity of wildland fuels consumed in a nuclear exchange. Factors that could limit the potential wildfire area and fuel consumption include the following: 1. Overlap of target zones, particularly in missile silo fields. 2. Multiple bursts over isolated "hard" military targets.

56 3. Smaller ignition areas for very low air bursts and ground bursts aimed at ~hard" targets. 4. Obscuration of bomb light by dust and smoke from previous explosions. 5. Restricted flammability of wildlands due to meteorological and seasonal effects. 6. Firebreaks and firefighting. Factors that could augment wildland fuel consumption include the following: 1. Multiple explosions, which would increase the probability of fire ignition and spread in forests because (a) the blast would knock down the vegetative canopy, which is a major factor limiting fire ignition by single explosions over forests, {b) the repeated thermal irradiation and winds would dessicate fuels, augmenting fire ignition and spread probabilities, (c) the explosion winds would spread firebrands and fan established fires, and (d) natural clouds and the dust clouds created by ground bursts would scatter back and intensify the thermal radiation field of very low altitude detonations. 2. Location of a substantial fraction of the hundreds of military bases (other than missile fields) within several kilometers of forested areas; up to 2000 Mt could be directed against these targets. 3. Tactical nuclear warfare (in Europe) involving explosions over heavily forested areas (used for camouflage); tactical weapons are particularly effective in forest fire ignition (Woodie et al., 1983~. About 500 Mt of tactical weapons are detonated in the baseline scenario. 4. Frequent spread of fires well outside of the ignition zone; occasionally, an individual fire might spread over 10,000 km2 or more. 5. Consumption of vast quantities of other wildland fuels such as grass and brush. 6. On military bases, particularly strategic air bases, burning of large stores of aircraft fuels, buildings, munitions, and other materials. It is assumed in the baseline case that 5000 Mt of explosions are detonated over widely dispersed military targets. Ignition probabilities are greatest in late spring, summer, and early fall {Schroeder and Chandler, 19661. Attacks at other times of the year would produce fewer fires. On a purely random basis, about 40 percent of the military explosions could occur over forests, with 40 or 50 percent of the remaining explosions occurring over brush- and grass-covered lands. Since, according to Huschke (1966), roughly 50 percent of such areas are capable of sustaining type II fires at any given moment under summertime conditions (Figure 5.2), an average total ignition yield of about 2000 Mt might be expected. Of course, many of these explosions would overlap, so that an effective ignition yield of about 1000 Mt may be more reasonable. The corresponding total ignition area would be 250,000 to 500,000 km2, corresponding to a thermal fluence of 10 to 20 cal/cm2 {e.g., Hill, 1961; Ayers, 19653.

57 Missile silo fields occupy an area of about 250,000 km2, counting fringe regions. The silos are spread out to reduce vulnerability. Based on a survey using Landsat photographs and coarse vegetation maps (Short et al., 1976), about 20 percent of the land housing silos appears to be forested (i.e., 50,000 km21. A 2000-Mt overlapping barrage against these silos would probably incinerate these regions almost entirely. Other military attacks would account for about 3000 Mt of additional explosions (after deducting the silo yield of 2000 Mt and the urban yield of 1500 Mt from the total baseline yield of 6500 Mt). If 40 percent of these explosions occurred over forests and ignited 250 km2/Mt half of the time, another 150,000 km2 of forest land could burn. This assumes that areas affected by only one explosion at irradiation levels of 20 cal/cm2 or greater have a 50 percent probability of ignition (Huschke, 1966) and that areas affected by two explosions, each at >20 cal/cm2, have a much greater probability of burning because of the drying effect of nuclear thermal radiation on exposed vegetation.* It is further assumed that, in this category of targeting, more than double overlapping at such high thermal intensities and overpressures is unlikely. Finally, fire spread is assumed to increase the overall area of burnout by 25 percent, allowing as well for some ignitions outside of the 20 cal/cm2 zone tHill, 1961; Ayers, 1965~. Military barrage attacks, in which overlap is purposefully minimized to optimize the area of impact, are omitted. Such attacks could be directed at the strategic bombers scrambling from air bases, at mobile missiles hidden in multiple bunkers or dispersed in forests. It should be obvious that such tactics would significantly increase the areas of wildfires. The total area of forest fires in the baseline nuclear war is thus taken to be 250,000 km2, with all other related fires neglected. The total combustible fuel within this area is about 5000 Tg, of which 20 percent is assumed to burn (see the previous sections). An area of 250,000 km2 seems reasonable, in view of the physical characteristics of the targets and the flammability statistics of wildlands. In previous generalized analyses, Crutzen and Birks (1982) estimated a forest fire area of 1,000,000 km2; Turco et al. (1983a), 500,000 km2; and Crutzen et al. (1984), 200,000 to 1,000,000 km . On the other hand, Small and Bush (1984), using a more detailed targeting and fire ignition methodology, have estimated a forest burn-off area of only 70,000 km2 {and a total fire area of 170,000 km2) in a summertime exchange of about 4000 Mt. The differences between the highest (~106 km2) and lowest (~105 km2) forest fire area estimates have yet to be resolved. *The latter assumption must be tested, but is based on the noticeable "greying n of live vegetation by bomb light during nuclear tests.

58 SMOKE EMISSIONS AND PROPERTIES The dense smoke plumes from massive urban and natural fires have been described by many observers. Downwind of large conflagrations there are often reports of dark skies, red suns, and black rain (e.g., Plummer, 1912; Lyman, 1918; Tshikawa and Swain, 1981~. Photographs show black plumes rising over industrial fires, and satellite images show wildfire plumes extending downwind for hundreds of kilometers-- direct evidence that large fires can cause profound local optical and physical perturbations of the atmosphere (e.g., Davies, 1959~. The important properties of smoke are the quantity generated per unit mass of fuel consumed, the particle composition and size distribution, the specific extinction and absorption coefficients (expressed in square meters per gram) at visible and infrared wavelengths and the heights of infection into the atmosphere. All of these properties are highly variable, depending on fuel type, moisture, burning conditions, and so on. Experimental evidence suggests that the smoke from a composite array of fuels in a fire may be roughly summed over the array (Rasbash and Pratt, 1979), although in certain instances the average smoking tendency of the mixture can be lower than the sum for the individual components (Rasbash and Drysdale, 1982~. Using an approximate linear addition rule, average values for smoke characteristics can be deduced if the fuel array is adequately defined. In terms of the optical properties of the smoke, an attempt is made here to be somewhat conservative in estimating the possible sooty, highly absorbing component. . . . Urban Smoke Emission Factors* Many of the flammable materials that are commonplace in homes, businesses, and industry can generate dense, sooty smoke and toxic gases when burned in an uncontrolled environment. The mechanisms of soot formation in flames (Calcote, 1981; Nakanishi et al., 1981) and smoldering fuels (Bankston et al., 1981) have been investigated, but no complete theory has yet been established. A variety of molecular organic neutral and ionic precursors in the flame may combine to form incipient soot (graphitic carbon) particles, which continue to grow by accretion of organic material while losing hydrogen relative to carbon. Several studies have been made of the soot produced by organic gas flames (Chippett and Gray, 1978; Pagni and Bard, 1979; Kent and *The smoke emiss ion factor can be exur eased as the mass of ~;mc~ke generated per unit mass of fuel burned--g-smoke/g-burned--or as a weight percentage yield of smoke--percent of burned mass converted to smoke. The factors are interchangeable; i.e., a 4 percent smoke yield by mass is equivalent to an emission factor of 0.04 g/g.

59 Wagner, 1982), anthropogenic liquid and gaseous fuel combustion (Day et al., 1979; Kittelson and Dolan, 1980; Wolff and Klimisch, 1982), and burning solid materials (Hilado and Machado, 1978; Ohlemiller et al., 1979; Jagoda et al., 1980; Bank s ton et al., 1981; Vervisch et al., 1981; Butcher and Ellenbecker, 1982; Tewar son, 1982~. Some properties of smoke produced by burning wood and polymers are given in Table 5.5. The quantity of smoke and soot generated by a fire depends on several factors, including the efficiency of ventilation of the fire and the average temperatures in the pyrolysis and burning zones (Quintiere, 19821. In controlled combustion devices, such as oxyacetylene torches, oil burners, and automobile engines, smoke and soot emissions are minimal (unless pretuned operating conditions are disturbed). Even a crackling fire in a fireplace represents well-controlled combustion (although particulate emissions can approach 1 to 2 percent (by weight) of the wood burned (Dasch, 1982~. In laboratory experiments on smoke generation, the smoke emission during flaming combustion is generally observed to increase rapidly as the oxygen available in the burning zone decreases (Saito, 1974; Morikawa, 1978; Tewar son et al., 1980; Tewar son and Steciak, 1982), or as the ventilating air supply is preheated (Morikawa, 1978; Powell et al., 1979; Bankston et al., 1981~. The smoke output of flaming wood decreases as the sample is heated artificially by radiation (Bankston et al., 1981), although the opposite effect is seen in other materials (Seeder and Einhorn, 1976; Tewar son, 19827. In nonflaming combustion, smoke emissions typically increase markedly when the samples are radiatively heated, but decrease in certain materials such as polyurethane foam when the air supply is preheated (Seeder and Einhorn, 1976; Bankston et al., 1981~. In unregulated fires, the efficiency of ventilation by winds and turbulence should tend to decrease as the fire area and fuel density increase. Accordingly, with an expansion in the scale of a fire, the smoke emission factor should increase (Rasbash and Drysdale, 19821. This effect is generally, but not always, observed under experimental conditions (Quintiere, 1982; Rasbash and Drysdale, 19821. On the other hand, the intensification of a fire with size tends to reduce smoke emissions through several mechanisms: augmented winds and turbulence, which enhance local ventilation; higher temperatures, which incinerate the smoke over the fire; and induced precipitation, which scavenges the particles. The smoke from very intense fires, although reduced in quantity, is enriched in graphitic carbon (soot) (Morikawa, 1978; Wagner, 1981), and thus absorbs light more effectively (see below). The complete incineration of soot in flames requires temperatures in excess of 1500 K (Wagner, 1981), whereas the temperatures above flames, even in very intense fires, are typically <1000 K. Hence it is not clear that violent burning would significantly reduce, through reduced smoke emissions, the long-term optical effects of large-scale fires. The complexity in predicting the smoke-forming properties of materials can be illustrated by using laboratory test data for construction lumber. During flaming combustion fully ventilated by room-temperature air, Bankston et al. (1981) found that Douglas fir irradiated by a supplementary heat source of 2.5 W/cm2 had a smoke

60 ~q o e~ · - o · - 0e ~n, CO · - a o U) ~q o tn X U' . U~ - n ~; E~ C · - a v - - d~ - v ~S - dP - · - ~a ~n o . o I O I · o ~r U~ ~ · O O I , ~ N · ~ ~ o o 1 1 -I 1 1 ~ o I_ ~ ~ 1 1 1 . · . I _ O O O · · U~ U~ 1 1 o 1 o t- N ~ N 1D d' N . · I · · 1 1 U~ ~ ~ O 00 1 1 1 1 1 I I 1 1 1 o I I · ~ O O '~ ~ ~V V ~ I ~ I _I O O 0 1 C5\ ~ a' 0 _1 0 · · · ~ · 1 · . . . . . O O O O O O O O O O O ~ ~ A U~ . o 1 In c ~1 ~ ~1 1 1 1 1 C ~o · I · · · 1 1 1 1 1 · 1 ~ O O O O O I O kD ~ · ~ C ~U~ o · I C~ ~ U ~C~ · 1 · · 1 1 1 1 vo 1 · 1 _~ 1 ~) _I _~ I I ~ ~ o 1 0 0 -I V ~ O {Q O ~CQ ~ ~ c: S ~ ~ ~C V ~ c ~ ~ -~ S ~ ~-~ Q Q Y Q' ~- - eq {Q V O S ~ V ~· - 1 .~ \- ~ ·~ 1 ~ -~ :>, ~>,~ ~X ~ ~Q ;~; ~ O ~ ~ Q ~ O O O O £ O Q4 - ~ ~ ·- :4 ~ ~ ~ p. ~r; 0 o Q S~ tQ o O ~ ~ tR O · - V · - S~ ~ s~ S~ .,, ~Q ~· a, U]o · ^ eq - Q ~n ~· - v s V ~V s~ a U~ ~ ~ ~ o .Q v~ a) ~ = -~ ~ Q ~ v ~Q ~ ~ £ O · - 1 1 1 1 1 O ·~- - 1 1 - _I I ~ ~ ~ ~ O c- - ~ ~ ~ O o~ v ~3 v ~ a, s 3 tQ eq ~s S ~o S~ · - Q. ~ V ~ _= s. - ~ ~· - m~ ~ ~ ~a s .,, ~, ~ _, O 3 Q. ~ eQ v O ~ a a, s ~ ~ · - ~ O ~ ~~ - ~ ~· - ~n ~ ~ ~ ~ o ~ . ~ ~ ~ ~r ~ ~ 0 ·. eq ~l · o V- - ~n [Q 3 ~ O es tn ~ ~ O ~ · - 11 - - - rn ' c, O s ~ £ ~ 3 ~0 s 0- - ~n ~ a: -~ - 3 ·^ ~ ·-^ S-' ·' ~ ~ O ~n Y ~.rl ~ ~ O ~ JJ V £ ~ ~ tn- - V u, .,,.,, ~- - ~ O V s ~ ~ v ~l ~n a, ~ ~ ~ {Q ~ v 0 ~ ~ e- - ~ ~ ~ O ~ ~ =- - V · - ~n ~ · - a, ~ ~u, ~ Q ~ £ £ ~ ~ £ N :D ~ [Q ~n a~ s 0 £ O ~ · ~' ~ 3 ',q c: bO tQ ~n ~ v ~ ~ ~ ~ ~ 0 ~ a, 0 ~ ~ ~ eQ · - ~ ~5 ~ v -, ~ 0 ~ ~ 0 ~ al Q V ~ ~ ~ ~ ~ ~ ~ · - a, ~ ~ · - ~ ~ ~ ~ S ·^ ~ ~S J~ U1 3 rQ ~ ~Q ~ ~ ~ ~ ~ O ~ ~ ~ O ~ ~ ~ O O ·- O E~ V ~ ~ ~ ~ ~ O C-) ~ E4 05 Q V ~ V V · - · - S ~0 - o 3 E~ ~D - o bO ~n · O ~ ·- m ~ · .-, ~. - 3 ~5 ~s ~ o. - s 3 v ~ ~ _ s [e = ~- O ·^ s ~ == s m- - ~v 3 Q eq a~ ~ ~ N .,' ·, O O ~S~ £ E' O c' ~5 tn tQ a,. - ~ ~ ~· - ·-1 ·^ ~:~: _ _ - ~ o {Q CQ ~ a, .,' . - 0 .,! .y : _ ~ O ~ ~ ~ s ·,l .,la ~a, ~: v v ~ ~ O QS V _ O O [Q O U] ~·. {Q a, ~c' V a ~, p<, H O u,

61 emission factor of about 2.5 percent. At an irradiation of 5 W/cm2, however, the emission factor was <1 percent. In other flaming tests with Douglas fir at irradiation levels of 5 W/cm2, Powell et al. (1979) observed a steady increase in smoke production as the air flow was heated from about 300 K to 600 K. By contrast, with smoldering Douglas fir, a substantial decrease in smoke production (from a reference yield of about 15 percent) was measured as the ventilating air was heated from 300 K to 600 K (Powell et al., 1979~. Saito (1974) noted that dry spruce and other wood products generated less smoke as flaming samples were heated from about 750 K to 850 K. In larger-scale chamber tests, however, 2 to 4 times as much smoke was produced at the higher temperatures as in the individual sample tests (Saito, 1974~. Finally, room-size experimental wood fires were found to emit, on the average, about 3 to 6 percent smoke by weight (Rasbash and Pratt, 19791. The smoke emissions from a variety of noncellulosic urban combustibles (e.g., oil, tar, plastics, and rubber) range from about 5 to 40 percent (of the weight of fuel consumed) under smoldering conditions to 1 to 15 percent in flames (Table 5.5~. In large building and area fires, all burning conditions exist simultaneously, and an average smoke emission factor of 6 percent is probably reasonable for these materials (Seador and Einhorn, 1976; Rasbash and Pratt, 1979~. For dry construction lumber, furniture, paper, and other cellulosic materials, an average flaming/smouldering smoke emission factor of 3 percent can be used (Rabash and Pratt, 1979~. In subsequent baseline estimates, an average smoke emission factor of 4 percent is adopted for all urban fires (before smoke scavenging and removal in the fire plumes--see below). This average emission figure assumes that two-thirds of the material burned is cellulosic with an emission factor of 3 percent, and one-third is noncellulosic or polymeric with an emission factor of 6 percent. For urban targets-- industrial complexes and fuel production, distribution, and storage centers--the latter class of combustibles is even more common, and thus the relatively high percentage of these materials in the burned inventory. In older cities and residential areas, wood construction dominates the flammable burden. Flame retardants applied to certain materials, such as flexible polyurethane foams, may actually increase their smoke output (Bankston et al., 1981~. Most recently, smoke retardants have been developed for commercial applications, but these are not yet in wide use. Size Distribution and Composition The size distribution of smoke particles shows a significant variability (Table 5.5~. Most sooty smokes are composed of nearly spherical ultrafine graphitic (carbon) crystals of about 0.005 to 0.02 um in radius and varying amounts of oils and tars. In dry smokes the carbon crystals coagulate into chain structures with dimensions of the order of 0.1 um. In oily smokes, droplets of heavy organic liquids laced with carbon nuclei can grow to sizes exceeding 0.2-pm radius. In either case, individual smoke particles may be composed of 10 to

