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.
The Medical Implications of Nuclear War, Institute of Medicine. ~ 1986 by the National Academy of Sciences. National Academy Press, Washington, D.C. Atmospheric Perturbations of Large-Scale Nuclear War ROBERT C. MALONE, PH.D. Los Alamos National Laboratory, Los Alamos, New Mexico Several of the papers in this volume have discussed nuclear winter, large fires, and the dynamics of smoke plumes from large hires. I would like to elaborate on this theme by describing new computer simulations of the atmospheric consequences of the injection of a large quantity of smoke. I will focus on what might happen to the smoke after it enters the atmosphere and what changes or perturbations could be induced in the atmospheric structure and circulation by the presence of a large quantity of smoke. To help in understanding the significance of these atmospheric pertur- bations and the manner in which they anse, I will start by breaking the nuclear winter phenomenon into its component parts. A very simplified view of nuclear winter is represented in Figure 1A, in which is shown a vertical column of processes and a box to the side that represents smoke injected into the atmosphere. Ignoring the rest of Figure 1 for the moment, it can be seen that there are two basic ingredients to nuclear winter: sunlight coming into the earth's atmosphere and smoke that has been injected into the atmosphere by fires. The smoke absorbs some of the incoming sunlight, causing a reduction in sunlight reaching the earth's surface. A radiation The material on which this presentation is based is drawn from two papers (Malone et al., 1985, 1986). Readers interested in a more comprehensive discussion that includes historical background, related research, technical details of the model, and more extensive references should consult these articles, particularly the latter. All figures are reprinted from Malone, R. C., et al., 1986, Nuclear winter: three- dimensional simulations including interactive transport, scavenging, and solar heating of smoke, J. Geophys. Res. 91 (D1): 1039-1053, @) 1986 by the American Geophysical Union. Reprinted with permission. 141
142 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES deficit at the surface results because the surface continues to emit infrared radiation (heat). The smoke particles do not trap infrared radiation effec- tively, so the heat goes out into space (not indicated in Figure 11. This continuing heat loss to space combined with reduced incoming sunlight causes the surface to cool. This is the origin of the so-called nuclear winter effect. It is apparent Mat the magnitude of the cooling would depend on the amount of smoke injected and that the duration of the cooling would depend on how long smoke remained in the atmosphere. The latter point brings us to the next element of complication in this picture, which is the removal of smoke from the atmosphere by rainfall (Figure 1B). Precipitation scavenging of smoke, as this is also called, was considered in the TTAPS study of nuclear winter (Turco et al., 1983) and ~ r -A , I ~ SUNLIGHT I , 1 ~ , ~ SURFACE COOLING ABSORPTION BY SMOKE HEATING ~OF SMOKE l l . I ! WINDS ~ I SMOKE LOFTING LOWERING OF TROPOPAUSE | ATTENUATION | OF SUNLIGHT . ,. . 1 DISTRIBUTION AMOUNT OF SMOKE I, B A | .________________J FIGURE 1 Interconnection of the processes which control the distribution and res- idence time of smoke in the atmosphere and the resulting surface climatic change. Some arrows indicate that one process causes another; other arrows indicate only that a process influences the operation of the process to which the arrow points. For example, the presence of smoke in the atmosphere results in the absorption of sunlight, which causes heating, which causes both lofting of the smoke and lowering of the tropopause. These two effects influence (decrease) the efficiency with which precip- itation removes smoke by changing the vertical distribution of both smoke and pre- cipitation. Removal by rain changes the amount of smoke. Heating also modifies the winds, which influence the distribution of smoke. Source: Malone et al. (1986, p. 1040).
