3
Sensitivity/Uncertainty Analysis and the Setting of Priorities

In this chapter, the goal is to present a framework capable of focusing disparate elements into a coherent and cost-effective whole (i.e., an integrating framework for setting priorities). This framework is constructed via sensitivity/uncertainty analyses, and we call the resulting program the Interagency Climate-Aerosol Radiative Uncertainties and Sensitivities (ICARUS) Program.

The proposed ICARUS program would be an additional major component of a larger program dealing with climate change, which in the United States is known as the Global Change Research Program (USGCRP). The goal of the USGCRP is ''to gain a predictive understanding of the interactive physical, geological, chemical, biological, economic, and social processes that regulate the total Earth system, and, hence, establish a scientific basis for national and international policy formulation and decisions relating to natural and human-induced changes in the global environment and their regional impacts." The ICARUS program should be incorporated and administered as a part of the USGCRP.

As a part of the USGCRP, priorities for ICARUS would be required to fit within the "prioritization scheme" used by USGCRP and would appropriately be judged relative to these existing priorities. Correspondingly, the overall budget for ICARUS would have to be consistent with the priorities USGCRP assigns to the scientific contributions required from ICARUS, weighting cost appropriately with feasibility and with the strategic and integrating priorities of the USGCRP.



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 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change 3 Sensitivity/Uncertainty Analysis and the Setting of Priorities In this chapter, the goal is to present a framework capable of focusing disparate elements into a coherent and cost-effective whole (i.e., an integrating framework for setting priorities). This framework is constructed via sensitivity/uncertainty analyses, and we call the resulting program the Interagency Climate-Aerosol Radiative Uncertainties and Sensitivities (ICARUS) Program. The proposed ICARUS program would be an additional major component of a larger program dealing with climate change, which in the United States is known as the Global Change Research Program (USGCRP). The goal of the USGCRP is ''to gain a predictive understanding of the interactive physical, geological, chemical, biological, economic, and social processes that regulate the total Earth system, and, hence, establish a scientific basis for national and international policy formulation and decisions relating to natural and human-induced changes in the global environment and their regional impacts." The ICARUS program should be incorporated and administered as a part of the USGCRP. As a part of the USGCRP, priorities for ICARUS would be required to fit within the "prioritization scheme" used by USGCRP and would appropriately be judged relative to these existing priorities. Correspondingly, the overall budget for ICARUS would have to be consistent with the priorities USGCRP assigns to the scientific contributions required from ICARUS, weighting cost appropriately with feasibility and with the strategic and integrating priorities of the USGCRP.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change Although these constraints on the ICARUS program as a part of the USGCRP would be substantial, the benefits would be enormous, including utilization of the existing USGCRP administrative structure and integration with existing USGCRP measurement and modeling capabilities. INTEGRATION VIA SENSITIVITY/UNCERTAINTY ANALYSES To define priorities within any research program, we accept and endorse the time-honored rational, quantitative procedure of performing sensitivity analyses and then weighting the results appropriately with feasibility and various other factors (such as cost) imposed by broader systems. However, major difficulties are encountered when sensitivity analysis is applied to climate forcing by aerosols. To guide decisions associated with potential climate change, sufficiently reliable models of the climate must be developed. To develop reliable climate models, reliable models of climate forcing by airborne particles from future emissions are needed. Therefore, central to the ICARUS program is development of the understanding of aerosol forcing, as embodied in climate models: to describe, realistically, existing concentrations and properties of aerosol particles throughout the global atmosphere, thereby to provide climate models with needed predictions of current and future anthropogenic aerosol radiative forcings, thereby to meet the prime goals of the ICARUS program and the USGCRP, plus to establish and continuously update rational choices of ICARUS research priorities via sensitivity analyses, and to integrate data, theory, and applications. AN EXAMPLE OF SENSITIVITY ANALYSIS: DIRECT RADIATIVE FORCING We present here an example that illustrates sensitivity analysis but not uncertainty analysis (uncertainties are not evaluated) and only for direct radiative forcing by sulfate particles. The flux calculations are based on a column version of the National Center for Atmospheric Research (NCAR) CCM2 radiation model. The model atmosphere employed three layers of cloud, a high cloud layer, a midlayer cloud, and a low cloud level. The cloud amounts were adjusted to yield a top of atmosphere planetary albedo of 0.3. The lowest cloud layer is located at 800 millibars (mb). The below-cloud calculation assumed that the aerosol layer was completely below this