62 1000 or more soot nuclei. At very high rates of pyrolysis, soot particles often agglomerate to sizes of 0.5 um or larger (Bankston et al., 1981~. In thick carbonaceous smokes, loose clusters of soot nuclei can grow to 10 Am, with occasional fluffy agglomerations reaching 100 um (Day et al., 1979~. (However, in these extreme cases it is suspected that individual soot clusters may aggregate on the filters used for collection.) The elemental (graphitic) carbon fraction of the smoke generated during the unregulated combustion of liquid fossil fuels and synthetic fibers and plastics can range from 5 to 90 percent by weight (e.g., Tewar son, 1982), but is typically greater than 20 percent (Table 5.5~. Burning wood generally produces less graphitic carbon (<20 percent). As was noted in the previous section, smoke from very hot, intense fires has a higher graphite content independent of the fuel. Considering the types of combustibles and burning conditions expected in urban nuclear fires, an average graphitic (soot) fraction of about 20 percent for the smoke seems conservative. Urban smoke generally absorbs sunlight more effectively than forest fire smoke, which is typically about 10 percent graphite (see below). The baseline urban smoke emission factor of 4 percent (0.04 g/g) implies a graphitic carbon emission factor of 0.8 percent. Accordingly, in the nominal case, the generation of light-absorbing particles in city fires is taken to be quite inefficient. The size distribution of smoke particles may be represented by a log normal distribution, nO 1 In ~ r/r nfr) = ~ exp _ (2~T) In ~ 2 In ~ where nfr~dr is the number of particles in the size interval r ~ r + dr per cubic centimeter of air (i.e., n has dimensions of particles per cubic centimeter of air per unit radius interval), Em is the number median (or mode)* radius, and ~ is a measure of the dispersion or width of the size distribution. Here, the radius is the spherical volume-equivalent radius for oddly shaded particles. The density, p, of pure graphitic carbon is ~2.25 g/cm ; for most oils, p < 1 g/cm3. Smoke particles that are composed predominantly of oils will be spherical and have a density of ~1 g/cm3. Smoke particles that are nearly pure elemental carbon will be characterized by a fluffy chained structure with a low effective ratio of mass to size (i.e., an effective density of <<2.25 g/cm33. *For the log normal size distribution chosen, rm is both the number median radius (above and below which lies exactly half of the particles) and the mode radius (at which lies the peak in the number distribution plotted in In r coordinates).

63 The value of y for a wide range of natural and man-made aerosols falls in the range of 1.5 to 2.5. A reasonable average choice for smoke is ~ = 2.0, which is consistent with a number of measurements (Turco et al., 1983b). In the log normal size distribution with ~ = 2, the mass median particle radius* is approximately 4.3 times the number median radius. Several factors may lead to changes in the size distribution of smoke particles within the plume of a mass urban fire: 1. Coagulation of the particles at high concentrations. 2. Collection of the particles by water droplets followed by coalescence of the smoke as the droplets evaporate. 3. Scavenging of the fine smoke aerosols on the surfaces of larger ash and debris particles. Coagulation. Coagulation by Brownian diffusion is undoubtedly responsible for many of the large soot particles observed in thick smoke plumes (Day et al., 1979~. (Aggregation of particles on filters and other sampling surfaces is responsible for occasional misinterpretations, however.) Because of coagulation, smoke concentrations in a plume are not expected to exceed approximately 105/cm3 after several hours, or approximately 104/cm3 after several days (Twomey, 1977~. Turco et al. (1983a,b) carried out detailed coagulation simulations for both confined and dispersed clouds, in which the evolving size distribution was predicted as smoke function of time and height. These size distributions were then used to perform optical calculations, assuming that the particles are always spherical. Turco et al. showed that, in artificially confined plumes with slow dispersion, the smoke particles could coagulate to average sizes of 0.5-pm radius, and that their optical depth, when related to the equivalent value for smoke distributed over the northern hemisphere, was reduced by about a factor of 2 after about 1 month. This relatively small difference in optical depths followed because (a) the initial smoke clouds at different altitudes were not equally dense and so the particles coagulated to different extents, and {b) the size distribution changed in dispersion (y) as well as in average size (rm), which reduced the overall impact on the optical properties. Such results indicate that the effects of coagulation on the optical characteristics of smoke are not straightforward. Turco et al. also showed that, even if the fire plumes are instantly dispersed over the hemisphere, the smoke particles eventually coagulate to sizes up to about 0.2 um (after about 1 month). Crutzen et al. (1984) analyzed the coagulation problem using "self-preserving" size distributions. However, for the present analysis, an even simpler approach is adopted which considers only the mean particle size. *This is the radius above which, and below which, half of the total particle mass lies.

64 A simple model of an expanding fire plume may be used to derive the following approximate expression for the change in the average particle radius with time due to coagulation: r/rO = ~ 1 + 2 KnO l ln(1 + fit) I 1/3 Cl The initial particle radius is rO, and the initial particle concentration is nO (particles per cubic centimeter). K is the Brownian coagulation kernel (-1 x 10-9 cm3/s for particles of about 0.1 um in radius), and a is the linear plume expansion coefficient (per second); i.e., a~1 is the time in which the initial plume volume doubles. Table 5.6 illustrates the dependence of the particle size on the parameters no, a, and t. It can be seen that, except in the most extreme cases of high initial particle concentration and retarded plume dispersion, the particles undergo only modest growth by coagulation in the first day after emission. This treatment assumes that Brownian coagulation is the dominant aggregation mechanism, and that the effects of electrical charge are small. A normal Boltzmann distribution of charge on the aerosols has little influence on the coagulation rate (Twomey, 1977~.* Turbulence in fire plumes, which might be comparable in intensity to the turbulence in natural convective systems, is also a secondary factor (Rosenkilde and Serduke, 19831. Baum and Mulholland (1979) simulated smoke coagulation in a buoyant plume using a coupled dynamical/microphysical model. They found that coagulation was significant only with very large initial particle concentrations, exceeding ~108/cm3. In such cases, the coagulation process still tended to "freeze out, n or turn off, rapidly. They estimated that particle radii could double under such extreme conditions. The data suggest that the dilution of smoke in a typical buoyant plume occurs within minutes of emission, which may not necessarily be the case in a very large plume. Tsang and Brock (1982) describe the effect of coagulation on the optical extinction of extended aerosol plumes propagating through the atmosphere; they show that the effect is quite small even for very strong sources of O.l-pm particles. Estimates of initial smoke particle concentrations just above a fire can be obtained from experimental data, and used to place realistic limits on nO for open, uncontrolled combustion. Palmer (1976) directly observed smoke particle concentrations of about 4 x 105/cm3 (with an average radius of about 0.1 vm) just above a large, intense test fire in the Flambeau series. Concentrations of smoke particles measured in prescribed forest fire plumes are typically about 105/cm3 in the plume core close to the fire (Packham and Vines, 1978~. Benech (1976) determined the air volume flow rate *In fact, electrical charge may inhibit the agglomeration of soot particles (Valioulis and List, 1984~.

65 TABLE 5.6 Relative Sizesa of Coagulating Particles in an Expanding Smoke Plume Coagulation Time = 1 hour Coagulation Time = 1 day no a (s~l) (s~l) (particles/ cm3) ~10-3 10-4 10-5 10-3 10-4 10-5- 1 x 106 1.2 1.3 1.4 1.5 2.3 3.2 1 x 105 1.02 1.05 1.06 1.07 1.3 1.6 1 x 104 1.00 1.00 1.01 1.01 1.04 1.1 aThe ratio of the final particle radius to the initial radius is given for each set of physical parameters. The particles are assumed to retain a spherical shape. . through a 600-MW fire at the Meteotron facility, which burns fuel oil. Since the rate of energy release in a fire can be related to the rate of fuel consumption and thus to the rate of smoke production, the calculated rate of air mass flow through the fire may be used to deduce an initial smoke particle concentration. Assuming a smoke emission factor of 0.04 g/g and a heat of combustion of 4 x 104 J/g for oil, Benech's data imply a maximum concentration of about 1 x 106/cm3 of 0.1-pm smoke particles just above the fire. Carrier et al. (1984) simulated the plume dynamics of a very large urban mass fire. Their results suggest a peak smoke concentration of about 2 x 106/cm3 of 0.1-pm mode radius particles. The calculations also indicate that the plume expands in volume by a factor of 10 in the first ~100 s. Observations and simulations of fire plumes suggest that maximum initial smoke particle concentrations (nO) are about 106/cm3. The initial expansion and dilution rate of the plumes are also very rapid, with a >> 10-3/s (e.g., Benech, 19763. Therefore, according to Table 5.6 therefore prompt coagulation should be rather limited in most plumes, unless the fires are unusually smoky or the plumes unusually compact. After a period of <1 hour, the maximum smoke concentration in a plume would probably be about 105/cm3, due primarily to dilution by the entrainment of ambient air. In a series of measurements of the sooty aerosols generated by the oil-burning Meteotron facility, Radke et al. (1980a) and Benech et al. (1980) recorded the process of smoke aging through coagulation. Initially (in the rising plume), the smoke was dominated by very small particles (<0.05-um radius). After 18 min. the concentration of these particles had begun to decrease in relation to the concentration of particles of >0.05 um. By 30 min. a well-defined size mode had developed, with a number mode radius near 0.1-pm radius. Between 30 and 40 min. the mode structure remained stable as dilution apparently

66 controlled the size distribution. After 40 min. essentially all of the smoke particles still had sizes below 0.5-pm radius. The most important direct effect of smoke coagulation is to reduce the number and optical efficiency of the particles. In a series of experiments, Seader and Ou (1977) measured the "optical density" {equivalent to the specific extinction coefficient, see below) of smoke from a variety of cellulosic and polymeric materials, for flaming and nonflaming combustion, at concentrations ranging from 15 to 2750 mg/m3, and for aging times up to several minutes or more. Smoke produced by smoldering had a specific extinction coefficient of about 5 m2/g with a maximum overall variation of a factor of 2, and smoke produced by flaming, a coefficient of about 8 m2/g with a much smaller variation. These results indicate that the optical properties of smoke are not particularly sensitive to the initial coagulation of the smoke particles. Sooty smoke, moreover, exhibits the same typical extinction and absorption coefficients for a wide range of combustion sources and aging periods (e.g., Janzen, 1980~. On the basis of this discussion of coagulation, the use of a (spherical) smoke particle mode radius of about 0.1 um for optical calculations appears to be reasonable. Only in unusual circumstances would the average smoke particle size exceed about 0.2 to 0.3 Am due to prompt coagulation in the fire plume. In such cases, variations in y should also be considered (and the particle morphology would play a role as well). In aged wildfire plumes, smoke particles are observed to have a mode radius of about 0.05 to 0.1 Am (the relevant data are reviewed later), which again suggests that prompt coagulation exercises only a secondary influence on average particle sizes in typical fire plumes. Droplet Scavenging. The collection of soot and windblown charcoal debris by water droplets and ice crystals nucleated in the rising fire plume can be manifested as black rain,. which fell at Hiroshima and Nagasaki (Ishikawa and Swain, 1981~. The water droplets and ice crystals that did not fall to the ground from the plume would eventually reevaporate, leaving behind their involatile cores. Many of these cores would contain one or more of the original smoke particles. In effect, the water particles can act as aggregation centers for the smoke. Not all fire plumes form thick condensation clouds. Those that do are usually associated with large intense fires in humid environments. Because the fire itself generates immense numbers of cloud condensation nuclei, the nature of the capping cloud can be very different from a normal cumulus (L. Radke, University of Washington, private communication, 1984~. Thus higher concentrations of smaller water droplets nucleated on smoke particles, ash, and windblown debris are expected. The total condensed water mass may not be significantly different from normal clouds, however (see Appendix 5-21. Strong updrafts in the fire plume may create substantial supersaturations (Twomey, 1977~. Hence smoke particles could be nucleated even though they tend to be hydrophobic when fresh (the nucleation of ~0.1-pm radius particles at supersaturations of .

67 <1 percent usually requires that the particle have at least a small hydroscopic component (Fitzgerald, 1973~. Radke et al. (1980a) noted that cloud condensation nuclei (can) concentrations in the Meteotron smoke plume were comparable to those in ambient air, even though the total number of particles was enormously enhanced. These results provide evidence that fresh soot is a poor can. If all of the smoke particles were to nucleate in the ascending fire plume, the concentration of droplets would exceed 105/cm3, resulting in a dense haze rather than a cloud. It is possible that only <10 percent of the submicron smoke particles are active as nuclei (Eagan et al., 1974), resulting in a more normal concentration of about 103 to 104/cm3 of small water droplets or ice crystals (the size and number depending on the quantity of water vapor entrained into the plume). The presence of a water cloud can alter the size {and spatial) distribution of smoke both through precipitation (by physical transport or removal of nucleated and collected smoke particles) and through coalescence of the collected smoke particles upon evaporation of the water. Precipitation removal from the fine plume is discussed under a separate heading in this chapter (see below). The flux of aerosol particles diffusing to a cloud droplet is given by {Twomey, 1977; Pruppacher and Klett, 1978), Fp = 4nRDpnp, where Fp is the number of particles impinging on the droplet per second, R is the droplet radius (in centimeters), ~ is the particle diffusion coefficient (about 10-5 to 10-6 cm2/s for particles of >O.l-pm radius), and np is the total particle concentration (number per cubic centimeters. Typically, Fp/np is of the order of 10-7 to 10~8 cm3/s for cloud droplets. If np is about 105/cm3, it might take about 1 h for an average droplet to collect 10 smoke particles. For a cloud droplet concentration of 1000/cm3, an aerosol depletion of about 10 percent would result. AS the droplet evaporated, the collected smoke would combine to form a single larger particle. Accordingly, cloud drop collection may on occasion be more efficient than coagulation in modifying the smoke particle size distribution in the early plume. In the long term, continuous processing of the smoke through transient water clouds would strongly affect the size distribution and trace composition of the smoke aerosols. L. Radke (private communication, 1984) may have observed this process of smoke transformation in a cap cloud over a prescribed forest fire. He compared smoke in air that passed through the cap cloud to smoke in air that passed under the same cloud. Size distribution measurements showed that the "cloud-processed" smoke was depleted by nearly a factor of 10 of particles smaller than about 0.05-pm radius, and by a lesser factor of particles larger than about 0.5-pm radius. In the size range from 0.1 to 0.5 Am, the particle populations were similar, although the apparent number mode radius increased from about 0.05 Am to about 0.1 um in the processed smoke. However, and most

68 critically, the mass of submicron particles was essentially the same in both air samples (as determined from the measured volume distributions) and the optical scattering coefficients had roughly the same magnitude (within 20 percent), indicating similar optical extinction efficiencies of the smoke aerosols. Ash and Debris Particle Scavenging. At Hiroshima, dry, dusty ash and other litter were observed to fall outside of the region of the black rain (Ishikawa and Swain, 19813. The mass of the larger debris particles in the fallout was, by implication, much greater than the mass of the attached submicron aerosol, suggesting relatively inefficient removal of smoke particles by scavenging on "fly ash. n It might also be inferred from such evidence that strong fire winds are capable of lifting substantial quantities of fine dust to high altitudes in addition to the smoke, a factor that has been ignored in the present analysis. The three potential agglomeration mechanisms for smoke particles in urban fire plumes (coagulation, cloud processing, and dust scavenging} are not explicitly taken into account in the present assessment. There is no simple way to quantify their effects on the smoke size distribution between emission at the fire and injection into the free atmosphere far downwind of the fire. However, by emphasizing observations of a variety of aged smokes, the adopted size distributions (and optical properties) already reflect the influence of these prompt physical processes. The smoke particle size distribution continues to evolve as the plume ages. Coagulation and cloud processing act to increase the particle size, while scavenging tends to remove the largest particles preferentially (Jaenicke, 1980~. Long-term coagulation has been included in some estimates of the optical and climatic effects of nuclear-generated smoke (Turco et al., 1983a,b; Crutzen et al., 1984}, and in the present baseline simulations. Optical Properties The optical properties of smoke clouds depend on the size distribution, composition, and morphology of the smoke particles. The optical properties determine the ultimate impact of nuclear smoke emissions on solar insolation and climate. There are two basic techniques for obtaining these properties--by direct measurement and by Mie scattering computation, which is accurate for spherical particles, but only approximate for nonspherical particles represented as Equivalent volume" spheres. In general, the assumption of volume-equivalent spherical particles tends to underestimate the absorptive power of nonspherical carbon particles (Roessler et al., 1983~. For a smoke aerosol the extinction optical depth, ~e, at visible and thermal infrared (8 to 20 ~m) wavelengths is the product of the specific extinction coefficient (or cross section), he (expressed in square meters per gram of smoke), the smoke mass

