2
Elements of a Research Program for Aerosol Forcing of Climate

The scientific problems that need to be addressed to narrow uncertainties in aerosol radiative forcing of climate require a highly coordinated observational, laboratory, and modeling program. Figure 2.1 shows a heuristic diagram of the general components of an integrated aerosol-climate research program. The required research is composed of (1) observations, both short-term intensive observational campaigns and long-term systematic monitoring (both satellite and in situ); (2) process studies, both theoretical and laboratory-based; and (3) model development and evaluation, ranging from process models directed at a single phenomenon to global climate models. The arrows connecting the boxes in Figure 2.1 indicate how results from each component influence other components (i.e., the needed integration).

The ultimate goal of the program is to provide quantitatively accurate calculations of direct and indirect anthropogenic aerosol radiative forcings, as indicated in Chapter 1. Knowledge of these aerosol forcings is necessary both to understand past trends in climate (such as surface temperature records) and to permit accurate projections of future climate change for climate assessments. To accomplish this goal requires improved methods of representing aerosol radiative forcing in global climate models. The purpose of this chapter is to recommend the components of an aerosol research program required to achieve that goal.

The ultimate integrator of our understanding of the effect of aerosols on climate is the global climate model. For this reason, we begin the chapter with a discussion of research needs in global climate modeling for aerosols.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change 2 Elements of a Research Program for Aerosol Forcing of Climate The scientific problems that need to be addressed to narrow uncertainties in aerosol radiative forcing of climate require a highly coordinated observational, laboratory, and modeling program. Figure 2.1 shows a heuristic diagram of the general components of an integrated aerosol-climate research program. The required research is composed of (1) observations, both short-term intensive observational campaigns and long-term systematic monitoring (both satellite and in situ); (2) process studies, both theoretical and laboratory-based; and (3) model development and evaluation, ranging from process models directed at a single phenomenon to global climate models. The arrows connecting the boxes in Figure 2.1 indicate how results from each component influence other components (i.e., the needed integration). The ultimate goal of the program is to provide quantitatively accurate calculations of direct and indirect anthropogenic aerosol radiative forcings, as indicated in Chapter 1. Knowledge of these aerosol forcings is necessary both to understand past trends in climate (such as surface temperature records) and to permit accurate projections of future climate change for climate assessments. To accomplish this goal requires improved methods of representing aerosol radiative forcing in global climate models. The purpose of this chapter is to recommend the components of an aerosol research program required to achieve that goal. The ultimate integrator of our understanding of the effect of aerosols on climate is the global climate model. For this reason, we begin the chapter with a discussion of research needs in global climate modeling for aerosols.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 2.1 General components of an integrated aerosol-climate research program. Aerosol processes that determine radiative forcing (direct and indirect) must be parameterized by way of process models that describe such phenomena as cloud formation and aerosol and cloud optical properties. Process models emerge from process research. Long-term systematic monitoring of aerosol properties can provide data to test not only process models but also global climate models. Intensive field campaigns provide an opportunity to measure a comprehensive set of chemical and physical parameters that govern features of aerosol forcing. A method is required for determining both the individual phenomena that most influence climate forcing by aerosols and those that contribute the most uncertainty. As discussed in more detail in Chapter 3, sensitivity analysis (determining the change in model output resulting from change in each input parameter) is the traditional method. Figure 2.1 indicates that sensitivity analysis should ultimately be performed using both global climate and process models, since the magnitude of the sensitivity may depend on the scale of the phenomenon. These sensitivities are then used in principle to define future research directions to narrow the uncertainties associated with specific processes. Since we are seeking to understand the impact of anthropogenic aerosols on climate, we should explain why we include research in the cleanest

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change parts of the remote marine atmosphere. There are several reasons for this. First, in polluted air masses, the natural and man-made aerosols are so intimately mixed that the exact nature of the anthropogenic contribution becomes blurred. We cannot quantify man's contribution without thoroughly understanding the background aerosol. Second, in many cases, pollution aerosols blow offshore, mingle with marine aerosols, and are removed to the oceans. Yet we have an inadequate understanding of those marine loss processes to achieve the needed accuracy in our models. Finally, the simplicity of remote aerosol systems makes it possible to perform much more rigorous closure and process experiments than are possible in complex, spatially heterogeneous polluted air masses. Since the same basic principles will govern aerosol behavior in both environments, process studies in the simplest remote environments are likely to be more cost-efficient and productive than working in air masses where a wide range of sources might be responsible for observed changes over time and space. We must learn to walk before we can hope to run. Ultimate success of the research plan advanced in this report rests on integration of the various components shown in Figure 2.1. The nature of this program as an integrated plan requires strong coordination. A structure to ensure this coordination is provided in Chapter 4. First, though, detailed information is given about the components shown in Figure 2.1. GLOBAL CLIMATE MODELS The geographic extremes in aerosol forcing are important to the response of the climate system to such forcing. Although globally averaged estimates of aerosol forcing may be reported, to properly construct a global average forcing requires a global-scale model. A global climate model is the most comprehensive tool available for studying the importance of any forcing on the climate system, including aerosol. Such a model consists of a number of components: an atmospheric general circulation model (AGCM), some form of an ocean model, a sea ice model, and a land surface model. An atmospheric chemical transport model (ACTM) is used to simulate the distribution of aerosol. These models continue to grow in complexity as more detailed chemistry and removal processes are added. The AGCM uses information on aerosols to calculate radiative forcing and supplies information (e.g., winds) to the chemical model to determine aerosol distribution. Calculation of direct radiative forcing may not necessarily require all processes described in an AGCM. Indeed, by its very definition, radiative forcing means that no atmospheric process has altered the climate state used to calculate the change in radiative flux caused by aerosol. Aerosol forcing calculations are typically performed in an AGCM, but employ only the