ATMOSPHERIC PERTURBATIONS OF GE-SC NUCLEI Wow 143 also in the report on this subject by the National Academy of Sciences (1985~. However, with the models that were available at the time, it was necessary to assume that the removal of smoke by rainfall occurred at a rate that was prescribed based on the observed lifetime of smoke particles in the unperturbed atmosphere. Although it was recognized that changes in the atmosphere would occur, it was not possible to take these changes into account in the models. It has now become possible to investigate these atmospheric changes with more complicated models that have been developed in the last few years. These changes are quite important because they influence the ability of precipitation to remove smoke from the atmosphere and, therefore, the duration of the climatic effects of smoke. Now the last elements can be added to the diagram (Figure 1C). The principal ingredient in Figure 1C is heating of smoke-filled air due to absorption of sunlight by smoke particles. This heating causes changes in the atmospheric circulation and structure (also indicated in Figure 1C) the atmospheric perturbations alluded to in the title of this paper. The first of these perturbations is a major change in the atmospheric circulation patterns that causes the heated air and the entrained smoke particles to rise. This carries some smoke particles well above the altitudes to which they were injected initially by the fires. The second change is one that takes place in the vertical thermal structure of the atmosphere, which is also brought about by the heating of smoke-filled air. As I will show in this paper, both of these effects inhibit the ability of the atmosphere to purge itself of smoke. Specifically, they reduce the efficiency of smoke removal by precipitation. In fact, there is a competition between rainfall, which removes smoke from the atmosphere, and these atmospheric perturbations, which act to isolate smoke from removal by rainfall. Precipitation scavenging begins to act as soon as smoke is injected into the atmosphere. In the model calculations, precipitation is able to remove a substantial amount of smoke during the first two weeks. During that time these perturbations develop and, at least for summertime conditions and large smoke injections, can become dominant. These changes in the atmospheric structure and circulation are important in their own right, but it should be noted that they form a feedback loop in which elements of Figures 1A, 1B, and 1C are interconnected. In the full diagram, the amount (and spatial distribution) of smoke remaining in the atmosphere at any time is influenced by the changes caused by solar heating of the smoke itself. In a given season of the year, the intensity of heating depends on the amount (concentration) of smoke. Consequently, if larger injections of smoke are postulated, stronger heating results and causes larger atmospheric perturbations and greater inhibition of smoke
44 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES removal by rain. Thus, the larger the amount of smoke injected, the greater is its ability to modify the atmosphere and, thereby, to inhibit its own removal. (For very large smoke injections, another effect, discussed by Malone et al. , modifies this conclusion.) For a given amount of injected smoke, the intensity of heating depends on the amount of sunlight that is available. Assuming that smoke would be initially injected only in the Northern Hemisphere, the heating of smoke and the resultant atmos- pheric perturbations would be greater in July than in January, simply because there is more sunlight in the Northern Hemisphere in July than in January. The computer model that we used for our studies is a general circulation model or global climate model, or simply a GCM. It is a three-dimensional model that solves on a computer the mathematical equations describing the evolution in time of the winds, temperature, moisture, and other quantities throughout the earth's global atmosphere. To study the nuclear winter problem, the capability of transporting aerosols (very small particles) with the simulated winds of the model was added. The model's solar radiation scheme was modified to allow for the absorption by smoke particles of sunlight coming into the atmosphere. A very simplified treatment of the removal of smoke from the atmosphere by rainfall was also added. For this rainfall was used as predicted by the model itself, so that changes in rainfall caused by the heat-induced at- mospheric perturbations could be taken into account. In the computer simulation studies that I will describe, smoke was injected into the model atmosphere over the United States, Europe, and the western part of the Soviet Union. The injection rate decreased linearly to zero at day 7. Half of the smoke was injected during the first two days. The sensitivity of smoke transport and removal to the assumed initial vertical distribution of smoke was considered by using two profiles: a low injection with smoke distributed between 2- and 5-km altitude in the lower troposphere, and an NAS injection (so-called because of its use in the study done by the National Academy of Sciences) with constant smoke mass density between the surface and a 9-km altitude (NAS, 1985) but still within the unperturbed troposphere. Both January and July conditions were used to reveal seasonal differences. The behavior of aerosols in the normal atmosphere was studied with a passive tracer which, like smoke, is transported by the model's winds and removed by the predicted rainfall but, unlike smoke, does not absorb sunlight. This last characteristic permits the model atmosphere to evolve unperturbed by the presence of the passive tracer. The contrasting behaviors of interactive smoke and passive tracer illustrate clearly the importance of atmospheric heating due to sunlight absorbed by smoke particles. The amount of smoke that is assumed to be injected into the atmosphere
ATMOSPHERIC PERTURBATIONS OF LARGE-SCALE NUCLEAR WAR 145 is an important parameter, but estimates of this quantity are quite uncertain. The study by the National Academy of Sciences (NAS, 1985) estimated a range from 20 teragrams (Tg; 1 Tg = 1012 grams = 1 million metric tons) up to as much as 640 Tg of smoke. I will present only results for 170 Tg, a value close to the NAS baseline value; results for other smoke amounts can be found in Malone et al. (19861. Now I would like to explain more fully some of the elements of Figure 1. Using July conditions, because the atmospheric changes are larger and more easily seen, I will first describe smoke lofting and then show how the structure of the atmosphere is changed. Next I will describe how these effects influence the removal of smoke by rainfall and the lifetime of smoke in the atmosphere. Finally, I will describe briefly the findings about the climatic impact of smoke. Figure 2 contains a comparison of two calculations that illustrate nicely the influence of solar heating on the dynamics of smoke. One calculation was done with interactive smoke; the results from it are shown with solid contours. The second calculation was done with a passive tracer; its results - 30 25 20 it, 15 10 5 to ' 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 DAY 20 _ rlO .1,: ~ UP 60 30 ED -30 -60 SP LA TITUDE 10 20 30 _` ~2 50 70 100 ~ CO 200 300 500 700 1000 FIGURE 2 Longitudinally averaged mass mixing ratios for July conditions at day 20. The dashed contours apply to a passive tracer, while the solid contours apply to interactive smoke. In each case 170 Tg (1 Tg = 1012 g = 1 million metric tons) of material was injected over the Northern Hemisphere continents with a low injection profile (see text). The contours of mixing ratio are labeled in units of 10-9 g material/g air. Source: Malone et al. (1986, p. 1044~.