69 concentration, M (in grams per cubic meter of air), and the optical path length, L (in meters), or* Te = ~eML. The specific extinction coefficient is simply related to the specific absorption and scattering coefficients by He = Ha + as Thus only oe and oa need to be determined. The absorption optical depth may be calculated using an equation analogous to that for the extinction optical depth. In the visible spectrum, measurements for very pure carbon soots and carbon black yield values of He ~ 10 to 12 m2/g and Ha ~ 8 to 10 m2/g (Twisty and Weinman, 1971; Janzen, 1980; Chylek et al., 1981~. At the opposite extreme, smoke from burning wood can have <;e ~ 4 to 6 m2/g and oa ~ 0.01 to 2 m2/g, varying with the conditions of combustion (Patterson and McMahon, 1983~. The optical coefficients of smokes generated from plastics and other polymeric compounds exhibit values intermediate between those of carbon black and wood smoke, but generally lie closer to the carbon black values (Tewar son, 1982~. The extinction and absorption of smoke from hydrocarbon gas flames also approach the pure carbon values (Roessler and Faxvog, 1979~. For a variety of white smokes, he is observed to vary from about 2 to 5 m2/g, and for black smokes, from about 5 to 9 m2/g (Ensor and Pilat, 19711. The long-wavelength infrared properties of smoke clouds may affect the heat balance of the perturbed atmosphere. In the infrared, the specific extinction coefficients of most smokes lie between O.1 and 1 m2/g; that is, roughly one order of magnitude below the visible extinction values (Volz, 1972; O'Sullivan and Ghosh, 1973; Roessler and Faxvog, 1979, 1980; Uthe, 1981; Vervisch et al., 1981; Bruce and Richardson, 1983~. The infrared extinction is dominated by absorption. Factors that could enhance the infrared opacity of a dispersed smoke plume include the following: 1. The presence of large windblown flyash and debris particles. 2. Water condensed as droplets. 3. Infrared active gases such as H2O, CO2, and unburned hydrocarbons. 4. Aggregated particles exceeding ~0.5-pm radius. 5. Highly eccentric particles. ~- *Note that the extinction of a direct beam of light obeys the law, I/Io = exp(-~e), where To and I are the incident and attenuated beam intensities, respectively. However, light scattered from the beam adds a diffuse background illumination that increases the overall transmitted intensity.

70 On the basis of information currently available, however, it does not appear likely that any of these factors would critically influence the long-term climatic impacts of extensive smoke clouds. For example, ash particles rapidly fall out of the clouds, and water evaporates as the cloud disperses. The persistent infrared-active gases are not produced in large enough quantities to affect the cloud infrared opacity significantly (see Appendix 5-21. Nevertheless, further analyses of the smoke infrared problem would be helpful. For very pure carbon soot, the index of refraction inferred from optical measurements shows only a slight variation with wavelength in the visible and infrared spectrums. The real part of the index is 1.6 to 2.0, and the imaginary part* is -0.4 to -1.0 (Twisty and Weinman, 1971; Chippett and Gray, 1978; Pluchino et al., 1980; Tomaselli et al., 19813. The observed variability, particularly in the imaginary part, is probably due to varying amounts of oils (with refractive indices of ~1.5 - 0 i) mixed with the solid carbon. Pure graphite powder can have a refractive index as large as 2.7 - 2.0 i (Tomaselli et al., 1981~. For oily smoke, the imaginary index of refraction may be roughly estimated as the elemental carbon volume fraction of the smoke emission. For example, if the smoke is composed of 20 percent solid carbon by mass and 80 percent oils, then the carbon volume fraction is ~10 percent and the smoke particle imaginary index of refraction is about -0.1. C:alculations of the optical properties of smoke require knowledge of the size distribution, refractive indices, and morphological structure of the particles. Using measured values for these physical parameters, extinction and absorption coefficients can be predicted that agree closely with the directly observed coefficients (Chylek et al., 1980; Ackerman and Toon, 1981~. Such calculations also reveal the sensitivity of the optical properties to changes in the smoke parameters. For example, the specific extinction of visible light by smoke consisting of spherical particles is slightly sensitive to ~ (varying by roughly a factor of 2 for limiting y values between 1.5 and 2.5), and more sensitive to the mode radius (varying approximately as rm1 for mode radii >0.2 um). At a wavelength of 10 Am, the overall variation in the calculated specific extinction coefficient is about a factor of 2 for all mode radii up to about 1 um (i.e., for spherical smoke particles that lie within the Rayleigh extinction regime; Deirmendjian, 1969; Kerker, 19693. If the smoke consists of solid carbon particles suspended in droplets of oil or water, or if fine soot particles coat the surfaces of larger soil particles, the apparent specific absorption coefficient of the smoke at visible wavelengths can increase considerably (Ackerman and Toon, 1981), although its lifetime may be reduced. *The index of refraction of a bulk material has a dispersive component (the "real" part) and an absorptive component (the "imaginary" part). Real and imaginary are mathematical terms that derive from the equations of light propagation. The imaginary index of refraction is generally proportional to the absorption coefficient of the material.

71 As was mentioned earlier, coagulation and other particle aggregation processes alter the size distribution of smoke particles in plumes and clouds. Laboratory and field measurements suggest that, for a wide range of conditions, the effect of coagulation on the optical coefficients of smoke should be less than a factor of 2 (e.g., Seader and Ou, 1977~. In general, the evolution of the size distribution of the smoke particles must be taken into account in predicting long-term optical properties (Turco et al., 1983a,bl. Also, most existing optical theories assume spherical particles, which probably leads to an underestimate of the optical coefficients of sooty smokes or soot/water mixtures (e.g., Pagni and Bard, 1979; Janzen, 1980~. The baseline smoke optical properties are summarized in a later section. Basically, an oily smoke is adopted with oe = 5.5 m2/g and oa = 2.0 m2/g. Compared to observational data, these values seem to be conservative, perhaps making urban smoke look more like forest fire smoke, which is less absorbent (see below). In the baseline climate calculations of Chapter 7, the full evolution of the smoke particle size distribution and optical properties are treated using the microphysical/optical model of Turco et al. (1983a). Forest Fire Smoke Smoke emissions from burning forests have been measured extensively (Watson, 1951; McMahon and Ryan, 1976; Packham and Vines, 1978; Radke et al., 1978; Sandberg et al., 1979; Ward et al., 1979; McMahon, 1983; Radke et al., 1983, 19843. The particulates are typically composed of 40 to 75 percent benzene-soluble organic compounds. About 5 to 15 percent of the collected material is elemental carbon. The smoke from heading fires (moving with the wind) is yellowish to dark brown and oily, while the smoke from backing fires is black and sooty. Smoke emission factors for heading fires are ~1 to 6 percent of the fuel burned, and for backing fires, ~0.5 to 3 percent. The most recent observations by Radke et al. (1983, 1984) of smoke emissions from a series of carefully monitored prescribed forest burns found submicron particle emission factors averaging about 0.5 to 1.0 percent and total particle emission factors of 1 to 3 percent (but with a relatively high graphitic carbon content). It is observed that uncontrolled (wild) fires in forests produce about twice as much smoke as prescribed forest fires, per unit mass of fuel burned (Sandberg et al., 1979; McMahon, 19831. An average baseline smoke emission factor of 3 percent, consistent with the wildfire data, is assumed below. Irrespective of the fuel type, the maximum in the smoke particle number size distribution falls near 0.05-pm radius, and in the mass distribution, near 0.15 um (which implies a y value of ~1.8 to 1.9 in a log normal size distribution). Intercomparisons of size distributions measured by networks of ground samplers placed around natural fires, and by aircraft-borne instruments traversing wildfire plumes, reveal that the size distribution is generally preserved in the convective column and downwind of the fires (Sandberg et al., 1979~. Even hundreds of miles from large forest fires, the smoke particles

72 appear to remain unaltered (Watson, 1951~. However, Ward et al. (1979) observed a steady increase in the abundance of optically active aerosols in the aging plume of a prescribed backing fire. While there is great variability in forest fire properties and emission rates, the greatest particle yields occur for low-intensity smoldering fires, heading fires, and fires in green and nonwoody fuels (Sandberg et al., 1979~. The latest laboratory and field measurements support these general conclusions (McMahon, 1983; Patterson and McMahon, 1983; Radke et al., 1983~. Radke et al. (1983) recently detected a substantial fraction of supermicron particles in the plumes of prescribed forest fires up to several kilometers downwind. The mass mean diameter of the large particle population was about 10 to 20 Em, and the apparent emission index was occasionally as high as 0.04 gig of fuel. Particles and embers of millimeter size were also observed. These larger particles are generally of secondary importance because of the following: 1. Low number concentrations (<l/cm3~. 2. Short atmospheric residence times (against sedimentation and washout). 3. Negligible optical effects and small infrared effects. While of less interest here, the smoke emissions from prescribed fires in grass stubble and straw amount to about 1 percent of the material burned (Boubel et al., 1969~. Wild grass fires could emit twice this amount. The observed specific scattering coefficient of forest fire smoke at visible wavelengths is 2.6 to 7.0 m2/g (Evans et al., 1977; Tangr en, 1982; Radke et al., 1983~. The specific absorption coefficient of smoke generated by prescribed burns of palmetto leaves and pine needles is found to lie in the range of 0.04 to 2.8 m2/g (Patterson and McMahon, 1983~. These low optical absorptivities are consistent with a graphitic carbon mass fraction of about 5 to 10 percent in the smoke. Radke et al. (1983) measured graphitic carbon fractions of 10 to 15 percent in smoke from prescribed forest fires. Some wildfires might have considerably higher solid carbon contents and optical absorptivities (Sandberg et al., 1979; McMahon, 19831. While the infrared properties of wildland smokes are not well established, it has been observed that solar near-infrared radiation is much less attenuated by forest fire smoke than is visible radiation (Wexler, 19501. Such a result is consistent with smoke plumes dominated by submicron particles. The presence of fly ash (Radke et al., 1983) and water vapor (Appendix 5-2) in the plumes would increase their infrared opacity, particularly near the fire sources. However, over the long term, the infrared effects of wildfire smoke would appear to have little significance. The optical anomalies produced by forest fire smoke plumes provide some information on the properties of the smoke particles. By far, the most commonly reported optical effects are red and yellow-green suns (e.g., Lyman, 1918), yellow skies, dry fogs, and dark days (Plummer, 1912~. These effects are all consistent with heavy emissions of

73 submicron, oily smoke particles like those that have been collected in wildfires. Occasionally, blue suns and moons are observed through smoke clouds. The most notable case is the Alberta, Canada, forest fire of September 1950 tWexler, 1950~. The blue sun phenomenon has been interpreted in terms of an aerosol composed of particles of about 0.5-pm radius (Watson, 1952~. The aerosol could be formed by water vapor condensed on smoke particles. Even though Watson (1951) measured smoke particle sizes of about 0.05 um in air that had descended to the ground from the Alberta smoke plume, the water that may have been condensed aloft could have evaporated as the air subsided and warmed. An alternative explanation for the appearance of blue suns involves the contrast of an occluded solar disk against a smoky sky and the color sensitivity of the eye. However, this argument may not apply to the Alberta smoke pall tWexler, 1950~. In any case, blue suns viewed through fire plumes are rare. Satellite imagery of wildfire smoke plumes at visible and infrared wavelengths is available (Matson et al., 1984), and might be analyzed to determine the optical properties of the smoke clouds as they disperse over the course of several days. Fire Burning Times It is assumed that urban fires would burn out in about 1 day, with the period of most intense burning confined to several hours (FEMA, 1982~. Fires in dense fuel arrays might persist for several days. Hence most of the smoke injection following a nuclear exchange would occur over a relatively short time (assuming that the exchange itself would be executed within a few days}. Except in cases of unusually dry and windy weather, nuclear forest fires would probably burn out within 1 week. If the nuclear detonations were distributed over a period of weeks to months, the principal findings discussed here need to be revised. Such a concept of controlled nuclear war fighting is not widely accepted by nuclear strategists (see Chapter 3~. Smoke Injection Altitudes The heights at which nuclear smoke clouds would stabilize can be estimated by using observational data and buoyant plume theory. Anecdotal information concerning the ascent of large fire plumes is reviewed in Appendix 5-1. During intense wildfires and prescribed forest fires, plumes of smoke generally reach altitudes exceeding 1 km and can easily rise to 6 km or more (Wexler, 1950; Taylor et al., 1973; Eagan et al., 1974; Packham and Vines, 1978; Radke et al., 1978, 1983~. In the late, smoldering stages of wildfires, the smoke is generally confined to the lowest kilometer of the atmosphere. However, the smoke emission during this phase of the fire is of secondary interest here.

74 Large-scale (>l-km diameter) urban nuclear fires probably would deposit much of their smoke in the ~free" troposphere above the planetary boundary layer (Miller and Kerr, 19651. Deposition at these heights is supported by World War II experiences (e.g., Hamburg and Dresden) and by simulations of urban fire plume dynamics (e.g., Brode et al., 1982; Larson and Small, 1982a,b; Carrier et al., 1984; Cotton, 1984). Smoke that is injected at high altitudes tends to remain aloft as it disperses over large areas. (Thus the invention and extensive use of smokestacks to reduce surface pollution.) When the smoke can penetrate the boundary layer, or a temperature inversion at any level, it may remain aloft for very long periods. An excellent example of this is afforded by the Alberta, Canada, forest fire of September 1950 (Smith, 1950; Wexler, 19507. Most of the Alberta smoke remained well above an altitude of several kilometers, and eventually reached the tropopause, as it dispersed over an area of about 107 km2 during the course of a week. In the past, there have been many fires that created dark days and dry fogs over distances of hundreds to thousands of kilometers (Plummer, 1912~. The theory of buoyant plumes in a stratified atmosphere is described by Briggs (1975~. He reviewed observational data and theoretical treatments of the problem for heat sources up to 1000 MW. For smoke rising in still air, the height of the center of the plume may be estimated as ZC ~ 1/5Q1/4, where Zc is in kilometers, and Q is the heat source in megawatts. This relation holds for a constant temperature lapse rate of 6.5°C/km (an average value) and horizontal wind speeds of <5 m/s. In stronger wind fields, the plume center height is given approximately by ZC ~ 1/9Ql/3U~l/3, where U is the wind speed in meters per second. The vertical thickness of the plume is typically <50 percent of the center height. Measured plume heights and fuel consumption rates (for example, by Taylor et al. (1973) in an intense forest-slash fire) may be correlated very well with this simple plume theory. If the plume equations are extrapolated to large-scale urban fires, a rough estimate of potential smoke injection heights can be obtained (Martins, 1984~. Q is determined in this case by the relation, Q - AmOfbeJ tb x 104, where A is the fire area in square kilometers, mofb is the mass of combustible material burned per unit area (in grams per square centimeter), ec is the heat of combustion of the fuels (20,000 J/g), and tb is the burning time. For a city fire with A = 100 km2, m0fb = 3 g/cm2, and tb = 104 s, Q = 6 x 106 MW, and Zc ~ 10 km.