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change radiative transfer part of the AGCM. There are now a number of global model studies of direct aerosol forcing for sulfate aerosols (Charlson et al., 1991; Kiehl and Briegleb, 1993; Taylor and Penner, 1994; Boucher and Anderson, 1995). Kiehl and Rodhe (1995) have reviewed several of these calculations and find roughly a factor of 2 difference in forcing, caused by chemical modeling differences. Calculations of the indirect effect (Chuang et al., 1994; Jones et al., 1994; Boucher and Lohmann, 1995) indicate a forcing magnitude similar to that of the direct effect. Atmospheric General Circulation Models Atmospheric general circulation models are numerical representations of the equations of motion and physical processes that define the workings of the atmosphere. These models are global in spatial extent and represent the equations of motion in either a physical grid space or spectral space (i.e., the dynamical fields are represented by a series of spherical harmonics). Climate models, by their very nature, require time integrations from one season to multiple decades. The horizontal resolution of the AGCMs is usually determined by computational resources. During the past decade there has been increased emphasis on higher spatial resolution. Current AGCMs employ resolutions from 8° X 10° down to 1° X 1°. The number of vertical levels in these models ranges from around 10 to 20. Physical processes included in the AGCMs are radiation, convection, and processes in the boundary layer and the surface layer. Cloud processes related to cloud amount and optical properties are typically included as a part of the radiation calculations. All of these physical processes operate on scales smaller than those resolvable in the AGCM and hence require parameterization. Much of AGCM development over the past 30 years has focused on improving these physical parameterizations, with much of the focus on clouds and convection. At present, there is a diverse range of approaches to the parameterization of these processes, which no doubt reflects limitations in understanding. A list, by no means complete, of some global models currently being used to study aerosol forcing is given in Table 2.1, where the atmospheric general circulation and chemical models are listed separately. The chemical transport models listed in Table 2.1B cover hemispheric to global scales and are suitable for coupling to AGCMs. Atmospheric Chemical Transport Models Predictions of aerosol distributions by an ACTM require information on magnitudes and geographic distributions of emissions of precursor gases and sources of primary particles, chemical reaction rates in the atmosphere, transport of these gases and aerosols by large-scale advection and subgrid-scale

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change TABLE 2.1 Global Models Currently Used to Study Aerosol Forcing Modela Horizontal Resolution Number of Vertical Levels Temporal Resolution (minutes) A. Atmospheric General Circulation Models for Aerosol Forcing Calculations GISS 4° × 5° 9 7.5 LMD   11 6 MPIb 2.9° × 2.9° 19 24 NCARb 2.9° × 2.9° 18 20 UKMO 2.5° × 3.75° 19 30 LLNL-CCM1 4.5° × 7.5° 12 30 B. Global and Synoptic Models for Chemical Transport of Aerosols LLNL Grantour 4.5° × 7.5° 12 6 Stockholm/Mainz Moguntia 10° × 10° 10 Monthly mean NCAR IMAGES 2.5° × 2.5° 25 Monthly mean GChM 1.0° × 1.0° 15 6 GISS 8.0° × 10° 9 6 MPI 2.9° × 2.9° 19 24 a GISS (Goddard Institute for Space Studies), LMD (Laboratoires de Météorologie Dynamique), MPI (Max Planck Institut für Meteorology), NCAR (National Center for Atmospheric Research), UKMO (United Kingdom Meteorological Office), LLNL (Lawrence Livermore National Laboratory), Stockholm (Langner and Rodhe, 1991), IMAGES (Intermediate Model for the Annual and Global Evolution of Species), GChM (Global Chemistry Model; Pacific Northwest Laboratory). b The MPI and NCAR models are spectral; the horizontal resolution listed is an equivalent Gaussian grid. convection, and removal mechanisms (both wet and dry). Links to an AGCM are contained in large-scale advection of gases and aerosols, which employs atmospheric winds, and in convection parameterization, which supplies convective fluxes for subgrid vertical transport. Links are also associated with the land and surface modules in climate models through the representation of dry deposition processes. Wet removal processes in the ACTM require precipitation information from the AGCM. Finally, aqueous-phase reactions require information on cloud water. AGCMs are just beginning to prognostically calculate amounts and distribution of cloud water. The development of many of the current ACTMs began with models that were not coupled to AGCMs. This uncoupling was necessary for accurate testing of transport and chemistry algorithms. These ''off-line" ACTMs employ either AGCM atmospheric state information (e.g., winds, precipitation, cloud variables) or meteorological or climatological state information.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change A number of the off-line models used for predicting sulfate aerosol distributions employ climatological data. Some questions remain about how accurate aerosol distributions are when calculated by using climatological data, since temporal changes in aerosol distribution occur on shorter scales than climatological and the resulting average aerosol distribution may not be well represented from climatological data. To date, much of the global ACTM modeling has focused on sulfate aerosols (Langner and Rodhe, 1991; Chuang et al., 1994; Pham, 1994). These models have treated the aqueous and gas-phase production rates of sulfate in a highly simplified manner, for example, by assuming that the rate of aqueous formation of sulfate is proportional to cloud amount and cloud lifetime. This assumption does not properly represent the rate of formation of in clouds when H2O2 is the limiting species (nor are the roles of trace metals and other oxidizing agents accounted for) and therefore does not represent the expected seasonal variation of sulfate aerosol. Other models that attempt to account for oxidant limitation in simplified ways predict the seasonal cycle of better in source areas over North America than in Europe (Penner et al., 1994a). Reasons for lack of observed seasonal variation in sulfate over Europe are not understood. Oxidation of SO2 on sea salt particles is another process not presently represented in ACTMs. This process operates mainly in the marine boundary layer, and the amount of sulfate produced in this manner is, as yet, poorly quantified. Yet it may be important to represent this process in global aerosol models, to obtain a more complete understanding of the atmospheric sulfur cycle. Only a few model calculations are available of global-scale distributions and climate forcing by components of the aerosol other than sulfate, and these have been limited to smoke aerosols from biomass burning and/or black carbon from fossil fuel combustion (Penner et al., 1993; Cooke and Wilson, 1995; Liousse et al., 1995). This latter component is especially important for determining the light absorption coefficient and single-scatter albedo (a measure of the relative magnitudes of aerosol scattering and absorption) of anthropogenic aerosols. As yet, no global-scale model calculation of the combined effects of all anthropogenic aerosol components (sulfate, nitrate, ammonium, organic carbon, black carbon, anthropogenic dust aerosol) and their climate forcing is available, although one-box model-based analysis of all these components has been carried out (Pilinis et al., 1995). Additionally, to test model predictions against measurements (thereby performing a "closure" experiment), it is necessary that natural components as well as anthropogenic components be represented. Major uncertainties surround the prediction of aerosol components other than sulfate, with the primary uncertainty related to inadequate specification of source rates. In the case of carbonaceous aerosols (organic and black carbon), little infor-