146 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES are shown with dashed contours. In both calculations the same amount of material (170 Tg) was injected over the Northern Hemisphere continents in July at altitudes between 2 and 5 km (low injections). The contours indicate the concentrations of material (in parts per billion by mass) re- maining at day 20 in the calculations, averaged over all longitudes. The display extends from the North Pole to the South Pole and from the surface of the earth up to about 30 km, which is in the lower stratosphere. (I will explain a little more about the normal atmospheric structure in connection with Figure 3.) These contours tell us how much of the material is left at day 20 and how it is distributed over latitude and altitude. Most of the passive tracer remains at low altitudes, where it was injected, because the passive tracer and surrounding air are not heated by sunlight. Since scavenging by rainfall is fastest in the lower atmosphere, the passive tracer is rapidly removed, as indicated by the relatively small concentra- tions (Figure 21. In the interactive case, on the other hand, the smoke does absorb sun- light. The heating drives vertical motions that carry smoke-filled air up- ward from the region of injection in the lower atmosphere. This takes some smoke up higher, completely out of reach of removal by precipi- tation. Also, the heating of the atmosphere inhibits the formation of pre- cipitation. This allows more smoke to remain, as can be seen by the larger concentrations on the solid contours. Before showing how the structure of the atmosphere is changed by the heated smoke, let me first describe the atmosphere as it normally exists. Figure 3A displays the longitudinally averaged temperature in the atmos- phere for normal July conditions. The temperature contours are labeled in degrees Kelvin (273°K = 0°C). The structure of the atmosphere in its normal state is such that the temperature is warmest at the surface and decreases upward with height to an altitude of about 10 km. This region is called the troposphere. At about 10-15 km, the temperature becomes relatively constant with height and then increases with height in the strat- osphere because of the absorption of sunlight by ozone. The heavy dashed line in Figure 3A shows the approximate position of what is called the tropopause, which is the boundary between the troposphere and the strat- osphere. For the purpose of this study, the most important characteristic of the troposphere is that it is the region of the atmosphere in which storms and rainfall occur. Since precipitation is the primary removal mechanism for smoke, this is where smoke removal will take place. Figure 3B also displays the longitudinally averaged temperature for July conditions, but with the atmosphere being perturbed by the injection of 170 Tg of smoke. The smoke was injected with constant density from the surface up to about 9 km (NAS injection), so that all of it is in the
ATMOSPHERIC PERTURBATIONS OF LARGE-SCALE NUCl FAR WAR 147 A 30 1 1 1 1 1 1 1 1 1 11 1 1 1 ( 25 23° ~210 ^20 _ ~ 15 I) 10 5 . O B 30 ;S, 20 n 15 A: 10 r '10 __ ) ) 1~1 1 ~: ) NP 60 30 ED -30 -60 SP 20 30 ._` 50 `§ 70 100 200 300 500 700 1000 ~70 \ _ 330~ ) - t'---4270~q o 30 ED -30 -60 SP LA TITUDE 20 30 ~ 50 `§ 70 100 ~ co co 200 300 500 700 1000 FIGURE 3 The longitudinally averaged temperature (degrees Kelvin) in the simulated unperturbed (A) and perturbed (B) atmospheres for July conditions. The perturbed distribution in (B) is a 5-day average beginning 15 days after the initiation of injection of 170 Tg of smoke with the NAS vertical injection profile. The unperturbed distri- bution in (A) is a long-term average. In each figure the approximate position of the tropopause is indicated by a heavy dashed line. Source: Malone et al. (1986, p. 10451.