75 Carrier et al. (1984) modeled the hydrodynamics of a large-scale fire convective column and concluded that a 1 x 106 MW fire could generate a plume reaching 8 to 10 km in height. They suggested that, under special circumstances, the plume could go even higher. The urban fires expected in a full-scale nuclear war would be unprecedented in number and size. Based on observations and simulations of buoyant plumes, it appears that smoke from nuclear fires could be injected well into the free troposphere, and even up to the tropopause (Cotton, 1984; Manins, 1984~. There are a number of factors that could decreases the fire plume heights, and others that could increase them. Decreases could be caused by the following: 1. Stiff crosswinds and turbulence. 2. Strong temperature inversion and atmospheric stability. 3. Infrared radiative cooling under nighttime conditions. 4. Low-level atmospheric divergence. 5. Slow burning. Factors increasing plume heights include the following: 1. High ambient surface humidity (and latent heat release). 2. Marginal atmospheric stability or conditional instability. 3. Solar absorption and heating in the plume. 4. Low-level atmospheric convergence. 5. Rapid burning. Davies {1959) describes the plume of an oil refinery fire that apparently acquired considerable buoyancy by solar absorption. The additional energy seems to have accelerated the dispersion of the plume. The height distribution of the injected smoke, summed over all urban and forest fires, depends on the statistical distribution of the plume heights, each of which depends on fire area and intensity and meteorological conditions, among other things. There is no reason to believe, a priori, that the smoke would be preferentially injected near the ground or at high altitudes. However, common experience even with rather small fires suggests that the smoke would be channeled upward by the fire convection column to a distinct stabilization height. Because forest fires generate only a small fraction (about 17 percent) of the total smoke emission in the present baseline case, attention is focused here on urban fires. It might be assumed, as a reasonable statistic, that the number of urban fires with areas that fall into the interval A ~ A + dA varies as dA/A. This is equivalent to distributing the total fire area uniformly over equal intervals of area. Thus, for example, there would be equal total areas of 0- to 10-km2 fires, 10- to 20-km2 fires and 20- to 30-km2 fires. Such an assumption may in fact overestimate the number of smaller fires in a world having only about 1000 major urban centers. It can be shown that an A~1 fire area probability distribution and the plume height equations given above lead to a fairly uniform injection of smoke mass with altitude between 1 and 10

76 km (assuming fire areas ranging from about 0.03 km2 to about 300 km2, and the baseline fire parameters discussed earlier). Urban fires smaller than about 0.01 km2 would most likely occur in clusters over limited areas, and their plumes could be expected to interact strongly. During the early and late phases of a large urban fire, smoke is deposited at low altitudes. However, the fraction of fuel consumed during these phases is small, and the smoke produced can be entrained into the plumes of nearby fires burning in a more intense phase. In accordance with the crude analysis of plume dynamics just presented, the smoke generated in the baseline nuclear war can be assumed to be uniformly distributed by mass density (i.e., g-smoke/m3-air) between the ground and 9-km altitude. No smoke is injected above 9 km. If there were a preponderance of very larae scale · · . . . . city fires, the smoke injection would be skewed upward; of smaller, low-intensity fires, downward. Prompt scavenging of smoke from the fire plumes should have the greatest effect on the highest plumes; there should be very little effect on plumes below about 4 km. In the context of the present baseline case, this implies that additional smoke would be injected below about 4 km, not that less would be injected above about 4 km (see the scavenging discussion that follows). The sensitivity of the optical and climatic impacts of nuclear smoke to its height of injection is discussed by Turco et al. (1983a,b) and in Chapter 7. In general, the lower that smoke is injected into the atmosphere, the shorter its lifetime (under ambient conditions). Atmospheric turbulence continually mixes the smoke to higher and lower altitudes, driving the distribution toward a constant mixing ratio (g-smoke/g-air). However, precipitation rapidly depletes the smoke below about 4 km, creating a cleaner layer near the surface (Turco et al., 1983a,b). By contrast, physical mechanisms exist that may suppress the long-term smoke removal processes at higher altitudes and cause the smoke clouds to rise and spread rapidly from lower altitudes (Chapter 7~. The sensitivity of climatic impacts to smoke injection heights may thus be considerably smaller than expected. Although the energetics of large-scale fires and firestorms suggest that some smoke could readily be injected as high as the tropopause, there is less reason to believe that massive stratospheric smoke injections would occur (Murgai, 1976; see also Appendix 5-1~. Evidence pointing to stratospheric injection includes collection of wildfire smoke particles in the stratosphere (Cadre, 1972), and hydrodynamic calculations of fire plumes punching well into the stratosphere (e.g., Br ode et al., 1982; Cotton, 19841. Contrary evidence involves anecdotal eyewitness accounts of mass fire plume heights (see Appendix 5-1~; and analytical studies of fire plumes and fire whirls (e.g., Carrier et al., 1982, 1984~. Nevertheless, in cases of exceptionally powerful city fires under conditions of high surface humidity, strong local wind vorticity, and low atmospheric stability, some smoke could be injected directly into the stratosphere.

77 Water in Nuclear Clouds Nuclear explosion clouds and fire plumes can hold large quantities of condensed and gaseous water. The ~fire" water can effect the removal of smoke, and affect the optical and infrared properties of the clouds. A brief evaluation of the quantities of water involved and the potential impact is given in Appendix 5-2. Removal from the Plume Fine smoke and dust particles may be removed from the plumes of large, intense fires by collection in precipitation (black rain) and by attachment to the surfaces of debris lofted by the fire winds. Both phenomena were observed at Hiroshima and Nagasaki (Ishikawa and Swain, 1981) and may have occurred at the Hamburg firestorm. During major volcanic eruptions, wet and dry dust balls are often seen to fall out of the eruption clouds, although these phenomena may not be closely analogous to scavenging in fire plumes (Turco et al., 1983b). In their work, Turco et al. (1983a,b) discussed possible scavenging processes, such as nucleation and collection by rain and cloud drops, and concluded that up to 50 percent of the smoke could be removed from the plumes of intense fires and 25 percent from the plumes of other large urban fires (however, on the basis of observations, no significant early scavenging was assumed in the case of forest fire plumes). Crutzen et al. (1984) argued that smoke scavenging in fire plumes would be inefficient, but still assumed that one-third of the smoke would be removed initially. The nucleation of water on aerosols of >0.1 um in radius can occur at relatively low supersaturations (about 1 percent). This appears to be the principal mechanism of aerosol incorporation into natural cloud water (Leaitch et al., 1983; Radke, 1983) and probably into precipitation. Typical background submicron aerosol concentrations are <103/cm3. Hegg and Hobbs (1983) used field data to determine a nucleation scavenging efficiency of 63 ~ 53 percent for sulfate in cumulus, stratocumulus, and stratus cloud formations. However, Isaac et al. (1983) also emphasized the importance of the air circulation associated with cumulus clouds in redistributing trace atmospheric constituents. They noted in particular the high efficiency at which even small clouds can raise material over substantial height intervals within updrafts. In forest fire plumes, very high concentrations of cloud condensation nuclei (can) have been detected (Eagan et al., 1974; Radke et al., 1978~. However, in an oil fire plume, can concentrations were found to be at background levels (Radke et al., 1980a). In the rising convective column of a fire, only a small fraction of the smoke particles (whose concentrations exceed 105/cm3) are likely to be nucleated into cloud droplets large enough to coalesce as precipitation. For example, generously assuming that the mass of water in the plume is 1000 times the mass of smoke (i.e., oil and soot) particles (see the discussion in Appendix S-2), all of the smoke

78 particles could grow uniformly by water vapor condensation only to an average size of about 1-pm radius. The coalescence of water droplets is extremely sensitive to droplet size, and 1-pm droplets normally require many hours to form precipitation (Twomey, 1977~. In fires that are intense enough to generate strong convection, the fire winds also sweep up large quantities of supermicron debris particles (soil, ash, and char). Windblown debris has been observed in the plumes of forest fires (Radke et al., 1983) and was seen to fall out of the clouds at Hiroshima, Nagasaki, and other large World War II fires. This debris could preferentially nucleate to form rain and cloud drops. It is possible, for example, that the most significant contribution to the recorded black rain" events of World War II was the prompt washout and rainout of charred, windblown fire debris. Water vapor nucleation and subsequent condensational growth might directly affect up to about 10 percent of the submicron smoke particles in strong plume updrafts with rapid cooling rates; as already noted, freshly formed smoke particles tend to be hydrophobic, or water repellent, and thus are inherently poor cloud nuclei (Charlson and Ogren, 1982~. Smoke can also be collected by cloud droplets that nucleate, grow, and later coalesce into raindrops. However, as brought out in the previous discussion of the smoke size distribution, the lifetime of the submicron smoke particles against diffusional collection (in a cloud of about 1000 10-pm droplets per cubic centimeter) is likely to be an hour or more. On the other hand, the black rain is probably formed in the rising convective column within minutes of the smoke emission. This is demonstrated by the fact that the mass centroid of the induced rainfall at both Hiroshima and Nagasaki was over, but just downwind of, the fire area (Ishikawa and Swain, 1981~. Smoke particles can be scavenged by rapidly growing ice crystals and water droplets through phonetic, inertial, and electrostatic forces (Pruppacher and Klett, 1978~. However, the absolute efficiencies for the collection of submicron particles is not well defined. In one particularly interesting study, Prodi (1983) noted that, while submicron particles of sodium chloride, which is quite hydroscopic, were readily scavenged by rapidly growing ice crystals, submicron droplets of Car anuba wax, which is hydrophobic, were essentially unaffected. In the case of inertial capture of submicron aerosols by precipitation (rain and ice), the collection efficiencies are exceedingly low {Pruppacher and Klett, 1978), because the small particles tend to be deflected bv the airstream that flows around a falling hydrometeor, thus preventing direct collisions. Evaporating cloud and precipitation drops and ice crystals would collect submicron particles principally by thermophoresis - -- _ and electrostatic attraction (Pruppacher and Klett, 1978~. This process would apply, for example, to raindrops falling through a subsaturated air layer, or to cloud water or ice in an evaporating cumulus cap cloud or anvil. However, since the water is evaporating, transformation rather than removal of the smoke is generally implied.

79 Field measurements of the collection efficiencies of aerosols by rain and cloud drops are often 10 to 100 times larger than theory indicates (e.g., Radke et al., 1980b). Laboratory measurements of collection efficiencies are also generally larger than theoretical values (e.g., Leong et al., 1982~. Electrical forces may play a significant role in increasing the collection efficiency (Wang et al., 1978), particularly in the case of violent updrafts where electrification conditions similar to those in thunderstorms might develop. Observations of large droplet collection efficiencies have been discussed recently by Slinn (1983), but their cause remains in dispute. The cloud/precipitation collection efficiencies are lowest for aerosols in the size range from about 0.1 to 1.0 um, the so-called ~Greenfield gap" (Greenfield, 1957~. Barlow and Latham (1983) estimated a submicron aerosol half-life in a 2 mm/in rainfall (at relatively humidities of 50 to 70 percents of 7 to 70 h, corresponding to collection efficiencies* of 10- to 10-3 (for charged and uncharged particles, respectively). In a thunderstorm, with an estimated raindrop collection efficiency of 0.1, Barlow and Latham estimated an aerosol half-life of about one-half hour. Turbulence could enhance the rate of smoke scavenging by cloud drops. However, in the region of the fire plumes where condensation and precipitation occur, the turbulence field should not be any more intense than in natural convective clouds. Rosenkilde and Serduke (1983) showed that such turbulence would not significantly augment the aerosol removal rate. Even after the smoke particles are captured in cloud droplets, they must still be removed in precipitation. There are only a few well-documented cases of significant prompt rainfall from fire plumes (e.g., Hiroshima). Mordy (1960) describes the singular lack of induced precipitation during the massive burning of sugar cane fields in Hawaii. In an air parcel rising over a large fire, the time available for the formation of precipitation is quite short (less than one-half hour). The cloud droplets may also be unnaturally small in size. Accordingly, the situation in a fire plume is fundamentally different from that in a natural convective/precipitation system, in that the buoyant instability is created in large part by ~dry" heat, and the rising air is seeded with unusually high numbers of cloud condensation nuclei. Typically, the ambient atmosphere in the vicinity of a nuclear fire would be found in a relatively stable initial state (i.e., without strong local convection and storms). During natural precipitation events, the half-lifetime of cloud droplets against removal as rain lies in the range of 103 to 104 s (Pruppacher and Klett, 1978~. In fire plumes, longer droplet lifetimes could be expected. For the baseline calculations, it is assumed that, on the average, 50 percent of the smoke emissions from urban fires is promptly removed *The collection efficiency is defined as the fraction of aerosols within the volume traced out by a falling raindrop that collide with and adhere to the drop.

80 by precipitation in the fire plumes. This estimate is conservative with respect to the previous discussion, which suggests that precipitation would not occur at all fires, and that removal efficiencies for submicron smoke particles could be quite low. Prompt scavenging reduces the overall baseline smoke emission factor from 0.04 g/g to 0.02 g/g for urban fires. The latter figure is utilized below to make smoke emission estimates. In reality, smoke scavenging might be concentrated in the most intense fires with the tallest convective columns, under conditions of high ambient surface humidity. However, lacking detailed quantitative information on the simultaneous probability of all of these conditions, the precipitation removal is applied uniformly to all urban smoke plumes. Some smoke would also be redistributed by the cloud and precipitation processes, but this effect · · IS lgnOreCt. Finally, it should be mentioned that the long-term removal of dispersed smoke by precipitation is explicitly taken into account in the baseline climate calculations (Chapter 7~. Global removal by wet deposition is the principal sink for the smoke that escapes from the fire plumes (Jaenicke, 1980; Charlson and Ogren, 1982; Turco et al., 1983c). Removal processes and typical atmospheric lifetimes for smoke particles are discussed in Chapter 7. ESTIMATING SMOKE EMISSIONS IN A MAJOR NUCLEAR EXCHANGE Baseline Estimates The nuclear war scenarios considered in this report are highly generalized. No detailed information is given regarding explosion yields or heights of burst for specific targets, or the duration of the exchange. The baseline scenario utilizes 6500 Mt. of which 1500 Mt is targeted on urban areas (Chapter 3), where key military and industrial targets are located (e.g., Kemp, 1974; Ball, 1981~. Moreover, it is reasonable to assume that a substantial fraction of the remaining 5000 Mt would be detonated over targets near forests, brush, and grass lands. Although the baseline case in this study does not postulate a season for the exchange, for the purposes of calculation of wildfire smoke, summer season values are used (urban fires, the principal source of smoke, would be largely independent of seasonal variation). The baseline smoke emission estimates are given here, and excursions from the baseline are discussed in the next subsection. As outlined earlier, it is assumed that 250,000 km2 of urbanized area is partially burned, which corresponds to 50 percent of the total urbanized area in the countries at war. Such an area could be ignited by 1500 Mt of air bursts, assuming an average ignition area of ~250 km2/Mt, no fire spread, one-third overlap of ignition zones, and no spreading beyond the 20 cal/cm2 ignition zone. Within the fire area, the average burden of combustible material is taken to be 4 g/cm2, and three-quarters of this is assumed to burn, in accordance with other estimates of urban fire damage in a nuclear attack (e.g., Miller et al., 1970; FEMA, 1982; Brode and Small, 1983~. Finally, the net smoke

81 emission factor is assumed to be 0.02 gig (grams of smoke per gram of fuel consumed) after scavenging and removal by coagulation and condensation processes in the convective fire plumes is taken into account (50 percent removed). Multiplying the appropriate factors together, the total urban smoke emission amounts to ~150 Tg (1.5 x loll g, Forest fires are also estimated to burn 250,000 km2 (i.e., roughly the area of irradiation at >20 cal/cm2 by 1000 Mt of air bursts). The basis for this estimate is discussed earlier in this chapter. The fuel consumed in forest fires is taken to be 0.4 g/cm2 (about 20 percent of the typical fuel loading), and the net smoke emission factor is taken to be 0.03 g/g, both values based on observations. Brush and grass fires, whose emissions are smaller per unit area burned, are not explicitly included in the analysis. The total forest fire smoke emission is then ~30 Tg. In winter, wildfire emissions might be reduced to a few teragrams; however, because urban fires contribute much more soot, the total emission would be reduced by no more than 20 percent. The composition and optical properties of the smoke in the baseline model must also be specified. Even though urban fires dominate the aggregate smoke emission in the baseline case, with potential soot fractions of up to 90 percent, it is assumed that the average graphitic carbon fraction is only 20 percent (compared to -10 percent in forest fire smoke). The smoke particle number size distribution is taken to be log normal with a number mode radius* of 0.1 um and ~ = 2.0; the effective particle density is 1 g/cm3. The smoke index of refraction is 1.55 - 0.1 i. At visible wavelengths the corresponding smoke specific extinction coefficient can be taken as 5.5 m2/g, and the specific absorption coefficient as 2.0 m2/g. The smoke infrared extinction and absorption coefficients (at 10 um) are both roughly 0.5 m2/g. These physical constants provide a consistent set for optical (Mie) calculations. Because the selected baseline optical extinction and absorption coefficients are much smaller than typical values for sooty {urban) smokes, the effect of "aging, n which can reduce the optical efficiency of the smoke, may be neglected in carrying out approximate optical-effects simulations. The optical efficiency is otherwise expected to decline with time. The smoke parameters for the baseline nuclear war scenario are summarized in Table 5.7. The total estimated smoke emission is 180 Tg, caused by roughly 30 percent of the nuclear explosions. The estimated smoke emissions are very uncertain, however; some of the sources of uncertainty are discussed below. The total quantity of combustibles consumed in the baseline war scenario is 8500 Tg (7500 Tg in urban fires and 1000 Tg in forest fires). For the urban flammables, about 5000 Tg of cellulosics, 1500 Tg of liquid fossil organics, and 1000 Tg of industrial organochemicals, plastics, polymers, rubbers, resins, etc., are burned. The *For volume-equivalent spherical particles.