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change mation regarding natural and anthropogenic source strengths is currently available (Penner, 1995). Existing models for aerosol forcing calculations do not explicitly carry detailed information on particle size; the approach is to predict aerosol mass. It is well established, however, that both direct and indirect radiative effects of aerosols can depend strongly on the size distribution of the aerosol (Boucher and Anderson, 1995; Nemesure et al., 1995; Pilinis et al., 1995). Thus, global aerosol models will eventually need to include explicit aerosol size information. Attempts to calculate indirect climate forcing by anthropogenic aerosols have all been based on predicted geographical distributions of anthropogenic sulfate aerosol (Boucher and Rodhe, 1994; Chuang et al., 1994; Jones et al., 1994; Boucher and Lohmann, 1995). These approaches empirically relate cloud drop number concentrations, and thus cloud optical properties, to sulfate aerosol mass concentration derived from a chemical transport model. Whereas this approach is arguably the best starting point to evaluate the indirect effect, the underlying assumptions do not address physical cause-and-effect relationships in a fundamental manner. For example, emissions of aerosol precursors and primary emissions are known to vary by region. Such variations will alter the ratio of to other aerosol components by region. Also, the functional relationship between cloud condensation nuclei (CCN) concentration and sulfate mass is expected to be nonlinear. For example, if most anthropogenic sulfate is formed via aqueous processes in clouds, then adding more SO2 will mainly form larger particles rather than additional CCN, and only a small fraction of anthropogenic sulfur emissions would form new CCN (e.g., see Leaitch et al., 1992), although the climate forcing might still be significant (Chuang and Penner, 1995). Most of these models have further assumed that observed relationships between either cloud water sulfate and cloud droplet concentrations or between cloud base aerosol concentrations and cloud droplet number concentrations could be used to represent the response of cloud droplets to new aerosol particles or sulfate mass. It is not clear that these relationships can be used to analyze the response of climate forcing to either increases or decreases in anthropogenic aerosol emissions, because such changes may involve a change in aerosol size distribution. However, one attempt to account for the effects of changes in size distributions, which notes the importance of the aqueous pathway, still finds significant forcing (Chuang et al., 1994). Data relating sulfate and/or other aerosols to cloud droplet concentrations are sparse to nonexistent for remote marine clouds and can give conflicting low susceptibility when account is taken of the type of mixing processes within the clouds (Novakov et al., 1994). These processes have not been accounted for in any model estimates of indirect climate forcing.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change Because prediction of both homogeneous and heterogeneous rates of formation of sulfate depends on understanding and predicting the photochemistry of ozone and other oxidants, especially H2O2, understanding sources of discrepancy between modeled abundances and observations may well require development of coupled global aerosol-photochemical models. Photochemical models of the global troposphere are just now becoming available (Crutzen and Zimmerman, 1991; Atherton, 1994; Müller and Brasseur, 1995), and none has been self-consistently integrated with an aerosol model, even in simplified form. Finally, the coupling of the atmospheric and chemical components into a complete system model will no doubt introduce new nonlinear interactions within the coupled model. For example, addition of aerosols to the system will have a tendency to lower surface temperatures. This reduction in surface temperature can, in turn, lead to changes in evaporative fluxes, which can alter the hydrologic cycle (e.g., reduction in precipitation). Changes in the hydrologic cycle will then modify chemical production and removal of aerosols. Each link in this feedback loop is highly nonlinear and can alter the sensitivity of the overall climate system. The addition of feedback processes involving atmospheric aerosols is imperative for a credible climate system modeling program and represents a major challenge for atmospheric science in the twenty-first century. Recommended Research on Global Climate Modeling of Aerosol Radiative Forcing Global climate modeling of aerosol radiative forcing requires the following: major advances in the representation of indirect climatic effects of aerosols in global climate models by treating in a fundamental manner the relationship between aerosol mass and aerosol number, aerosol number and CCN number, CCN number and cloud drop number concentration (CDNC), and CDNC and cloud optical properties; development and evaluation of coupled global aerosol-photochemical models; coupling of atmospheric chemical transport and aerosol models into a global climate model system; improvements in models of the sulfur cycle and development of models for other aerosol types and for mixtures—it is particularly necessary to develop models with a time resolution that is short compared to those of the relevant meteorological processes (e.g., synoptic time scale); and evaluation and representation of aerosol sources and precursor gases for aerosol chemical models.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change PROCESS RESEARCH Process studies are an essential part of the overall program design, because they provide the means to identify, understand, and quantitatively describe the processes controlling aerosol forcing of climate, as well as to develop and test parameterizations used in models, and because existing knowledge of the pertinent processes is inadequate to the required level of certainty. Evaluation of the sensitivity of process model results to uncertainties in model formulation and input parameters is necessary for guiding observational programs. Some process studies are appropriately conducted in the laboratory, and some require small-scale field experiments, whereas others involve major, multi-investigator, and multiplatform field campaigns. Because of the significant costs associated with large field campaigns, such studies are treated separately from smaller-scale process research. For both direct and indirect forcing of climate by aerosols, the connection between properties such as aerosol mass and the actual radiative forcing is mediated by a series of intervening linkages, some of which are poorly characterized. For one family of important species, sulfates, these linkages are shown schematically in Figure 2.2. Process studies seek to elucidate the individual linkages, some of which are very complex and nonlinear, such as that between sulfate aerosol mass and CCN concentration, as noted earlier. Although process studies are generally focused on a single linkage or process, they often benefit from the context provided by conducting them alongside related observations in an integrated field program. As noted earlier, many of the properties, both physical and chemical, that determine aerosol forcing result from processes that occur on scales unresolved by general circulation models (GCMs). These processes therefore need to be parameterized in terms of prognosticated large-scale variables (e.g., temperature, specific humidity, cloud water). A complicating factor in these parameterizations is that many of the processes are highly nonlinear, for example, the relationship between changes in SO2 emissions and CDNCs. To develop and test these parameterizations requires process models. For example, fine-scale cloud models (two or three dimensional) that explicitly resolve scales of motion down to 500 to 1000 m, with explicit treatment of cloud microphysics, have been used to study the effect of increased CCN on cloud droplet number (Flossmann and Pruppacher, 1988). We begin this section with a summary of optical properties, then proceed to aerosol formation, transformation, and removal processes, and finally consider aerosol process models that integrate all of these features. Optical Properties The basic quantities needed to describe the direct interaction of aerosol particles with solar radiation are the aerosol optical depth δ(λ), single scattering