48 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES unperturbed troposphere and is initially subject to removal by rainfall. The heating by sunlight of this smoke, some of which is carried higher (Figure 2), is quite intense and changes the vertical thermal structure of the atmosphere significantly. Figure 3B shows a S-day average of the temperature during the third week after smoke injection began. There is still a region in the lower atmosphere in which temperature decreases with height; that is, there is still a troposphere. However, the top of the tro- posphere is now at about 5 km, rather than at 10-12 km as in the normal atmosphere. Higher up the solar heating of smoke has raised the temperatures by as much as 50-80°K above normal. A situation now exists in which the smoke has created its own "stratosphere." Above the lowered tropopause [heavy dashed line in Figure 3B], warm air overlies cooler air, a condition that inhibits convective motions that would bring about precipitation. Consequently, precipitation is confined below the tropopause and most of the remaining smoke is above it, as illustrated in Figure 4. The heavy dashed line, taken from Figure 3B, again represents the tropopause, the boundary between the troposphere and the heated region. The cross-hatch- ing shows where precipitation is occurring; clearly, it is confined below the tropopause. The black stippling, which indicates various concentra- tions of smoke, shows that smoke now resides primarily above the tro- popause. Smoke that was below the lowered tropopause largely has been removed by precipitation. Because the remaining smoke is now separated physically from its primary removal mechanism, its lifetime in the at- mosphere is greatly increased. This increased lifetime can be seen in Figure 5, which shows the tem- poral evolution of the total mass of material remaining in the atmosphere. The upper four curves apply to interactive smoke calculations with vertical injection profiles, as indicated, while the lower pair of curves apply to passive tracer calculations with low injection profiles. The vertical axis has a logarithmic scale. The total injection in all of these cases was 170 Tg, a value that is near the top of the diagram. As a result of scavenging by rainfall, none of the curves ever reaches the 170-Tg level. A substantial amount of material is removed while the injection proceeds. The passive tracer curves in Figure 5 approximately represent normal aerosols in the unperturbed atmosphere. Following the cessation of injec- tion at day 7, these curves fall in almost straight lines, which means that material is removed exponentially in time. These two curves provide a useful validation of our model. They tell us that aerosols in the normal atmosphere, as calculated by the model, have a residence time on the order of one week. This is in good agreement with observations. Now, contrast that with the behavior of interactive smoke indicated for July by the upper pair of dashed curves. During the first week or two, a
ATMOSPHERIC PERTURBATIONS OF ~GE-SC~ NUCLEI Wit 149 - ~ ~ I: l::::! ~:l::.:::l::.::.:::l.: :l :! in I I I I I I OAYS 15-20 25 20 C) 15 10 5 o Ad.. ...~/////////5~////~ .. : : : : At/ / / / / 7 / / ~ - -~d~ UP 60 30 ED -30 -60 so LA TITUDE 10 20 30 _ 50 70 100 ~ CO 200 300 500 700 1000 FIGURE 4 The relative positions of the modified tropopause (heavy dashed line) and the precipitation distribution (cross-hatched region below the tropopause), both averaged over days 15-20, and the smoke distribution at day 20 (stippled area above the tropopause) for the 170-Tg NAS case portrayed in Figure 3B. Darker stippling indicates greater smoke loading; the smoke contour intervals correspond to mixing ratios of to x 10-9, 40 x 10-9, and 70 x 10~9 g smoke/g air. These can be compared win We solid contours in Figure 2, which apply to a low injection July case, also at day 20. Source: Malone et al. (1986, p. 10451. substantial amount of smoke has been removed from the atmosphere. This is mostly smoke down low that can be easily removed by rain. But because there is strong solar heating in the Northern Hemisphere in July, the rate of removal of smoke is greatly decreased after the first two weeks. As explained above, this occurs because some smoke has been carried higher in the atmosphere and because the atmospheric structure has been mod- if~ed. Approximately one-third of the mass of smoke initially injected still remains in the model atmosphere after 40 days of the July calculations. This smoke has a very long lifetime in the atmosphere, as indicated by the near constancy after day 15 of the upper pair of dashed curves in Figure 5. Up to this point, I have only talked about July because it is easier to illustrate the interesting effects for July conditions than for January. The upper pair of solid curves in Figure 5 show the interactive smoke results