82 0 a~ COe ~: 4 :~ V z a) V`e S U]e Ye U]e ·rl m E~ tQ o e~ ~e V ~3 e~ U'e m e~ - y ~ _ 1 Y O O O ~~ O tf, 1 - o C~ ~ 1 O O O U~ eD O eD I _I d' ~ ~ U-'- 1 1 1 O O O C~ ~ - y o o O O O· O · O O 1 o 1 _I u~ O ' · · 1 1 0 1 o 1 O _O O~ O · ·· ~ ~ 0 - 0 0 0 ~ o V? °e o u~ O O O O O O _ ,,~, t-, a~ c ~° V? o o U. V' ~e ~Y C ~Q Ei e ~Y ~E~ ~t~) ~l O) V ~ l ~O Ue. _ ' ~ E ~ O~ ~ Y U] E ~ ~ 03V (1) Oe O rl ~V ~ e'_| ~e E ~ ~ ~3 ~e U] ~O ~ ~O ~e ~O e' l O ~e e ~eC ~>1 e ~55 ~et_| ~ < ~e ~eC ~ t ~O ~Q e ~V ~O ~ Ye Q. ~O ~ e ~V ~ Ye ~l ~O ~e e~ 0) ~e 41J ~~ l ~tn e~ ~Ue) ~e~ (U Y ~e ~ ~ ~ ~COe ~ ~Cle, ~n5 O Y 0~ - c- - ~ m- - ~ ~ ~ ~ ~ · - O ~ £- - ~ ~ ~ ~ ~ O ~ ~ ~ U]e `1) V ~l O ~ eO ~ V ~ ~ Ue~ V ~l ~ V ~ 6) U] ~ ~ ·-l ~ G) ~J _I eY (V Q (1) I~ y Il ~ (l) ~--1 eY dV ~l (1) S.J · - O ~ S ~ ~ O ~1 · - ~ ~ · - ~ · - ~ Q' ~ 3 ~ ~ ~| Q~e ~O Ue ~ U]e tU ~U Ue ~V ~n 0 ~ ~ ~ ~ te Ue, _1 Q ~ ~e) Ge, tQ 0 41) 41) ~ ~ft {~ ~4) 4) ~eY Y~ Q ~O S~ eQe Q Q Q S ~s ~4 ~O O ~1 0 0 )~ ~4 4 ~0 0 0 0 ~ 3 ~I E ~E ~e 3 3 ~3 ~4 Ecl 1 ~E4 U3e U)e e' | eCO O eQe ~O ns1 Vt~e . V · - · - S · - e4 dP o 1 1 eD eC o £?e V o C~ eD O · O O O · O · ° 1 ~ ~ 1 1 1~0 1 1 1 eD O ~O O C~ · · C~ · ~ O ~ ~ O V ·^ · - O S Q ~lOe~ _1 0 ~-~1 1 0 1~ 0 eD _I O eD O eD dP dP · ..... O O _I O C~ eD {~ O C~ c0 U?e tn V U'e · - JJ O ::, o V Q ·rl Ue, e~ Q. ~e · - S ·^ S~ S e~ Q' Ue, ~ eO O ~P. `: ·rl ~ O O ~ eD 41) eD V L. ~ · ~ O O a, 0 eD . - S ~ ~ J~ dJ 41) ~ · - 3 CGe _I _1 (IJ e-1 V ~ et_| 0) O e'_ V ~l O U ~e Q C) e~ erl V o V C~ C~ V V ~ ~ _ _ _= V - N 'Q Q ~ 4~ erl1 S e~ e~l 4 _I 0 ~ U] U] Q ~ e~ e~ ~ .~1 ~- - ~ >- - ~e GS ~ ~ ~ ~ ·~ ·~ ~ ~ _I C C eC -- (I, ~ O O O ~ ·~_| ·- e~ X Ll O V Q. C 4 \_l ~eQ~ O O e ~55~ J~ U?e 0 eO O X Q Q ~ ~-I ~ IT' ~ e'_| ~ ~ V V V _~ e~ e~ e~ V V V ~ - - e~ e~ e~ e~ e~ V V V Q4 S-I Q~ Q~ ~e U] Ue~ ~ 41) ~ (D y Y Ye eY eY O~ ~O ~O ~Q ~O U] U e~ C~?e U3e U)e e~ V tle,

83 corresponding total energy release is about 5 x 1019 cal, or 50,000 Mt. assuming an average heat of combustion of 6000 cal/g. (Note, by comparison, that one day's solar insolation amounts to about 3,000,000 Mt of energy.) The energy release drives the buoyancy of the fire plumes and may create strong surface winds. Because the initial nuclear detonations over cities would pulverize large quantities of masonry and plaster into fine dust, it is likely that a significant burden of submicron particulates would be drawn up into the fire plumes. Even if only 1000 tons of fine (submicron) dust were raised for each megaton of thermal energy released, the dust injection could total 30 Tg. However, because there are few data pertaining to this source of particulates, it is ignored in the baseline assessment; future consideration seems worthwhile. As was discussed earlier, the smoke mass insertion is assumed to be uniform with height between the ground and 9-km altitude, and to occur over a period of several days to 1 week. Excursions from the Baseline Case In order to place some limits on the possible range of smoke emissions in the baseline scenario, reasonable excursions of the fire parameters are investigated. These excursions are not meant to represent an absolute range of possibilities, but a range that seems to be consistent with current scientific knowledge. In the case of urban fires, the area burned is varied between 25 percent and 75 percent of the urbanized area of the NATO and Warsaw Pact countries (neglecting possible urban damage in other industrialized nations such as China and Japan), the net smoke emission factor is varied between 0.01 g/g and 0.04 g/g, and the fuel burden is varied between 2 g/cm2 and 4 g/cm2. None of these assumptions appears to be extreme. The resulting urban smoke emission varies from ~20 Tg to ~450 Tg. This range of emissions is in rough accord with the range estimated by Broyles (19841. In the case of forest fires, it is assumed, on the low side, that no smoke emissions would occur. On the high side, a fourfold increase in the burned area and a smoke emission factor of 0.05 g/g are assumed, yielding a forest smoke emission of ~200 Tg. Accordingly, the present estimate of a potential range of smoke emissions following the baseline nuclear exchange is ~20 to ~650 Tg. This is not an uncertainty range for the emission, but an excursion range based on plausible parameter variations. Sources of uncertainty in these estimates are discussed in the next section. Because it is possible that the smoke plumes of massive urban fires would penetrate into the stratosphere, it is worthwhile to consider the implications of smoke injections in the lower stratosphere. The injection of up to 10 Tg of smoke (just over 5 percent of the baseline estimate) between 12 and 20 km at northern mid-latitudes may be assumed. Even though such an injection is not included in the baseline calculation, it represents a potentially interesting excursion {Turco et al., 1983a,b). Turco et al. (1983a,b) pointed out that massive smoke emissions

84 would be possible in nuclear exchanges that involved only a limited total yield detonated over or near major urban centers. This conclusion is based on the observation that most urban areas tend to have dense "cores" in which combustible materials are concentrated. Thus about 100 Mt (say, in 50- and 100-kt weapons) would be sufficient to attack all of the major urban centers in the NATO and Warsaw Pact countries. Such a purposefully destructive strategy is currently thought to be unlikely. However, an equivalent result is possible. For a scenario of any size in which 100 Mt of explosions were to burn an urban area of 25,000 km2 (about 50 percent of the city cores of the combatant nations), consume 20 g/cm2 of combustibles, and emit 2 percent (net) of the burned mass as particulate in the process, ~100 Tg of smoke would be generated. This is similar to the baseline urban smoke emission of 150 Tg. However, the emission would be patchier for a longer time in the 100-Mt case due to a reduced number of smoke sources. In accordance with the estimates presented above, one may deduce that smoke emissions from nuclear-initiated wildfires scale very roughly with the total yield of the exchange, including tactical weapons, and are very sensitive to season, with maximum emissions in summer and early fall and minimum emissions in winter. Smoke production by urban fires, on the other hand, may be rather insensitive to total yield, if the urban centers, or the military and industrial sites within urban zones, are systematically targeted. The effect of seasonal and meteorological conditions on nuclear urban fires (as with everyday urban fires) is also less important, owing to the general protection of urban combustibles from the weather. Optical Depth Excursions Given the range of smoke emissions just described and the possible variations in smoke optical properties summarized in Table 5.7, ranges of average optical depths (at visible wavelengths) can be calculated. Assuming hemispherical dispersion of the smoke, the baseline extinction optical depth is 4, and the absorption optical depth is 1.4. If the baseline smoke optical constants are accepted, the hemispherical extinction optical depth can range from 0.44 to 14.3; the optical depth for smoke confined to the northern mid-latitudes can range from 1 to 36. The corresponding absorption optical depths are about one-third of these values. Taking into account possible variations in the smoke optical constants, the optical depth range is even greater. Significant radiative and climatic perturbations might be expected whenever the hemispheric-scale optical depth is >1. volcanoes, which generate only nonabsorb~ng aerosols, can produce noticeable global disturbances at optical depths of about 1 (Chapter 81. Accordingly, the major segment of the optical depth range derived above can lead to serious environmental effects (see Chapter 7 for an exposition of these effects). However, given the large plausible ranges of fire and smoke parameters, subcritical optical depths clearly lie within the range of uncertainty.

85 UNCERTAINTIES Uncertainties are recognized in each of the key parameters pertaining to fires and smoke emissions in a nuclear war. Although only very rough estimates of the uncertainties may be deduced, even these may be useful in evaluating the weaknesses in current knowledge. Accordingly, a subjective assessment of uncertainties, based on consideration of the limited set of data available to the committee, is spelled out below. 1. The areal extent of nuclear urban fires per megaton of yield (factor of 2 to 3~. Potential overlap of fire zones, and fire spread, dominates the uncertainty. 2. Quantities and distributions of flammable materials in cities and surrounding areas (factor of 3 in the average central-city fuel burden, factor of 2 in the average suburban fuel burden, factor of 3 in the worldwide urban-area average fuel burden). 3. Urban smoke emissions per unit mass of combustible loading (factor of 2 in the fraction of fuel burned in urban nuclear fires, factor of 2 to 3 in the quantity, or mass, of smoke generated per unit mass of material burned, factor of 3 in the graphitic carbon mass fraction, factor of 2 to 3 in the mean particle size, and factor of 1.5 in the average particle bulk density). 4. Optical (visible wavelength) properties of urban fire smoke (factor of 2 in the specific extinction and scattering coefficients (square meters per gram), factor of 3 in the specific absorption coefficient (square meters per gram), factor of 3 in the imaginary part of the refractive index). 5. Infrared properties of urban fire smoke (factor of 3 in the late-time specific extinction/absorption coefficient; factor of 5 in the early-time extinction/absorption coefficient which may be controlled by condensed water and fly ash). 6. The areal extent of nuclear forest fires (factor of 3 to 4, neglecting sensitivity to the explosion scenario). 7. Forest fire smoke emissions per unit area burned (factor of 2 to 3 in the fraction of biomass fuel consumed, factor of 2 in the mass of smoke emitted per unit mass of fuel burned, factor of 3 in the graphitic carbon mass fraction, factor of 2 to 3 in the mean particle size, and factor of 1.5 in the average particle bulk density). 8. Optical (visible wavelength) properties of forest fire smoke (factor of 1.5 to 2 in the specific extinction and scattering coefficients (square meters per gram), factor of 3 in the specific absorption coefficient (square meters per gram), factor of 3 in the imaginary part of the refractive index). 9. Infrared properties of forest fire smoke (factor of 2 to 3 in the specific extinction/absorption coefficient at intermediate and late times). 10. Heights of smoke plumes from mass nuclear urban and forest fires (factor of 1.5 to 2 in both cases). 11. Extent of precipitation scavenging (black rain) and coagulation in the most intense fire plumes (the overall precipitation scavenging efficiency could vary from 25 to 75 percent; the reduction

86 of the optical extinction and absorption coefficients by prompt coagulation in the densest plumes could vary from 20 to 50 percent). 12. Quantity of submicron masonry dust raised in urban fire plumes following pulverization of buildings by nuclear blast (injection of O to 105 tons/Ml of explosive yield); the extent of smoke production from burning aluminum and other "nonflammable" materials in very intense fires is unknown. 13. Effect of massive smoke emissions on the subsequent meteorology and particle removal rates (factor of 3 to 10; see Chapter 7~. The uncertainty factors defined above cannot simply be multiplied to estimate absolute ranges of equally likely values for composite parameters such as smoke emissions and optical depths. The factors do not correspond to intervals of statistical significance, in which the central (or baseline) values are the most probable values. Because the various smoke parameters are largely uncorrelated, the uncertainty in combinations of the parameters must be deduced by statistical means. A precise determination of the overall uncertainty in the smoke emission and optical depth estimates cannot be made at this time, because the nature of the statistical dispersion has not yet been ascertained. The propagation of uncertainty into the radiative transfer and climate calculations has an exponential component, because those calculations involve terms of the form, emu. Using the present baseline case as a reference, an increase in the smoke emissions would have less impact than a decrease, inasmuch as the light absorption by the smoke is already about 90 percent, averaged over the northern hemisphere. The duration of significant effects would be prolonged, however. Patchiness, or light leakage through "holes in the smoke clouds, also has an exponential dependence. Nevertheless, average smoke optical depths of even -1 would still imply major perturbations of the postwar environment (for example, volcanic scattering optical depths ~1 can produce significant climate anomalies). The climatic aspects of the light transmission problem are discussed in Chapter 7. Turco et al. (1983a,b} carried out a large number of sensitivity tests in which the physical parameters of smoke and dust and the explosion scenarios were varied to investigate the nature of the uncertainty in the smoke emission, light transmission, and climate variation. They concluded that as many uncertain factors could act to aggravate the effects as could act to ameliorate them. SUMMARY A full-scale nuclear exchange of 6500 Mt. involving a variety of military and urban targets, would ignite numerous fires and could generate as much as 180 Tg of smoke. Considering the substantial uncertainties involved in estimating the smoke emission, however, the plausible range of emissions extends from 20 to 650 Tg. The optical properties of the dispersed smoke clouds have been deduced principally

87 from observational data. At visible wavelengths, a specific extinction coefficient of 5.5 m2/g and a specific absorption coefficient of 2.0 m2/g are selected for optical and climate calculations. The infrared extinction coefficient is an order of magnitude smaller than the visible extinction coefficient. The baseline optical absorptivity is conservative (on the low sidel, in view of the strong absorption of light by typical sooty smokes. Even so, the implied disturbances in solar transmission on a global scale appear to be serious. REFERENCES Ackerman, T.P., and O.B. Toon (1981) Absorption of visible radiation in atmosphere containing mixtures of absorbing and nonabsorbing particles. Appl. Opt. 20:3661-3668. Ayers, R.U. (1965) Environmental Effects of Nuclear Weapons. Vols. 1-3. Report HI-518-RR. Harmon-on-Hudson, N.Y.: Hudson Institute. Backovsky, J., S.B. Martin, and R. McKee (1982) Experimental Extinguishment of Fire by Blast. Report PYU-334. Menlo Park, Calif.: Stanford Research Institute. 97 pp. Baldwin, R. (1968) Flame merging in multiple fires. Combust. Flame 12:318-324. Ball, D. (1981) Can Nuclear War Be Controlled? Adelphi Paper 169. London: International Institute Strategic Studies. Bankston, C.P., B.T. Zinn, R.F. Browner, and E.A. Powell (1981) Aspects of the mechanisms of smoke generation by burning materials. Combust. Flame 41:273-292. Barlow, A.K., and J. Latham (1983) A laboratory study of the scavenging of sub-micron aerosol by charged-raindrops. Pages 551-560 In Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Baum, H.R., and G.W. Mulholland (1979) Coagulation of smoke aerosol in a buoyant plume. J. Colloid. Interface Sci. 72:1-12. Benech, B. (1976) Experimental study of an artificial convective plume initiated from the ground. J. Appl. Meteorol. 15:127-137. Benech, B., J. Dessens, C. Charpentier, H. Sauvageot, A. Druilhet, M. Ribon, P.V. Dinh, and P. Mery (1980) Thermodynamic and microphysical impact of a 1000 MW heat-released source into the atmospheric environment. Pages 111-118 ~n Proceedings of the Third WMO Scientific Conference on Weather Modification. Geneva: World Meteorological Organization. Boubel, R.W., E.F. Darley, and E.A. Schuck (1969) Emissions from burning grass stubble and straw. J. Air Pollut. Control Assoc. 19:497-500. Briggs, G.A. (1975) Plume rise predictions. Pages 59-111 ~n Chapter 3 of Lectures on Air Pollution and Environmental Impact Analyses. Boston, Mass.: American Meteorological Society. Brode, H.L., and R.D. Small (1983) Fire Damage and Strategic Targeting. PSR Note 567. Santa Monica, Calif.: Pacific Sierra Research Corp. 58 pp.