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 2.2 Direct and indirect forcing mechanisms associated with sulfate aerosols.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change albedo ω(λ), and angular scattering function β*(φ). The aerosol optical depth is the vertical integral of the aerosol extinction coefficient σe(λ) = σsp(λ) + σap(λ). The single scattering albedo, a measure of the relative magnitudes of aerosol scattering and absorption, is defined as ω = σsp/σe. It can be measured directly in situ by separate measurement of σsp (with an integrating nephelometer) and σe (with a long-path extinction cell or atmospheric path) or from separate determinations of σsp and σap. The angular scattering function describes the angular distribution of the intensity of light scattered by particles. Parameterizations in radiative transfer models use an integral property of the angular scattering function, either the asymmetry factor g, defined as or the upscatter fraction β(μ), defined as the fraction of solar radiation scattered in the upward direction (back to space), where μ is the cosine of the solar zenith angle (Wiscombe and Grams, 1976). The angular scattering function β*(φ) can be determined with a polar nephelometer (Hayaska et al., 1992; Jones et al., 1994), but commercial versions of this instrument are not available. No instruments exist that directly determine g or β(μ). However, an integrating nephelometer with a special backscatter shutter can determine the hemispheric backscattering fraction b =β(1), which can be used to estimate the value of the asymmetry factor. The notion of describing sulfate forcing separately results in a negative forcing term (δF/δ), whereas δF/δ soot is positive (see Table 1.3). Absorbing aerosols can have either a positive or a negative radiative forcing (i.e., produce either a heating or a cooling effect), depending on the surface albedo and the ratio of aerosol light absorption to hemispheric backscattering coefficients (1-ω)/b (Chylek and Coakley, 1974; Haywood and Shine, 1995). For a given surface albedo, there exists a critical value of (1-ω)/b below which the aerosol leads to cooling, and above which to heating. The few measurements that are available (Waggoner et al., 1981) yielded values of (1-ω)/b in the range 1-3 in rural areas and 3-9 in urban areas. For the radiative transfer model used by Chylek and Coakley (1974), these results would imply a positive forcing in rural areas with surface albedos larger than 0.13-0.27. This range of surface albedos includes values expected for rural areas and points out the need for systematic, combined measurements of aerosol single scattering albedo and hemispheric backscatter fraction in order to represent properly the radiative effects of absorbing aerosols. Liousse et al. (1995) deduce global maps of ω that show, in general, a cooling effect of aerosols. Given particle size distribution and chemical composition, Mie theory