50 100 CO O 10 CO o 1 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES l _ JANUARY JUL Y 0 5 NAS LOW_ -'~PASSIVE, LOW 10 ~-0 TIME (DA YS) FIGURE 5 The mass of material remaining in the global atmosphere as a function of time. The upper four curves apply to smoke; the lower pair apply to the passive Lacer. Solid and dashed curves indicate January and July conditions, respectively. Labels indicate low and NAS injections. The slopes of the passive Lacer curves at late times yield 1/e-residence times of 5 to 6 days, which agree well with observed residence times of aerosols in the lower troposphere. Source: Malone et al. (1986, p. 10461. for January. Smoke is removed faster in January than in July simply because there is less sunlight in the Northern Hemisphere to drive the atmospheric perturbations that enhance the lifetime of smoke. By the end of six weeks in our January calculations, the remaining fraction of smoke injected with the low and NAS profiles is about 5 and 15 percent, re- spectively, compared with 35 percent in the July cases. Nevertheless, solar heating of smoke does have a significant effect even under winter conditions. After three weeks, there is approximately a factor of three more smoke present in the atmosphere in January than would have been the case without the influence of solar heating (compare the passive tracer curve). In July the comparable ratio of smoke to passive tracer mass is about 10 after 3 weeks. Figure 6 consists of two maps of the world showing the distribution of smoke looking down through the atmosphere at days 20 and 40. Most of the smoke is still concentrated in the Northern Hemisphere. Transport of smoke by the winds has made the geographical distribution of smoke fairly
ATMOSPHERIC PERTU~ATIONS OF ~GE-SC~E NUCLEI Wow A 90 60 o -30 -60 _ Ott -90 II ~I··I.,I I,,I -180 -150 -120 -90 -60 -30 0 30 DAY 20 -¢-~0.1~ ~J 1 _! I . . I . , , . 60 90 B 60 30 o -60 120 150 180 DAY 40 r ~ ...... ; ., 1,, 1, I _90 ~ I I I I I I . -180 -150 -120 -90 -60 -30 , I,, I ., 1,, I . . I . . I 0 30 60 90 120 150 180 FIGURE 6 The vertically integrated solar absorption optical depth of smoke at day 20 (A) and day 40 (B) of the interactive July simulation with 170 Tg injected with the NAS vertical profile. The contours are presented at intervals of 0.1, with the lowest value being 0.1 on the southernmost contour. If ~ is the absorption optical depth, the light reaching the surface from the sun overhead is reduced by a factor of e-T. For ~ = 0.1, 0.3, 0.5, and 0.7, the factor e-T is 0.90, 0.74, 0.61, and 0.50, respectively. Source: Malone et al. (1986, p. 1047~.
152 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES uniform in longitude, although some nonuniforrnities remain. Some low- level smoke lingers over the continents. This is possible because the surface cooling (Figure 7) causes evaporation and precipitation to decrease over the continents. Air over the oceans is clearer. Some smoke has reached the Southern Hemisphere. The quantity displayed in Figure 6 is called the absorption optical depth and can be used to determine the attenuation at the surface of sunlight coming down through the atmosphere. The fractional attenuation is about 10, 25, 40, and 50 percent for optical depths of 0.1, 0.3, 0.5, and 0.7, respectively. Figure 7 shows the changes in surface air temperature, relative to nor- mal, predicted by the model when 170 Tg of smoke is injected in July. A 5-day average of the temperature change near the end of the first week is displayed in Figure 7A. It shows cooling by 15°C or more over large areas of the interiors of the North American and Eurasian continents during this period when the smoke clouds are particularly dense over the regions of injection. The long lifetime of smoke under summer conditions causes significant reductions in the surface air temperature to last through the end of the calculation at day 40. Figure 7B shows the simulated temper- ature changes during week 6; reductions of 5-15°C persist over the north- ern midlatitude continents. The features in the Southern Hemisphere have nothing to do with what is going on in the Northern Hemisphere; they are due simply to normal weather fluctuations in the winter (Southern) Hemi- sphere. For 170 Tg of smoke injected in January, simulated surface air tem- perature reductions of 5-15°C occur over portions of the northern mid- latitude continents during the first few weeks. However, the faster removal of smoke allows the temperatures to recover toward normal more rapidly than in July. The discussion so far has focused on a baseline value of 170 Tg of injected smoke. However, it was pointed out in connection with Figure 1 that the intensity of heating, the magnitude of the atmospheric pertur- bations, and the smoke removal rate all depend on the concentration (hence, total mass) of injected smoke. A very small amount of smoke has little impact on the atmosphere, which allows the smoke to be quickly removed from the troposphere, much like the passive tracer results in Figures 2 and 5. As the injected mass is increased in the simulations into the range estimated for a major nuclear exchange, the solar heating of smoke and the atmospheric perturbations increase in magnitude. The frac- tional mass remaining in the atmosphere at late times also increases, and its rate of removal decreases. This trend continues up to injected masses comparable to the baseline value (170 Tg). With still larger values, another effect comes into play that causes the fractional mass remaining to stop increasing and even to decrease somewhat (Malone et al., 19861.