88 Brode, H.L., D.A. Larson, and R.D. Small (1982) Time-Dependent Model of Flows Generated by Large Area Fires. PSR Note 483. Santa Monica, Calif.: Pacific Sierra Research Corp. Broido, A. (1960) Mass fires following nuclear attack. Bull. At. Sci. 16:409-413. Broyles, A.A. (1984) Smoke generation in a nuclear war. Submitted to Am. J. Phys. Bruce, C.W., and N.M. Richardson (1983) Propagation at 10 um through smoke produced by atmospheric combustion of diesel fuel. Appl. Opt. 22:1051-1055. Butcher, S.S., and M.J. Ellenbecker (1982) Particle emission factors for small wood and coal stoves. J. Air Pollut. Control Assoc. 32:380-384. Cadle, R.D. (1972) Composition of the stratospheric sulfate layer. Eos 53:812-820. Calcote, H.F. (1981) Mechanisms of soot nucleation in flames--A critical review. Combust. Flame 42:215-242. Carrier, G.F., F.E. Fendell, and P.S. Feldman (1982) Firestorms. TRW Report 38163-6001-UT-00. Redondo Beach, Calif.: TRW Systems. 83 pp. Carrier, G.F., F.E. Fendell, and P.S. Feldman (1984) Big fires. To appear in Combust. Sci. Technol. Chandler, C.C., T.G. Storey, and C.D. Tangren (1963) Prediction of Fire Spread Following Nuclear Explosions. Research Paper PSW-5. Washington, D.C.: U.S. Forest Service. 110 pp. Charlson, R.G., and J.A. Ogren (1982) The atmospheric cycle of elemental carbon. Pages 3-18 ~n Particulate Carbon: Atmospheric Life Cycle, edited by G.T. Wolff and R.L. Klimisch. New York: Plenum Press. Chippett, S., and W.A. Gray (1978) The size and optical properties of soot particles. Co~'bust. Flame 31:149-159. Church, C.R., J.T. Snow, and J. Dessens (1980) Intense atmospheric vortices associated with a 1000 MW fire. Bull. Am. Meteorol. Soc. 61:682-694. Chylek, P., V. Ramaswamy, and M.K.W. Ko (1980) Effects of aerosol size distributions on the extinction and absorption of solar radiation. Pages 225-235 ~n Environmental and Climatic Impact of Coal Utilization, edited by J.J. Singh and A. Deepak. New York: Academic Press. Chylek, P., V. Ramaswamy, R. Cheng, and R.G. Pinnick (1981) Optical properties and mass concentration of carbonaceous smokes. Appl. Opt. 20:2980-2985. Cotton, W.R. (1984) A simulation of cumulonimbus response to a large firestorm--Implication to a nuclear winter. Submitted to Science. Crutzen, P.J., and J.W. Birks (1982) The atmosphere after a nuclear war: Twilight at noon. Ambio 11:114-125. Crutzen, P.J., C. Brohl, and I.E. Galbally (1984) Atmospheric effects from post-nuclear fires. Climatic Change, in press. Culver, C.G. (1976) Survey Results for Fire Loads and Live Loads in Office Buildings. Report NBS BSS-85. Gaithersburg, Md.: National Bureau of Standards. 157 pp.

~9 Dasch, J.M. (1982) Particulate and gaseous emissions from wood-burning fireplaces. Environ. Sci. Technol. 16:639-645. Davies, R.W. (1959) Large-scale diffusion from an oil fire. Pages 413-415 in Atmospheric Diffusion and Air Pollution, edited by F.N. Frenkiel and P.A. Sheppard. New York: Academic Press. Day, T., D. MacKay, S. Nadeau, and R. Thurier (1979) Emissions from in situ burning of crude oil in the Arctic. Water Air Soil Pollut. 11:139-152. Deirmendjian, D. (1969) Electromagnetic Scattering on Sperical Polydispersions. New York: Elsevier. 290 pp. Dessens, J. (1962) Man-made tornadoes. Nature. 193:13-14. Eagan, R.C., P.V. Hobbs, and L.F. Radke (1974) Measurements of cloud condensation nuclei and cloud droplet size distributions in the vicinity of forest fires. J. Appl. Meteorol. 13:553-557. Ensor, D.S., and M.J. Pilat (1971) Calculation of smoke plume opacity from particulate air pollutant properties. J. Air Pollut. Control Assoc. 21:496-501. Evans, L.F., I.A. Weeks, A.J. Eccleston, and D.R. Packham (1977) Photochemical ozone in smoke from prescribed burning of forests. Environ. Sci. Technol. 11:896-900. FEMA (1982) What the planner needs to know about fire ignition and spread. Chapter 3 ~n FEMA Attack Manual. Publication CPG 2-1A3. Washington, D.C.: Federal Emergency Management Agency. Fitzgerald, J.W. (1973) Dependence of the supersaturation spectrum of CCN on aerosol size distribution and composition. J. Atmos. Sci. 30:628-634. Glasstone, S. (1957) The Effects of Nuclear Weapons. Washington, D.C. U.S. Department of Defense. Glasstone, S., and P.J. Dolan (eds.) (1977) The Effects of Nuclear Weapons. Washington, D.C.: U.S. Department of Defense. 653 pp. Goodale, T. (1971) The Ignition Hazard to Urban Interiors During Nuclear Attack due to Burning Curtain Fragments Transported by Blast. Report URS-7030-5. San Mateo, Calif.: URS Research Corp. 25 pp. Greenfield, S.M. (1957) Rain scavenging of radioactive particulate matter from the atmosphere. J. Meteorol. 14:115-125. Hegg, D.A., and P.V. Hobbs (1983) Preliminary measurements of the scavenging of sulfate and nitrate by clouds. Pages 79-89 ~n Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Hilado, C.J., and A.M. Machado (1978) Smoke studies with the Arapahoe chamber. J. Fire Flammability 9:240-244. Hill, J.E. (1961) Problems of Fire in Nuclear Warfare. Report U 120869. Santa Monica, Calif.: RAND Corp. Huschke, R.E. (1966) The Simultaneous Flammability of Wildlands and Fuels in the United States. RM-5073-TAB. Santa Monica, Calif.: RAND Corp.

Do Isaac, G.A., J.W. Strapp, H.A. Wiebe, W.R. Leaitch, J.B. Kerr, K.G. Arlauf, P.W. Summers, and J.I. MacPher son (1983) The role of cloud dynamics in redistributing pollutants and the implications for scavenging studies. Pages 1-13 in Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Ishikawa, E., and D.L. Swain (Transl.) (1981) Hiroshima and Nagasaki, The Physical, Medical and Social Effects of the Atomic Bombings. New York: Basic Books. 706 pp. Issen, L.A. (1980) Single-Family Residential Fire and Live Loads Survey. Rep. NBSTR 80-2155. Gaithersburg, Md.: National Bureau of Standards. 176 pp. Jaenicke, R. (1980) Atmospheric aerosols and global climate. J. Aerosol Sci. 11:577-588. Jagoda, I.J., G. Prado, and J. Lahaye (1980) An experimental investigation into soot formation and distribution in polymer diffusion flames. Combust. Flame 37:261-274. Janzen, J. (1980) Extinction of light by highly nonspherical strongly absorbing colloidal particles: Spectrophotometric determination of volume distributions for carbon blacks. Appl. Opt. 19:2977-2985. Jaycor, (1980) Forest Environment (U). Report 5477 F. Washington, D.C.: Defense Nuclear Agency. Kanury, A.M. (1976) The science and engineering of hostile fires. Fire Res. Abstr. Rev. 18:72-96. Kemp, G. (1974) Nuclear Forces for Medium Powers. I. Targets and Weapons Systems. Adelphi Paper 106. London: International Institute of Strategic Studies. 41 pp. Kent, J.H., and H.G. Wagner (1982) Soot measurements in laminar ethylene diffusion flames. Combust. Flame 47:53-65. Kerker, M. (1969) The Scattering of Light and Other Electromagnetic Radiation. New York: Academic Press. Kerr, J.W. (1971) Historic fire disasters. Fire Res. Abstr. Rev. 13:1-16. Kerr, J.W., C.C. Buck, W.E. Cline, S. Martin, and W.D. Nelson (1971) Nuclear Weapons Effects in a Forest Environment--Thermal and Fire. Report N2:TR2-70. Washington, D.C.: Defense Nuclear Agency. Kittelson, D.B., and D.F. Dolan (1980) Diesel exhaust aerosols. Pages 337-359 ~n Generation of Aerosols and Facilities for Exposure Experiments, edited by K. Willeke. Ann Arbor, Mich.: Ann Arbor Science. Krinov, E.L. (1966) Giant Meteorites. New York: Pergamon Press. Pages 125-265 and 383-387. Larson, D.A., and R.D. Small (1982a) Analysis of the Large urban Fire Environment. II. Parametric Analysis and Model City Simulations. PSR Report 1210. Santa Monica, Calif.: Pacific Sierra Research Corp. Larson, D.A., and R.D. Small (1982b) Analysis of the Large Urban Fire Environment. I. Theory. PSR Report 1210. Santa Monica, Calif.: Pacific Sierra Research Corp.

91 Leaitch, W.R., J.W. Strapp, H.A. Wiebe, and G.A. Isaac (1983) Measurements of scavenging and transformation of aerosol inside cumulus. Pages 53-69 In Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Leong, K.H., K.V. Beard, and H.T. Ochs III (1982) Laboratory measurements of particle capture by evaporating cloud drops. J. Atmos. Sci. 39:1130-1140. Lewis, K.N. (1979) The prompt and delayed effects of nuclear war. Sci. Am. 241:35-47. Liou, K.-N. (1980) An Introduction to Atmospheric Radiation. New York: Academic Press. Lyman, H. (1918) Smoke from Minnesota forest fires. Mon. Weather Rev. 46:506-509. Manins, P.C. (1984) Cloud heights and stratospher ic in jections resulting from a thermonuclear war. Submitted to Atmos. Environ. Martin, S.G. (1974) The Role of Fire in Nuclear Warfare. Report URS 764 (DNA Report 2692F). San Mateo, Calif.: URS Research Corp. 166 pp. Matson, M., S.R. Schneider, B. Aldridge, and B. Satchwell (1984) Fire detection using the NOAA-series satellites. NOAA Tech. Rep. NESDIS 7. Washington, D.C.: U.S. Department of Commerce. McMahon, C.K. (1983) Characteristics of forest fuels, fires, and emissions. Paper presented at 76th Annual Meeting, Air Pollution Control Association, Atlanta, June 19-24. McMahon, C.K., and P.W. Ryan (1976) Some chemical and physical characteristics of emissions from forest fires. Paper presented at the 69th Meeting, Air Pollution Control Association, Portland, June 27 to July 1. Miller, C.F. (1962) Preliminary Evaluation of Fire Hazards from Nuclear Detonations. Memorandum Report, Project IMU-4021-302. Menlo Park, Calif.: Stanford Research Institute. 63 pp. Miller, C.F., and J.W. Kerr (1965) Field Notes on World War II German Fire Experience. Report MU-5070. Menlo Park, Calif.: Stanford Research Institute. 69 pp. Miller, R.K., M.E. Jenkins, and J.A. Keller (1970) Analysis of Four Models of the Nuclear-Caused Ignitions and Early Fires in Urban Areas. Report DC-FR-1210. Albuquerque, N.Mex.: Dikewood Corp. Mordy, W.A. (1960) Discussion. Geophys. Monogr. 5. Am. Geophys. Union 5:399. Morikawa, T. (1978) Evolution of soot and polycyclic aromatic hydrocarbons in combustion. Combust. Toxicol. 5:349-360. Murgai, M.P. (1976) Natural Convection from Combustion Sources. New Delhi: Oxford and IBH Publishing. Nakanishi, K., T. Kadoa, and H. Hiroyasu (1981) Effect of air velocity and temperature on the soot formation by combustion of a fuel droplet. Combust. Flame 40:247-262. Office of Technology Assessment (1979) The Effects of Nuclear War. Washington, D.C.: Office of Technology Assessment.

92 Ogr en, J.A. (1982) Deposition of particulate elemental carbon from the atmosphere. Pages 379-391 In Particulate Carbon: Atmospheric Life Cycle, edited by G.T. Wolff and R.L. Klimisch. New York: Plenum Press. Ohlemiller, T.J., J. Bellan, and F. Rogers (1979) A model of smoldering combustion applied to flexible polyurethane foams. Combustion and Flame 36:197-215. Oil and Gas Journal (1984) Statistics. Oil Gas J. 82:114. O'Sullivan, E.F., and B.K. Ghosh (1973) The spectral transmission, 0.5- 2.2 Em, of fire smokes. Pages 195-200 In Combustion Institute European Symposium 1973, edited by F.J. Weinberg. New York: Academic Press. Packham, D.R., and R.G. Vines (1978) Properties of bush fir e smoke: the reduction in visibility resulting from prescribed fires in forests. J. Air Pollut. Control Assoc. 28:790-795. Pagni, P.J., and S. Bard (1979) Particulate volume fractions in diffusion flames. Seventeenth Symposium on Combustion. Pittsburgh, Pa.: The Combustion Institute. Pages 1017-1028. Palmer, T.Y. (1976) Absorption by smoke particles of thermal radiation in large fires. J. Fire Flammability 7:460-469. Palmer, T.Y. (1981) Large fire winds, gases and smoke. Atmos. Environ. 15:2079-2090. Patterson, E.M., and C.K. McMahon (1983) Absorption Characteristics of Forest Fire Particulate Matter. Report FS-SE-2110-8. Atlanta, Gal: U.S. Fire Service. Patterson, E.M., B.T. Marshall, and K.A. Rahn (1982) Radiative properties of the Arctic aerosol. Atmos. Environ. 16:2967-2977. Pluchino, A.B., S.S. Goldberg, J.M. Dowling, and C.M. Randall (1980) Refractive-index measurements of single micron-sized carbon particles. Appl. Opt. 19:3370-3372. Plummer, F.G. (1912) Forest Fires: Their Causes, Extent and Effects, with a Summary of Recorded Destruction and Loss. U.S. For. Serv. Bull. 117:39 pp. Powell, E.A., C.P. Bankston, R.A. Cassanova, and B.T. Zinn (1979) The effect of environmental temperature upon the physical characteristics of the smoke produced by burning wood and P~C samples. Fire Materials 3:15-22. Prodi, F. (1983) The scavenging of submicron particles in mixed clouds: Physical mechanisms-laboratory measurements. Pages 505-516 ~n Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Pruppacher, H.R., and J.D. Klett (1978) Microphysics of Clouds and Precipitation. Boston: D. Reidel. Quintiere, J.G. (1982) An assessment of correlations between laboratory and full-scale experiments for the FAA aircraft fire safety program. Part 1. Smoke. Report NBSIR 82-2508. Washington, D.C.: National Bureau of Standards. 33 pp.

93 Radke, L.F. (1983) Preliminary measurements of the size distribution of cloud interstitial aerosol. Pages 71-78 in Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Radke, L.F., J.L. Stith, D.A. Hegg, and P.V. Hobbs (1978) Airborne studies of particles and gases from forest fires. J. Air Pollut. Control Assoc. 28:30-34. Radke, L.F., B. Benech, J. Dessens, M.W. Eltgroth, X. Henrion, P.V. Hobbs, and M. Ribon (1980a) Modifications of cloud microphysics by a 1000 MW source of heat and aerosol (the Meteotron project). Pages 119-126 In Proceedings of the Third WHO Scientific Conference on Weather Modification. Geneva: World Meteorological Organization. Radke, L.F., P.V. Hobbs, and M.W. Eltgroth {1980b) Scavenging of aerosol particles by precipitation. J. Atmos. Sci. 19:715-722. Radke, L.F., J.H. Lyons, D.A. Hegg, P.V. Hobbs, D.V. Sandberg, and D.E. Ward (1983) Airborne Monitoring and Smoke Characterization of Prescribed Fires on Forest Lands in Western Washington and Oregon. EPA Report 600/X-83-047. Washington, D.C.: U.S. Environmental Protection Agency. 104 pp. Radke, L.F., J.H. Lyons, D.A. Hegg, and P.V. Hobbs (1984) Airborne monitoring and smoke characterization of prescribed fires on forest lands in western Washington and Oregon. Seattle: University of Washington. Randhawa, J.S., and J.E. Van der Laan {1980) Lidar observations during dusty infrared Test-l. Appl. Opt. 19:2291-2297. Rasbash, D.J., and D.D. Dryedale (1982) Fundamentals of smoke production. Fire Safety J. 5:77-86. Rasbash, D.J., and B.T. Pratt (1979/80) Estimation of the smoke produced in fires. Fire Safety J. 2:23-37. Roessler, D.M., and F.R. Faxvog (1979) Optoacoustic measurement of optical absorption in acetylene smoke. J. Opt. Soc. Am. 69:1699-1704. Roessler, D.M., and F.R. Faxvog (1980) Optical properties of agglomerated acetylene smoke particles at 0.5145 um and 10.6 um wavelengths. J. Opt. Soc. Am. 70:230-235. Roessler, D.M., D.-S.Y. Wang, and M. Kerker (1983) Optical absorption by randomly oriented carbon spheroids. Appl. Opt. 22:3648-3651. Rosen, H., and T. Novakov {1983) Combustion-generated carbon particles in the Arctic atmosphere. Nature 306:768-770. Rosenkilde, C.E., and F.J.D. Serduke (1983) Turbulent aspects of rainout. Pages 589-596 in Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Saito, F. (1974) Smoke generation from building materials. Fifteenth Symposium on Combustion. Pittsburgh, Pa.: The Combustion Institute. Pages 269-279. Sandberg, D.V., J.M. Pierovich, D.G. Fox, and E.W. Ross {1979) Effects of Fire on Air. Technical Report WO-9. Washington, D.C.: Forest Service, U.S. Department of Agriculture. 40 pp.