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change the troposphere, given supporting in situ measurements, can meet the required accuracy demands. Nonetheless, the diversity of current and future spaceborne instruments, such as MODIS (the Moderate-Resolution Imaging Spectroradiometer) and MISR (the Multiangle Imaging Spectroradiometer), is expected to enhance overall understanding of tropospheric aerosol characteristics throughout the atmosphere. Undoubtedly, the most significant shortcoming in the current and planned suite of instruments is the lack of any instrument designed specifically to profile tropospheric aerosols. This task requires new and innovative techniques such as a spaceborne lidar system or some as yet unspecified approach. In order to fulfill the need for a global data set of aerosol measurements, we believe that a limb instrument operating in visible and near-infrared wavelengths, such as SAGE III, in an inclined orbit will provide the stratospheric data required. However, the next flight of SAGE III in this type of orbit does not occur until 2001 or later. This virtually ensures a substantial break in the stratosphere aerosol climatology initiated by SAM II in 1978. We strongly support a flight of opportunity for SAGE III at the earliest possible date in an inclined orbit (like SAGE II's). In addition, we believe that spaceborne lidar offers the greatest likelihood of producing the data set required to understand the impact of tropospheric (and anthropogenic) aerosols on climate. The strengths of such a system relative to other nadir-viewing instruments include high vertical and horizontal resolution; an indifference to land/ocean effects; only a second-order model influence in the retrieval of the backscatter profile (estimation of the extinction-to-backscatter ratio); and the ability to measure between, and often through, clouds. Since it is unlikely that cross-track scanning will be a part of any first-generation spaceborne lidar, the cross-track horizontal resolution is reduced relative to nadir imagers such as AVHRR and MODIS. The combination of a carefully planned in situ measurement network, and a satellite platform with a cross-track scanning imager sensitive to aerosols and a spaceborne lidar, such as SPARCLE, would be the required approach for depicting global tropospheric aerosols. In Situ Monitoring of Aerosols Satellite measurements are appealing because of their global coverage. However, satellites alone cannot yield the necessary chemical or microphysical data. The strength of surface-based, in situ observations is in providing temporally continuous and detailed information on chemical and microphysical properties; their weakness is limited spatial coverage: in situ measurements at fixed sites can cover only a limited geographical area and only at the surface. Surface-based remote sensing techniques permit measurements of aerosol optical depths and vertical profiles of aerosol backscattering.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change Adding these measurements to in situ surface measurements allows ground-level results to be placed in context with vertical variations. In situ observations provide mean values and temporal variability of key parameters required by global-scale models. Measurements of intensive aerosol properties at a number of sites, carefully stratified according to aerosol type (by using air mass trajectories and aerosol chemical composition, as well as gas-phase tracers of anthropogenic sources), can guide choices for numerical values of key parameters for models. Additionally, spatial distributions determined from a network of ground-based measurements, supplemented by regular horizontal transects obtained from aircraft and by regular vertical profiles obtained from aircraft and balloons, allow large-scale closure experiments on three-dimensional distributions of chemical species calculated from models. Such a set of observations is also needed to develop and validate the algorithms that are used to derive aerosol properties from satellite remote sensing data, as well as the algorithms that are used to remove the confounding effects of aerosols from other remotely sensed properties (e.g., sea surface temperature). Two different strategies for in situ observations are required to address different uses of intensive and extensive aerosol properties. Measurements of a few extensive aerosol properties, with aerosol optical depth having the highest priority, are needed at a sufficient number of locations to provide the geographical variability suitable for evaluating predictions of chemical transport and radiative transport models. Additional extensive properties (such as size-resolved mass concentrations for dominant chemical species) are highly desirable, but the cost and complexity of the measurements limit the number of sites at which these could be made. Instead, efforts should be made to use data from existing networks to validate model predictions of the spatial and temporal distribution of the mass concentrations of important chemical species. In a similar vein, existing networks that monitor aerosol wet deposition could provide data for validating model predictions of removal fluxes. Much more complex instrumentation is required for measuring intensive aerosol properties, requiring a different monitoring strategy. Table 2.3 lists aerosol parameters that we recommend be monitored continuously to determine the most important intensive aerosol properties. Table 2.4 lists the categories of sites that are needed. The recommended measurements can provide a continuous time series of all intensive aerosol properties needed for calculating aerosol radiative forcing, except angular scattering function (or asymmetry parameter) and humidity dependence of aerosol light absorption. Methods for direct determination of the latter two do not presently exist, and instruments to determine the angular scattering function are not commercially available for routine monitoring applications. Inclusion of the backward hemispheric component of aerosol light scattering will

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change TABLE 2.3 Aerosol Properties Needed at Continuous Monitoring Sites Description Recommended Approach Mass concentration of important chemical species including major ions, organic carbon, elemental carbon, and trace elements; the total mass concentration should also be determined Impactor/filter sampler; ion chromatography, combustion, PIXE.a Measurements should be obtained in two size fractions (sub-and supermicrometer) Scattering, hemispheric backscattering, and absorption components of the aerosol light extinction coefficient, in the range 0.35-0.90 µm Integrating nephelometer, continuous light absorption photometer Total number concentration Condensation particle counter Vertical profile of aerosol backscatter Lidar Aerosol optical depth at ∼5 wavelengths in the range 0.35-0.90 µm Tracking sun photometer or shadowband radiometer Surface radiation budget Pyranometer, pyrgeometer, pyrheliometer Hygroscopic growth factor for scattering and hemispheric backscattering components of the aerosol light extinction coefficient Humidity-controlled integrating nephelometer Number size distribution, 0.05-5 µm diameter Differential mobility analyzer, optical particle counter, aerodynamic particle spectrometer NOTE: Measurements of hygroscopic growth factor and number size distribution are not required continuously. These can be surveyed intermittently with a sampling package that circulates among the sites. a PIXE = proton-induced x-ray emission analysis. at least provide a substantial constraint on the range of possible values for the asymmetry parameter. It may become appropriate to add measurements of CCN spectra at a later date, but considerable process research and instrument development are necessary before a suitable sampling strategy can be devised. These different strategies for continuous measurements of intensive versus extensive aerosol properties dictate different station densities. A high-density network is needed for aerosol optical depth, whereas a limited number of sites, each located in an area dominated by a different aerosol type, is appropriate for monitoring intensive aerosol properties. Model results of