ATMOSPHERIC PERTURBATIONS OF ~GE-SC~E NUCLEI Wit 153 A 9o 1 30 -30 _ -60 B 60 30 n -30 _ -on _ DAYS 5- 10 I 1 1 ~I ~ 1 1 1 ~1 1 _ _ . - L~ -90 -180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180 ~ ~Do ~;~ a MOO ~) _ to or -5to-15 t ~>+5~/~ DAYS 35 - 40 · '~:: L~ -90 -180 -150 -120 -90 -60 -30 0 30 60 90 120 150 180 -5 to -15 it ~ >~ '1 ~ + 5 // FIGURE 7 The change in surface air temperature relative to the unperturbed atmosphere in July for 170 Tg of smoke injected with the NAS profile. Five-day averages of the perturbed case, minus the long-term average of the unperturbed case, are shown: (A) days 5-10, (B) days 35-40. Only changes larger in mag- nitude than 5°C are shown. Values are indicated at the bottom of the figure; the designation <-15 refers to temperature reductions below normal in excess of 15°C. Note that the warm and cool regions near Antarctica are simply manifes- tations of storms which occur naturally in the wintertime circumpolar flow; they have no connection with the changes occurring in the Northern Hemisphere. Source: Malone et al. (1986, p. 1049~.
154 PHYSICAL EFFECTS AND ENVIRONMENTAL CONSEQUENCES In summary, solar heating of smoke is a very important factor. It produces two effects. One is that some smoke is carried well above its initial injection height. The second is a modification of the atmospheric structure in which heating pushes the tropopause downward. Both effects contribute to isolation of smoke above the tropopause from precipitation below and cause an increase in the lifetime of that smoke relative to what one would find if solar heating of smoke were neglected. The magnitude of these effects depends on the season of year and the amount of smoke injected into the atmosphere by fires. There would be substantial cooling of the Northern Hemisphere con- tinents during the first few weeks in both January and July. In the July case only, the prolonged lifetime of smoke suggests that significant tem- perature reductions could persist for many weeks. Smoke would spread into the Southern Hemisphere in July as a result of the strong circulations driven by the solar heating of smoke. There would be very little spread into the Southern Hemisphere for January conditions; the smoke simply is not heated enough and is removed too fast. Interested readers should consult the paper of Malone et al. (1986) for a more complete discussion that includes the simulated surface climate impact of various smoke amounts. REFERENCES Malone, R. C., L. H. Auer, G. A. Glatzmaier, M. C. Wood, and O. B. Toon. 1985. Influence of solar heating and precipitation scavenging on the simulated lifetime of post- nuclear war smoke. Science 230:317-319. Malone, R. C., L. H. Auer, G. A. Glatzmaier, M. C. Wood, and O. B. Toon. 1986. Nuclear winter: Three-dimensional simulations including interactive transport, scaveng- ing and solar heating of smoke. J. Geophys. Res. 91 (D1):1039-1053. National Academy of Sciences, Committee on the Atmospheric Effects of Nuclear Explo sions. 1985. The Effects on the Atmosphere of a Nuclear Exchange. Washington, D.C.: National Academy Press. Turco, R. P., O. B. Toon, T. P. Ackerman, J. B. Pollack, and C. Sagan. 1983. Nuclear winter: Global consequences of multiple nuclear explosions. Science 222:1283-1292.