94 Schroeder, M.J., and C.C. Chandler (1966) Monthly Fire Behavior Patterns. U.S. For. Serv. Res. Pap. PSW-112:15 pp. Seader, J.D., and I.N. Einhorn (1976) Some physical, chemical, toxicological and physiological aspects of fire smokes. Sixteenth International Symposium. Pittsburgh, Pa.: The Combustion Institute. Pages 1423-1445. Seader, J.D., and S.S. Ou (1977) Correlation of the smoking tendency of materials. Fire Res. 1:3-9. Sehmel, G.A. (1980) Particle and gas dry deposition: A review. Atmos. Environ. 14:983-1011. Seller, W., and P.J. Crutzen (1980) Estimates of gross and net fluxes of carbon between the biosphere and the atmosphere from biomass burning. Climatic Change 2:207-247. Short, N.M., P.D. Lowman, Jr., S.C. Freden, and W.A. Finch, Jr. (1976) Mission to earth: Landsat views the world. NASA SP-360. Washington, D.C.: National Aeronautics and Space Administration. Shostakovitch, V.B. (1925) Forest conflagrations in Siberia. J. For. 23:365-371. Slinn, W.G.N. (1977) Some approximations for the wet and dry removal of particles and gases from the atmosphere. Water Air Soil Pollut. 7:513-543. Slinn, W.G.N. (1983) A potpourri of deposition and resuspension questions. Pages 1361-1416 ~n Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Small, R.D., and B.W. Bush (1984) Smoke production from multiple nuclear explosions in wildlands. Submitted to Science. Smith, C.D., Jr. (1950) The widespread smoke layer from Canadian forest fires during late September 1950. Mon. Weather Rev. 78:180-184. Tangren, C.D. (1982) Scattering coefficient and particulate matter concentration in forest fire smoke. J. Air Pollut. Control Assoc. 32(7):729-732. Taylor, R.J., S.T. Evans, N.K. King, E.T. Stephens, D.R. Packham, and R.G. Vines (1973) Convective activity above a large-scale brushfire. J. Appl. Meteorol. 12:1144-1150. Tewar son, A. (1982) Experimental Evaluation of Flammability Parameters of Polymeric Materials. Pages 97-153 ~n Flame Retardant Polymeric Material, Vol. 3, edited by M. Lewin, S.M. Atlas, and E.M. Pierce. New York: Plenum Press. Tewarson, A., and J. Steciak (1982) Fire Ventilation. Technical Report. Washington, D.C.: U.S. Department of Commerce. Tewar son, A., J.L. Lee, and R.F. Pion (1980) The influence of oxygen concentration on fuel parameters. Paper presented at the 18th International Symposium on Combustion, University of Waterloo, Canada, August 17-11. Tien, C.L., D.G. Doornink, and D.A. Rafferty (1972) Attenuation of visible radiation by carbon smokes. Combust. Sci. Technol. 6:55-59. Tomaselli, V.P., R. River a, D.C. Edewaard, and K.D. Moller (1981) Infrared optical constants of black powders determined from reflection measurements. Appl. Opt. 20:3961-3967.

95 Tsang, T.H., and J.R. Brock (1982) Effect of coagulation on extinction in an aerosol plume propagating in the atmosphere. Appl. Opt. 21:1588-1592. Turco, R.P., O.B. Toon, T.P. Ackerman, J.B. Pollack, and C. Sagan (1983a) Nuclear winter: Global consequences of multiple nuclear explosions. Science 222:1283-1292. Turco, R.P., O.B. Toon, T.P. Ackerman, J.B. Pollack, and C. Sagan (1983b) Global Atmospheric Consequences of Nuclear War. Interim Report. Marina del Rey, Calif.: R&D Associates. 144 pp. Turco, R.P., O.B. Toon, R.C. Whitten, J.B. Pollack, and P. Hamill (1983c) The global cycle of particulate elemental carbon: A theoretical assessment. Pages 1337-1351 In Precipitation Scavenging, Dry Deposition and Resuspension, edited by H.R. Pruppacher, R.G. Semonin, and W.G.N. Slinn. New York: Elsevier. Twitty, J.T., and J.A. Weinman (1971) Radiative properties of carbonaceous aerosols. J. Appl. ~teteorol. 10:725-731. Twomey, S. {1977) Atmospheric Aerosols. New York: Elsevier. U.N. (1981) Demographic Yearbook, 1979. New York. U.S. Department of Defense (1973) DCPA Attack Environment Manual, Chapter 3. Washington, D.C. USDA (1972) A mathematical model for predicting fire spread in wildland fuels. U.S. For. Serv. Res. Note INT-115. USDA (1981) Tree biomass--A state-of-the-art compilation. U.S. For. Serv. Tech. Rep. W0-33. Uthe, E.E. (1981) Lidar evaluation of smoke and dust clouds. Appl. Opt. 20:1503-1510. Valioulis, I.A., and E.J. List (1984) Collision efficiencies of diffusing spherical particles: Hydrodynamic, Van der Waals and electrostatic forces. Adv. Colloid. Interface Sci. 20:1-20. Vervisch, P., D. Peuchberty, and T. Mohamed (1981) The spectral transmission of 0.4-4.5 um of fire smokes. Combust. Flame, 41:179-186. Volz, F.E. (1972) Infrared absorption by atmospheric aerosol substance. J. Geophys. Res. 77:1017-1031. Wagner, H.Gg. (1981) Soot formation--An overview. Pages 1-29 ~n Particulate Carbon Formation During Combustion, edited by D.C. Siegla and G.W. Smith. New York: Plenum Press. Wang, P.K., S.N. Grover, and H.R. Pruppacher (1978) On the effect of electric charges on the scavenging of aerosol particles by clouds and small raindrops. J. Atmos. Sci. 35:1735-1743. Ward, D.E., R.M. Nelson, Jr., and D.F. Adams (1979) Forest fire smoke plume documentation. Paper 79-6-3 presented at the 72nd Annual Meeting, Air Pollution Control Association, Cincinnati, Ohio. Watson, H.H. (1951) Alberta forest-fire smoke. Weather 6:253. Watson, H.H. (1952) Alberta forest-fire smoke. Weather 7:128. Wein, R.W., and D.A. MacLean (1983) The Role of Fire in Northern Circumpolar Ecosystems. New York: John Wiley and Sons. 322 pp. Wexler, H. (1950) The great smoke pall--September 24-30, 1950. Weatherwise 3:129-134.

96 Wiersma, S.J., and S.B. Martin (1973) Evaluation of the Nuclear Fire Threat to Urban Areas. Report AD779-340. Menlo Park, Calif.: Stanford Research Institute. 131 pp. Williams, D.W., J.S. Adams, J.J. Batten, G.F. Whitty, and G.T. Richardson (1970) Operation Euroka: An Australian Mass Fire Experiment. Report 386. Maribyrnor, Victoria, Australia: Defense Standards Laboratory. Wolff, G.T., and R.L. Klimisch feds.) (1982) Particulate Carbon: Atmospheric Life Cycle. New York: Plenum Press. 411 pp. Woodie, W.L., D. Remetch, and R.D. Small (1983) Fire spread from tactical nuclear weapons in battlefield environments. PSR Note 566 Santa Monica, Calif.: Pacific Sierra Research Corp. 53 pp. Wright, H.A., and A.W. Bailey (1982) Fire Ecology, United States and Southern Canada. New York: John Wiley and Sons. .

97 APPENDIX 5-1: OBSERVATION OF PLUME HEIGHTS AND ASH TRANSPORT IN LARGE FIRES, by F.E. Fendell Plume Heights The altitude achieved by a plume over a maintained source of buoyancy depends largely on the strength of the source (heat released per unit time), the stratification and humidity of the ambient air, the strength of the crosswind (if any), and the size of the region of exothermicity. Rarely are all the desired inputs known for a single event. As a reference, one of the more dramatic persistent plumes of the last quarter century was that associated with the creation of Surtsey off Iceland. An effective heat source estimated at 1011 J/s (with upflow at the base of roughly 120 m/s) was initiated at 7 A.M. on November 14, 1963 (the energy release rate was equivalent to about 250 kt every 3 h). By 10:30 A.M. the plume was at 3.5 km; by 3 P.M., at about 6.3 km; and by the next day, at over 9.3 km (i.e., to the height of the tropopause near Iceland). Vapor columns rose from neighboring sites on the sea to 2.5 km, and ash-laden steam burst upward to 0.6 km in a gigantic, ink-black column (Bourne, 1964; Thorarinsson and Vonnegut, 1964; Thorarinsson, 19661. As another reference, the series of artificial convection experiments conducted at the Centre de Recherches Atmospheriques Henri Dessens, on the Lannemezan plateau in the French Pyrenees, entailed 105 fuel oil burners deployed in a three-arm spiral within a 140 m x 140 m square (the Meteotron). The heat release rate was about 109 J/s for 20 to 30 min (a total energy release of about 0.5 kt), and the plume reached 1 to 2 km {Benech, 1976; Church et al., 19801. Plumes of most small-scale fires reach only a few kilometers into the troposphere. The black plume of a 101° J/s oil fire that persisted for days near Long Beach, California, rose to 4 km (Henna and Gifford, 1975~. The convection column associated with the bombing of Leipzig in World War II, an event severe enough to give 15 m/s ground-level radial inflow at 4 km from the center and 34 m/s closer in, rose to only 3.9 km (Broido, 1960~. The first thousand-bomber raid by the British in World War II (on Cologne, on May 30-31, 1942) produced a column of smoke that rose to 4.5 km (and hung as a huge pall at daybreak) (Barker, 19651. Taylor et al. (1973) reported a brushfire near Darwin River, Australia, on September 10, 1971, in which the ambient temperature fell almost linearly from 301 K at ground level to 268 K at 6 km. Whereas the plume rose to 3 to 4 km for a heat release rate of 1011 J/s, during a 10- to 15-min interval the plume advanced to 5.8 km when the heat release rate doubled. A small cloud above the plume was sucked down into it 10 min after this rapid additional ascent. However, the fuel loading for this case was about one-tenth that in portions of the American Pacific Northwest, which has the highest loadings in the continental United States. Thus one is motivated to examine severe burning events more closely. Of the acreage burned in the United States annually, 95 percent comes from 2 to 3 percent of the total number of fires; these exceptional fires tend to occur in dry, hot, windy weather, can jump

98 rivers and lakes, and decay only with wind shifts, the arrival of precipitation, and/or the exhaustion of fuel. Thirteen fire complexes in the recorded history of North America have each taken 4000 km or more. Twelve thousand square kilometers were burned by the Maramichi and Maine fires of 1825, the North Carolina fire of 1898, and the Idaho and Montana fires of 1910; the Alaskan fires of 1957 consumed 20,000 km2. Fire complexes in Michigan in 1871, in Wisconsin in 1894, and in Washington and Oregon in 1910 each burned 8000 km2. Southern states lead the national fire statistics annually in both frequency and area burned; however, natural decomposition is slower in the North and fuel loads accumulate, so while the number of fires is fewer, with droughts come holocausts. As for extremes in spread rate, an 1887 Texas grass fire spread 26 km in 2 h, and crown fires propagating at 16 km/in have been recorded (Pyne, 1982~. At one time the August 1933 fire in Tillamook County, Oregon, was regarded as the most intense in recorded American experience. On August 24, 1933, hurricane-like winds arose, and 800 km2 were burned in 20 h. The plume, which had reached 3 km (Holbrook, 19431, pierced an inversion, and the smoke column reached 11.1 to 12 km, near-tropopause-level altitude (Pyne, 19821. In recent years, several events perhaps comparable in intensity to the Tillamook fire have been recorded. The Sundance fire in the northern Idaho area of Pack River and McCormick Creek advanced 14.5 km and burned 200 km2 from 2 to 11 P.M. on September 1, 1967. The energy release rate is estimated at 5 x 105i J/s, and the convection column reached 10.7 km, even though a 32- to 80-km/in wind was blowing (Anderson, 1968~. The peak rate was achieved during saturation spotting in a valley somewhat sheltered from the wind. A fire at an Air Force bombing range in North Carolina in 1971 was characterized by a crosswind of 32 km/in, a heat release rate of 1.2 x 1011 J/s, and a plume height of 4.6 km. A fire in the Sierra National Forest on July 16, 1961, burned 20 km2 in 5 h, and a convective column rose to 6 to 9 km. The so-called Mack Lake fire in the Huron National Forest, Michigan, on May 5, 1980, burned 100 km2 in 6 h; though the highly bent plume rose to only 4.6 km in the intense crosswind, the heat release rate has been estimated at 1.6 x 1011 J/s. However, the highest free-burning-fire heat release rates are associated with firestorms, the exceptional heat-cyclone consequences of massive incendiary air raids on urban targets during World War II. The rareness of these events is evidenced by the fact that the U.S. Strategic Bombing Survey characterizes only four firestorms (Hamburg, Kassel, Darmstadt, and Dresden) arising from the 49 major German cities subjected to incendiary bombing (SPRI, 1975~. No firestorm arose as a consequence of fifteen massive incendiary raids from March to June 1945 on Osaka, Kobe, Nagoya, Tokyo, or Yokohama, although the atomic bombing of Hiroshima produced a firestorm. At Hamburg, during the raid on July 27-28, 1943, a cumulonimbus- cloud-like plume with an anvil top, of 3-km thickness, rose to 10 km (Ebert, 1963; Morton, 1970) in a near-adiabatic lapse rate in the lowest few kilometers of the troposphere; this altitude was ascribed by a meteorologist 6 km away, although Brunswig (1982, page 245) ascribes

99 a height of only 7 km. Thick black smoke reached 6.9 km in half an hour after the onset of bombing; later-arriving crews reported severe turbulence, and some aircraft returned to base soot-covered (Middlebrook, 1981; Musgrove, 1981~. Large black greasy raindrops fell along the outskirts of the fire (Caidin, 1960~. Smoke and dust blotted out the sky for 30 h after the attack; the sun was not seen by Hamburg residents the next day (Rumpf, 19631. Dresden was subjected to two massive raids on February 13-14, 1945, though stratocumulus clouds caused a total overcast above 3 km for most of the night, and strong winds persisted. In these raids, 12.4 km2 were 75 percent destroyed, and an additional 4 km2 were 25 percent destroyed, by fires that persisted 7 days and 8 nights. A firestorm occurred in a quarter circle of 2.2-km radius around the time of the raid. At daybreak on February 14, the city was obscured by a column of yellow-brown smoke filled with lifted flotsam; this column appeared particularly dark up to 4.8 km. Sooty ash showered downwind as far as 29 km for several days (Irving, 19651. Smoke Obscuration There are accounts of smoke so thick from Pacific Northwest forest fires that navigation on the Columbia River and other inland waterways was brought to a standstill in 1849 and 1868. An instance of sun obscuration is given by the Peshtigo fires (October 8-9, 1871), in which 5000 km2 were burned along both banks of the Green Bay. The sun was obscured for 320 km, and gloom persisted, even at noontime, for a week {Holbrook, 19431. Paper lofted from Michigan crossed Lake Huron and landed in Canada. On August 20, 1910, some 1750 separate fires in Idaho and Montana blew up and 12,000 km2 were burned, such that the sun was blotted out (Holbrook, 1943~. However, the time scale for reduced daytime visibility was days, not weeks. References Anderson, H.E. (1968) Sundance Fire: An Analysis of Fire Phenomena. Research Paper INT-56. Ogden, Utah: Intermountain Forest and Range Experiment Station, Forest Service, U.S. Dept. of Agriculture. Barker, R. (1965) The Thousand Plan. London: Chatto and Windus. Benech, B. (1976) Experimental study of an artificial convection plume initiated from the ground. J. Appl. Meteorol. 15:127-137. Bourne, A.G. (1964) Birth of an island. Discovery 25 (Aprill:16-19. Broido, A. (1960) Mass fires following nuclear attack. Bull. Atmos. Sci. 16~10~:409-413. Brunswig, H. (1982) Feuerstrum uber Hamburg. Stuttgart: Motorbuch Verlag. Caidin, M. (1960) The Night Hamburg Died. New York: Ballantine.