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change TABLE 2.4 Categories of Sites to Monitor Intensive Properties Category Aerosol and Sampling Characteristics Polluted continental Industrial and other anthropogenic aerosols Polluted marine Anthropogenic aerosols, sampled after one or more days of transport and transformation after leaving the source regions Clean continental Characterize continental aerosols under conditions of minimal anthropogenic influence, for comparison with polluted conditions; given the wide variety of continental conditions (forest, desert, plains), it will be necessary to sample at a number of sites, although perhaps not simultaneously Clean marine Characterize marine aerosols under conditions of minimal anthropogenic influence, for comparison with polluted conditions Biomass combustion Aerosols formed by combustion of biomass Mineral dust Windblown dust, sampled after one or more days of transport and transformation after leaving the source regions Free troposphere Characterize aerosols above the boundary layer under conditions of minimal anthropogenic influence, for comparison with polluted conditions Stratosphere Surface-based monitoring impossible; primary in situ monitoring effort should be balloon-based measurements of the vertical profile of aerosol size distribution Benkovitz et al. (1994) yielded a characteristic autocorrelation distance for sulfate column burden over the Atlantic Ocean of around 1000 km. To characterize aerosol optical depth distributions over North America, where even shorter characteristic autocorrelation distances are expected because of the proximity to anthropogenic sources, would require a network of at least 30 stations. The now-defunct aerosol optical depth network operated by the National Oceanic and Atmospheric Administration (NOAA) in the 1960s and 1970s (Flowers et al., 1969) had about 40 stations spanning the United States. A similar estimate of the number of stations needed to characterize the aerosol intensive properties cannot be made, because of the lack of data or model results on the expected variability. Instead, the recommended strategy is to begin with one station in each category and use the results from the initial stations and the airborne survey flights (see discussion of mobile platforms below) to decide whether additional stations are needed. Although many local, state, national, and international monitoring networks include aerosol measurements, many of the available data are of limited utility for climate forcing calculations because particle sampling

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change lacks an appropriate upper size limit. Measurements of total suspended particulate mass and measurements made according to the U.S. PM-10 standard (particulate matter of diameter less than 10 µm) are largely unsuitable because they include particles that often dominate total mass but contribute little to optical or cloud-nucleating properties. Furthermore, production mechanisms for submicrometer and supermicrometer particles are very different, and a total sample will contain contributions from both types of sources. Such uncontrolled measurements may be useful in some cases for comparing measured mass concentrations of specific chemical species with model predictions, but extreme care must be taken to ensure that all sources contributing to the total sample are modeled. This problem is somewhat reduced, but not eliminated, with measurements made according to the U.S. PM-2.5 standard, which excludes particles larger than 2.5 µm in diameter. Extensive aerosol properties determined with this size cut are suitable for comparisons with model predictions of aerosol mass concentrations. These data are also useful for determining aerosol mass balance by chemical species, which can be used to estimate relative contributions of different species to light extinction. Measurements of mass scattering or absorption efficiencies made with a size cut larger than about 1-µm diameter, however, are very sensitive to local variations of supermicrometer particles. Measurements of the surface radiation budget (upward and downward, solar and terrestrial radiation) are recommended at sites where the full complement of aerosol properties is measured. Aerosol changes have been indicted as major contributors to changes in the radiation budget, and dedicated networks exist for monitoring changes in solar radiation reaching the surface (WCRP, 1991). These networks, however, are not measuring aerosol properties; thus, interpretation of the effects of aerosols on results from these networks is severely limited. By complementing fully instrumented aerosol monitoring sites with radiation budget measurements or by co-locating new aerosol monitoring sites at existing radiation budget [e.g., Atmospheric Radiation Measurement (ARM) Program] sites, it will be possible to evaluate the effects of aerosol on the surface radiation budget at least at a few locations. Measuring the surface radiation budget together with aerosol properties is needed, also, to evaluate effects of changes in aerosol on cloud optical properties. Satellite observations of cloud albedo and the radiation budget at the top of the atmosphere, plus surface-based observations of aerosols and the radiation budget, are expected to permit initial estimates of the sensitivity of cloud optical properties to below-cloud aerosol. In addition to obtaining long-term measurements, the continuous monitoring program must also include a research component into the way monitoring is conducted. Use of standardized sampling protocols is essential for

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change obtaining comparable results, but no single protocol is optimal for all conditions and chemical species. For example, humidity-controlled sampling is necessary to ensure that the results are not controlled by variations in atmospheric relative humidity. However, the process of lowering the relative humidity also can affect the chemical composition of the particles, possibly introducing systematic biases into the measurements. The magnitude of such effects must be evaluated over the broad range of conditions encountered at the sampling sites, and revised or supplemental sampling protocols must be developed to remove any such biases. The measurement strategy also needs a periodic evaluation of the results of the monitoring program, to address the questions of the duration of measurements and necessary changes in network density or experimental approach. Meteorological variability from year to year suggests that an initial observational period of 5-10 years is needed, and may have to continue even longer if significant trends or large variability in the results is observed. Recommended Surface-Based Monitoring Programs We recommend establishing a dual-density network of surface-based stations for continuous aerosol monitoring, consisting of a high-density network of ca. 30 stations for aerosol optical depth and a low-density network of 7 stations (Table 2.4) to provide detailed information on means, variability, and trends of key aerosol radiative, chemical, and microphysical properties (Table 2.3) for different aerosol types. Mobile Platforms The notion of regularly using a single aircraft to support surface observations in a nearly continuous circuit of geographically distributed sites is both novel and very attractive. Such an airplane could operate essentially continually, shuttling to sites across the United States with a roughly weekly schedule. Instrumentation on the aircraft should be kept fairly simple: a suite of spectral radiometers; in situ measurements of aerosol scattering, hemispheric backscattering, and absorption coefficients; and size-resolved chemical samplers. This suite of measurements covers the most important extensive and intensive aerosol properties. By shuttling between surface monitoring sites (where cooperative vertical soundings would be made), not only would the surface program be routinely augmented with soundings, but observation over a unique geographical-scale component could be added to the program. Similarly, a routine monitoring program from ships would add a valuable component to model validation studies (Figure 2.4). This approach is

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 2.4 Ship tracks off the coast of Northern California. an integral part of flask-sampling networks for monitoring atmospheric greenhouse gases from ships carrying freight along regular routes. For logistical reasons, the suite of measurements would have to be quite limited. As a minimum, size-resolved samples and scattering coefficients of the aerosol should be collected in fixed, predetermined regions along the track, with subsequent analyses for total mass, major ions, and light absorption. If logistical constraints allow, measurements of aerosol optical depth and aerosol light scattering and absorption coefficients should also be obtained. The recommended approach for routine monitoring of the stratospheric aerosol is via satellite. However, satellite observations must be complemented with a regular program of balloon-borne observations to provide ground-truth information. The 20-year record of vertical profiles of aerosol size distribution at Laramie, Wyoming, should be continued to satisfy this need.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change Recommended Mobile Monitoring Programs We recommend a systematic, limited-duration (three-to five-year) program of ship-and airborne surveys to characterize the horizontal and vertical distributions, over the global oceans and North America, of the same aerosol properties studied in the low-density, surface-based network. In addition, we recommend continuing the long-term record of vertical profiles of aerosol size distribution at Laramie, Wyoming. We recommend that a minimum of one aircraft make weekly flights to systematically make vertical profile measurements of a subset of aerosol properties. We also recommend that a least two ships each in the Atlantic, Pacific, and Indian Oceans be outfitted with continuous and flask-monitoring systems. RECOMMENDED TECHNOLOGY DEVELOPMENTS The proposed systematic program of closure studies (aimed at testing the internal consistency of measurements and models) will undoubtedly reveal situations in which current measurement technology is inadequate. Furthermore, there are areas in which development efforts are needed immediately to provide instrumentation that is suitable for continuous monitoring and process/closure studies. Also, the integrated program proposed herein will require continued advances in modeling capability in both computational hardware and software. Parallel Architectures: One of the most computer-intensive calculations with regard to modeling aerosol behavior is prediction of aerosol size evolution. At present, it is not feasible to include aerosol size resolution in global-scale models; however, massively parallel computers have the potential to greatly speed calculations and perhaps eventually allow size and composition resolution in global aerosol models. Research should be initiated on the use of parallel computers for coupled atmospheric chemistry-aerosol models. Coupled with the use of massively parallel computers, numerical methods for solving the basic equations of the models should be reassessed. In Situ Measurement of Aerosol Light Absorption: Suitable methods exist for determining the dependence of aerosol light scattering on humidity. For aerosol light absorption, however, humidity dependence has not been measured and suitable methods are unavailable. Existing methods for determining absorption in all but the most polluted situations require particle concentration on filters to produce a measurable change in transmitted light. Unfortunately, the humidity dependence of light absorption will likely be different for deposited versus airborne particles because of the optical and physical effects of contact with a filter substrate. We recommend