100 Church, C.R., J.T. Snow, and J. Dessens (1980) Intense atmospheric vortices associated with a 1000 MW fire. Bull. Am. Meteorol. Soc. 61(7):682-694. Ebert, C.H.V. (1963) The meteorological factor in the Hamburg fire storm. Weatherwise 16~2~:70-75. Hanna, S.R., and F.A. Gifford (1975) Meteorological effects of energy dissipation at large power parks. Bull. Am. Meteorol. Soc. 56~1~:1069-1076. Holbrook, S.H. (1943) Burning an Empire. New York: Macmillan. Irving, D. (1965) The Destruction of Dresden. New York: Ballantine. Middlebrook, M. {1971) The Battle of Hamburg. New York: Charles Scribner's Sons. (1970) The physics of fire whirls. Fire Res. Abstr. Rev. Morton, B.R. 12:1-19. Musgrove, G. (1981) Operation Gomorrah--The Hamburg Firestorm Raids. New York: Jane's. Pyne, S.J. (1982) Fire in America--A Cultural History of Wildland and Rural Fire. Princeton, N.J.: Princeton University Press. Rumpf, H. (1963) The Bombing of Germany. New York: Holt, Rinehart, and Winston. Stockholm Peace Research Institute (1975) Cambridge, Mass.: MIT Press. Incendiary Weapons. Taylor, R.J., S.T. Evans, N.K. King, E.T. Stephens, D.K. Packham, and R.G. Vines (1973) Convective activity over a large-scale bushfire. J. Appl. Meteorol. 12:1144-1150. Thorarinsson, S. (1966) Surtsey, the New Island in the North Atlantic. Reykjavik, Iceland: Almenna Bokafelagio. Thor arinsson, S., and B. Vonnegut (1964) Whirlwinds produced by the eruption of Surtsey volcano. Bull. Am. Meteorol. Soc. 45~8~:440-444.

101 APPENDIX 5-2: WATER IN NUCLEAR CLOUDS Clouds produced by nuclear explosions and by the fires they ignite can hold large quantities of water. The injection of this water into the upper air layers, and the consequences of the injection, are discussed in this appendix. Explosion Clouds Nuclear explosion clouds hold water that is vaporized and engulfed by the fireball. Surface bursts over deep water are expected to be relatively rare in a nuclear exchange and will be neglected (based on Pacific test data <3 x 106 tons of condensed water per megaton of yield are expected in the stabilized clouds (Gutmacher et al., 1983~. Subsurface ocean bursts do not generate high-altitude clouds (Glasstone and Dolan, 1977~. Surface bursts over land can raise about 3 x 105 tons/Ml of soil to the stabilized cloud height. About an equal amount of groundwater and mineralized water of hydration might be assumed. The fireball also entrains ambient water vapor as it rises through the lower troposphere. Adopting a fireball expansion rate such that dR/dz ~ 0.2 (that is, the increase in the fireball radius is about one-fifth of the height traversed), and a U.S. Standard (1976) mid-latitude water vapor profile, the entrained water vapor could vary from <1 x 105 to about 1 x 106 tons/Ml for a surface burst, depending on surface humidity. Accordingly, an average stabilized- cloud-water content of 1 x 106 tons H2O Mt is generous. The water concentration in stabilized nuclear clouds would be <1 g/m3, which is generally too small to cause precipitation but large enough to form an optically thick (ice) condensation cloud. As the nuclear cloud disperses, the ice particles would either settle out or evaporate. Air bursts above about 2 to 3 km would hold <1 x 105 tons of H2O per megaton. The total water injected by explosion clouds in the baseline exchange would almost certainly be less than 6000 Tg. Most of the water would be deposited in the troposphere. Fire Plumes There are three sources of moisture for fire plumes: water of combustion, evaporated surface water, and entrained water vapor. Most combustible materials generate <1 g-H2O/g-burned. Thus, in the present baseline exchange, up to 8500 Tg of H2O would be produced directly by fires, and could disperse with the plumes. Even if 1 cm of water were evaporated over the entire fire area in the baseline scenario, only 5000 Tg of additional water would enter the plumes; the actual amount would be much less, of course. Entrainment of ambient humidity into the plume, particularly at ground level where air is often efficiently sucked into the fire, could add >1 g-H2O/g-burned. At Hiroshima, a crude estimate suggests that about 10 g-H2O/g-burned

102 were entrained due to the high humidity at the time of the fire (R.P. Turco, private communication, 1984~. However, most of this water fell as precipitation (the "black rainy. Due to condensation and precipitation, only a limited quantity of water can remain suspended in the fire plumes and be carried long distances. This quantity is assumed to be S g H2O/g-burned, which consists primarily of moisture drawn into the fire near the ground. The total fire plume water injection in the baseline exchange may then be estimated as 40,000 Tg. The water is injected uniformly between 0 and 9 km (as is the smoke in the baseline case), or about 4000 Tg/km of altitude. Note that the injection represents primarily a redistribution of water vapor from the boundary layer into the free troposphere--as occurs during natural convection--because very little "new. water vapor is introduced by the combustion process. The water concentration (condensate plus vapor) in the stabilized high-altitude plumes of large fires is expected to be about 1 g/m3, based on the analysis of the water budget of a fire plume discussed above, air inflow rates obtained from plume theory, and direct measurements in fire cumulus cap clouds (L. Radke, private communication, 1984~. The onset of condensation in the convective column of a fire may occur above the level expected for condensation in surface air lifted adiabatically, due to the added heat of combustion and the entrainment of dry ambient air aloft (Taylor et al., 1973~. Low surface humidity, induced precipitation and entrainment of dry air can all limit the water concentration in fire plumes. The column abundances of water in fire clouds could be 1000 to 5000 g/m2, compared to about 10,000 g/m2 in natural cumulus clouds and 10 to 100 g/m2 in cirrus clouds. An upper limit to the water injection by fires in a nuclear conflict is in the vicinity of 500,000 Tg. This figure assumes that the initial fire plumes occupy a volume of 1017 m3 (about one-tenth of the volume of the northern hemisphere mid-latitude troposphere), all of the air in the plumes originates in the surface layer and holds an average of 5 g H2O/m3, and no rainout occurs. Obviously, these circumstances are highly unlikely. Water Perturbation Table 5.2-1 gives the average ambient profile of water vapor at mid-latitudes. The global troposphere holds roughly 107 Tg of water vapor and the stratosphere, about 3000 Tg. If all of the water in nuclear explosion clouds were confined to the mid-latitude stratosphere, H2O concentrations could increase by a factor of <10 there. Because the stratosphere normally is very dry, with a relative humidity of only 1 to 5 percent, and injected smoke and dust clouds can be heated by solar and infrared radiation, any condensed water would soon evaporate as the individual explosion clouds dispersed. A factor of 10 increase in stratospheric H2O would affect ozone photochemistry and the infrared radiation balance of the stratosphere. The

103 TABLE 5 . 2-1 Amb lent Atmospher ic Water Vapor a Equivalent Cumulative H2O Altitude Water Vapor Air Global H2O Mass up to the Interval Mixing Ratiob Density2 in the Layer Top of Layer (km) (ppmm) (kg/m3) (Tg) (Tg) 0-0.5 0.5-1.5 1.5-3.0 3.0-5.0 5.0-7.0 7.0-9.0 9.0-11. 11.-13. 13.-15 15.-17. 4686 3700 2843 1268 554 216 43.2 11.3 3.3 3.3 1.225 1.112 1.007 0.8194 0.6601 0.5258 0.4135 0.3119 0.2279 0.1665 1.4x106 2.1x106 2.1x106 l.Ox106 3.7x105 1.1x105 1.8x104 3500 750 550 1.4x106 3.5x106 5.6x106 6.6x106 7.0x106 7.1x106 7.1x106 7.1x106 7.1x106 7.1x106 he aCondensed water, which may reach concentrations of 10 g/m3 (5 x 106 Tg globally in a 1-km-thick layer), is neglected. U.S. Standard Atmosphere (1976) Midlatitude Mean Model. The water vapor mixing ratio is given in parts per million by mass (ppmm). Local water vapor fluctuations typically exceed 10 percent. photochemical effect of the H2O, however, would probably not be any more important than the photochemical effect of the explosion-generated NOX. The radiation perturbations are discussed below. The fire plume water injection of about 4000 Tg/km up to 9 km is typically <1 percent of the ambient water vapor at any level in this height interval. The total fire H2O injection is <0.5 percent of the global water vapor burden, and represents about 45 min of the normal global atmospheric water budget. The maximum perturbation could occur in the 7- to 9-km layer, where the average mid-latitude water vapor burden could increase by about 20 percent. If all of the fire water were put into this layer at northern mid-latitudes, the water burden would increase by about 0.20 g/m3, or about 400 g/m2. However, most of this water originates in lower regions of the atmosphere; the redistribution of water is likely to be less significant than an increase in the total water burden of the atmosphere. The improbable "upper limits water injections discussed in the previous section would lead to more substantial effects. Nevertheless, in view of the large ambient quantities of water vapor in the atmosphere, and the indirect water vapor perturbations to be discussed below, even the maximum credible water injections by fires could turn out to be of secondary interest.

104 CO2 Perturbation Carbon dioxide injections by nuclear fires are much less important than water injections. Because CO2 is uniformly mixed throughout the troposphere and stratosphere (at about 340 parts per million by volume), the transfer of air between different altitude levels by nuclear explosions and fires has little effect on the CO2 distribution. The global atmosphere holds about 3 x 106 Tg of CO2. Nuclear fires could generate about 1 x 104 Tg of CO2, roughly the amount produced in 1 year from fossil fuel combustion. Carbon dioxide is transparent in the visible spectrum, does not condense, and has only a limited infrared opacity (Liou, 1980~. On the other hand, CO2 perturbations could result from indirect disturbances in the global biospheric carbon cycle in the aftermath of a nuclear war (a subject that is not pursued in this report). Effects of Water Injections The water injected into the upper atmosphere with dust and smoke can have a number of important effects: 1. Modification of the photochemistry of ozone (see the previous discussion and Chapter 6~. 2. Scavenging and washout of dust and smoke particles (see the discussion in Chapter 5~. 3. Perturbation of the visible and infrared radiation balance by the condensed and vapor states of water. During the first week after the start of a nuclear war, the localized explosion clouds and fire plumes could hold significant quantities of condensed water. The visible and infrared opacities of these clouds could be very large (>>1~. Light levels below the clouds would be very low, particularly when heavy soot loadings are present. The infrared energy balance of the clouds would be complex, and some degree of thermal blanketing could result. Nevertheless, without solar insolation, the ground should still tend to cool. A strong greenhouse effect is not likely (at least in the case of smoke plumes) because solar absorption and heating would occur above most of the infrared opacity of the clouds (see Chapter 7~. In daylight, the smoke clouds would warm up rapidly, possibly inducing strong vertical and horizontal mixing of the cloud tops and edges, and perhaps causing some of the condensed water to evaporate. At night the clouds would cool by infrared emission, and subsidence might occur. The turbulence created by these heating and cooling cycles would be confined primarily to the upper cloud layers where precipitation is less probable. The major effect might therefore be to accelerate the dispersion of the smoke clouds. Some of the extended fire plumes would hold sufficient water to form thick cirrus anvils.

105 These cirrus could greatly increase the albedo above the smoke plumes, but would also hold in upwelling infrared radiation. The large-sca~e advection and spreading of smoke clouds by self-induced heating has been studied on different size scales. Chen and Orville (1977) investigated cumulus-scale convection of carbon-black clouds. R. Haberle et al. (private communication, 1983) and M. MacCracken (private communication, 1984) simulated the motions of hemispherical-scale soot clouds. In each case, the same general behavior was predicted. The clouds tended to rise and spread horizontally at a faster rate than would be expected if only ambient air motions were acting. Direct observations of large sooty smoke clouds reveal the same behavior (Davies, 1959~. Thus it is expected that some of the energy absorbed in the dust and smoke clouds would be converted into the kinetic energy of winds, which eventually dissipates as frictional heat. Within about 2 weeks, the nuclear dust and smoke clouds would be sufficiently dispersed that their infrared opacities would be quite small (<1~. The atmosphere could then approach the radiative regime analyzed by Turco et al. (1983a,b), Crutzen et al. (1984), and others, in which the infrared properties of the injected nuclear debris are less important than the visible properties. Water vapor, particularly in the stratosphere, can affect the infrared radiation balance of the atmosphere. It has been estimated, for example, that a five-fold increase in stratospheric H2O (with all other factors unchanged) would eventually lead to a 2°C surface warming (e.g., Manabe and Wetherald, 1967~. However, in the perturbed atmosphere, even this modest effect is unlikely to occur, because the surface temperatures and infrared radiation fluxes of the lower atmosphere would already be greatly reduced. Indirect Water Perturbations Changes in surface air temperatures, winds, and atmospheric stability would disturb the "normal" hydrological cycle. Such disturbances could be more important than the primary water injections of the explosions and fires. Among the hydrological perturbations that might develop: 1. Increased low-level storminess and precipitation near ocean-continent margins, induced by exaggerated sea-land temperature contrasts. 2. Formation of widespread ground fogs over continents due to rapid radiative cooling of surface air. 3. Suppression of deep convection and upper-level precipitation caused by soot-induced heating of the upper troposphere. 4. Decrease in general cloudiness above several kilometers altitude as a result of warming and reduced relative humidity. 5. Reduction in the global water vapor burden associated with a general decrease in surface air temperatures. 6. Increase in water vapor concentrations above several kilometers altitude due to the enhanced moisture capacity of the heated air.

106 It is not likely that all of these effects would occur. A partial discussion of the possibilities is given in Chapter 7. Further research into these problems will be necessary to determine their importance. References Chen, C.-S., and H.D. Orville (1977) The effects of carbon black dust on cumulus-scale convection. J. Appl. Meteorol. 16:401-412. Crutzen, P.J., C. Brahl, and I.E. Galbally (1984) Atmospheric effects from post-nuclear fires. Climatic Change, in press. Davies, R.W. (1959) Large-scale diffusion from an oil fire. Pages 413-415 In Atmospheric Diffusion and Air Pollution, edited by F.N. Frenkiel and P.A. Sheppard. New York: Academic Press. Glasstone, S., and P.J. Dolan (eds.) (1977) The Effects of Nuclear Weapons. Washington, D.C.: U.S. Department of Defense. 653 pp. Gutmacher, R.G., G.H. Higgins, and H.A. Tewes (1983) Total mass and concentration of particles in dust clouds. Rep. UCRL-14397. Livermore, Calif.: Lawrence Livermore Laboratory. 22 pp. Liou, K.-N. (1980) An Introduction to Atmospheric Radiation. New York: Academic Press. Manabe, S., and R.T. Wetherald (1967) Thermal equilibrium of the atmosphere with a given distribution of relative humidity. J. Atmos. Sci. 24:241-259. Taylor, R.J., S.T. Evans, N.K. King, E.T. Stephens, D.R. Packham, and R.G. Vines (1973) Convective activity above a large-scale brushfire. J. Appl. Meteorol. 12:1144-1150. Turco, R.P., O.B. Toon, T.P. Ackerman, J.B. Pollack, and C. Sagan (1983a) Nuclear winter: Global consequences of multiple nuclear explosions. Science 222:1283-1292. Turco, R.P., O.B. Toon, T.P. Ackerman, J.B. Pollack, and C. Sagan (1983b) Global Atmospheric Consequences of Nuclear War. Interim Report. Marina del Rey, Calif.: R&D Associates. 144 pp. U.S. Standard Atmosphere (1976) Washington, D.C.: U.S. Government Printing Office.

Next: 6 Chemistry »
The Effects on the Atmosphere of a Major Nuclear Exchange Get This Book
×
Buy Paperback | $55.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Most of the earth's population would survive the immediate horrors of a nuclear holocaust, but what long-term climatological changes would affect their ability to secure food and shelter? This sobering book considers the effects of fine dust from ground-level detonations, of smoke from widespread fires, and of chemicals released into the atmosphere. The authors use mathematical models of atmospheric processes and data from natural situations—e.g., volcanic eruptions and arctic haze—to draw their conclusions. This is the most detailed and comprehensive probe of the scientific evidence published to date.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!