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change that research be initiated on the development of in situ techniques for measuring aerosol light absorption. Angular Scattering Function and Asymmetry Factor: A few polar nephelometers, which measure the angular scattering function β(φ,λ), are in existence (Hayasaka et al., 1992; Jones et al., 1994), but none is being used for continuous monitoring or closure studies. These instruments are needed for determining the asymmetry parameter g(λ), for which no method of direct determination exists. Size distribution measurements and numerical inversion measurements of other aerosol optical properties can be used to estimate the asymmetry factor, but thorough closure experiments for these approaches require a direct measurement of either β(φ,λ) or g(λ). Systematic Vertical Profiling: Interpretations of observations from surface-based facilities require assumptions about representativeness of the observations to conditions throughout the vertical column. Systematic profiles obtained from the proposed airborne observational program will allow these assumptions to be tested, but more frequent observations at surface stations will be needed to allow evaluations of the importance of vertical inhomogeneities to measured aerosol properties. Tropospheric aerosol lidars measure vertical profiles of the volume aerosol backscatter cross section βπ (Hoff, 1988; Radke et al., 1989b; Ansmann et al., 1990). Relatively constant values of βπ in the boundary layer, with lower values aloft, would support the assumption that surface-based measurements represent optical properties of the part of the vertical column that dominates the direct aerosol radiative forcing; conversely, layers aloft with higher values of βπ would indicate that surface observations may be unrepresentative. Although methods exist for estimating aerosol extinction profiles from lidar measurements of βπ (Kovalev, 1993), the relationship is ambiguous and additional measurements (e.g., from the airborne observational program) are needed to resolve the ambiguity. However, for the purpose of identifying cases in which the surface measurements may be nonrepresentative, this ambiguity is acceptable. Analysis of tropospheric data from existing operational lidars would be a logical first step. Refractive Index: Most methods for determining the real part of refractive indices of aerosol particles require particle collection on filters for subsequent chemical or optical analysis. For some closure studies, these methods provide insufficient time and/or size resolution for adequate inference of the refractive index. New instruments for this measurement have the potential to surpass these limitations, but they cannot yet be used routinely for process/closure studies or continuous monitoring programs. Cloud Nucleating Particles: Several different types of CCN counters are in use today. Most have their roots in diffusion cloud chambers. In the thermal gradient cloud chamber, the aerosol is processed through a channel in which wetted parallel walls are maintained at slightly different temperatures.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change Each wall is coated with a water-absorbing material, so that saturated conditions are maintained at each surface. Because of the temperature gradient between the two walls and the nonlinear dependence of the equilibrium vapor pressure on temperature, the saturation ratio increases from unity at the walls to a maximum in the middle of the channel. To obtain a distribution of supersaturations, a sequence of supersaturations is produced for successive measurements. The limited time and supersaturation resolution of such measurements motivated Fukuta and Saxena (1979a,b) to develop a transverse thermal gradient CCN spectrometer capable of producing a range of supersaturations typical of those encountered in natural clouds and-yielding near real-time measurements of the CCN spectrum. The range of supersaturations achievable with this instrument is about 0.1 to 2 percent (i.e., the high end of those encountered in cloud environments). Hudson (1989) developed an instrument that both increased the time resolution beyond that achievable with the transverse gradient CCN counter and extended the range of supersaturations to values as low as 0.02 percent. The instrument does not provide an absolute measurement of critical supersaturation. Instead, it must be calibrated by using particles of known critical supersaturation (i.e., known composition and size). Moreover, the use of the final size to infer critical supersaturation makes the longitudinal gradient CCN counter sensitive to anything that might cause variations in growth rates. Experimental evidence that the condensation coefficient decreases with time following the onset of cloud formation (Hagen et al., 1989) suggests that the response of the instrument might vary with cloud age. None of the research-grade instruments that exist for determining number concentration of CCN as a function of supersaturation Nccn(S) is suitable for routine operation in a monitoring network or for unattended operation in an aircraft. There is a critical need for a compact, robust instrument to measure CCN spectra in a monitoring network or in airborne mode. Lightweight Samplers: Several new sampling platforms are being developed to reduce the cost of obtaining vertical profile information. Remotely piloted aircraft, balloons, and kites are (or will soon be) used to make measurements above the surface. Unfortunately, many of the sampling devices now in use on other platforms are unsuited to these low-power environments. It is particularly difficult to collect samples for chemical analysis since the large weight and power requirements of typical filter samplers cannot be flown by these systems. An aerosol collection device optimized for low weight and power would make it possible to sample aerosols above the boundary layer from an inexpensive kite or balloon at a much lower cost than flying a research aircraft. In view of the need for data on concentrations above the surface, the development of lightweight particle samplers needs to be pursued.

OCR for page 35
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change Buoy-Mounted Instrumentation: Over most of the Earth (the oceans), it is impossible to make long-term in situ observations of aerosol properties. The oceanographic and meteorological community has addressed similar sampling problems by investing in the engineering necessary to make some instrumentation compatible with long-term operation of buoys. Traditionally touchy devices, such as sonic anemometers and instruments that measure salinity and some chemical species, are now routinely operating in the middle of the oceans on buoys that are visited only annually. If one were to engineer a seawater DMS probe, an optical particle counter, or related aerosol-measuring devices to work unattended for months at a time on buoys, one could vastly improve the climatology of aerosol concentrations in remote regions. The engineering capability already exists and simply needs to be applied to the most important sensors. SUMMARY OF A RESEARCH PROGRAM ON AEROSOL FORCING OF CLIMATE We recommend an integrated program of research on aerosol forcing of climate that includes advances in the representation of aerosols in global climate models, particularly with respect to indirect climatic effects; laboratory, theoretical, and field research on aerosol optical properties; identification of aerosol molecular composition, particularly the organic fraction; development of an understanding of aerosol formation and growth in the atmosphere; elucidation, through laboratory, theoretical, and field studies, of the aerosol-CCN-cloud droplet-albedo relationship; execution of atmospheric closure experiments to test theoretical understanding; development of a new satellite system for remote sensing of tropospheric aerosols; establishment of in situ aerosol research measurement stations to provide continuous data on aerosol amounts and properties in key global areas; advancement of instrumentation technology for measuring aerosol properties in situ; and system integration and assessment. Addressing these needs will require a systematic and patient approach. A crash program is not called for; rather, a systematic development of capabilities should be pursued over a period of the order of a decade.