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change low cloud layer. The above-cloud calculation assumed that the entire aerosol loading was located above the lowest cloud layer. Details of the radiation model are given by Kiehl and Briegleb (1993). Table 3.1 shows the sensitivity of radiative forcing to changes in either aerosol properties or atmospheric conditions. The second column in Table 3.1 presents this sensitivity as a change in absorbed solar flux per percentage change in a property, that is, [ΔSa(x + 0.2x)-Δ Sa(x - 0.2x)] ÷ 0.4x, where ΔSa is the globally averaged forcing from an increase in sulfate aerosol and x denotes a particular aerosol or atmospheric property. Note that the control forcing ΔSa is -0.3 W m-2. The near-linear dependence of the forcing on loading is apparent from the first sensitivity (i.e., -0.3 W m-2 is obtained approximately by multiplying the listed value by 100 percent). The third column ranks the various sensitivities relative to the sensitivity to changes in total loading. For Table 3.1, the aerosol is assumed to reside in the boundary layer and hence below clouds. Table 3.2 shows similar results for conditions where the aerosol is placed above the lowest cloud layer (located at 800 mb). The results are shown graphically in Figures 3.1 and 3.2. A negative sensitivity implies an increase in aerosol forcing (i.e., more negative), because the anthropogenic aerosol forcing is negative (i.e., the presence of aerosols decreases the column-absorbed shortwave flux). A positive sensitivity implies a reduction in aerosol forcing from an increase in the property. For example, an increase in the asymmetry parameter g leads to more solar radiation scattered toward the Earth and hence less back to space, which reduces the anthropogenic aerosol effect. According to Table 3.1, for a sulfate aerosol located below clouds, the most important aerosol property affecting the forcing is the asymmetry parameter. For this parameter, the visible values are more important than the near-infrared values. Second in importance is the single scattering albedo, where the visible values are again of greatest importance. As for atmospheric properties, aerosol forcing is quite sensitive to total cloud cover. Importantly, the ranking is dependent on the vertical location of the aerosol layer. Table 3.2 indicates that the single scattering albedo is of greater importance for a sulfate layer above the lowest cloud base. In this case, however, the forcing is still very sensitive to the asymmetry parameter. Sensitivity to other aerosol properties, such as the width of the size distribution and the chemical composition, is discussed in Kiehl and Briegleb (1993). It is important to note that the direct aerosol forcing is sensitive to total cloud cover. This places a stringent constraint on cloud predictive capabilities of global climate models and demonstrates the importance of developing ICARUS priorities within the priority framework of the entire USGCRP. More to the present point, this analysis reveals some essential features of sensitivity/uncertainty analyses. Even though uncertainties were not

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change TABLE 3.1 Sensitivity Calculations for a Sulfate Aerosol Layer Below Clouds   Sensitivity of Forcing (W m-2 %-1) Rank Relative to Loading Aerosol Property Loading -0.00297 1.00 Single scattering albedo (ωo) -0.00473 1.59 ωo (visible) -0.00360 1.21 ωo (near infrared) -0.00098 0.33 Asymmetry parameter (g) +0.00607 -2.04 g (visible) +0.00484 -1.63 g (near infrared) +0.00106 -0.36 Atmospheric and Surface Properties Surface albedo +0.00048 -0.16 Total cloud cover +0.00236 -0.79 Water vapor amount +0.00051 -0.17 FIGURE 3.1 Sensitivity of aerosol forcing for an aerosol layer below cloud.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change TABLE 3.2 Sensitivity Calculations for an Aerosol Layer Above Lowest Cloud Layer   Sensitivity of Forcing (W m-2 %-1) Rank Relative to Loading Aerosol Property Loading -0.00227 1.00 Single scattering albedo (ω0) -0.00948 4.19 ω0 (visible) (≤ λ ≤) -0.00705 3.11 ω0 (near infrared) (≤ λ) -0.00202 0.89 Asymmetry parameter (g) +0.00457 -2.02 g (visible) +0.00358 -1.58 g (near infrared) +0.00086 -0.38 Atmospheric and Surface Properties Surface albedo +0.00035 -0.15 Total cloud cover +0.00180 -0.79 Water vapor amount +0.00050 -0.22 FIGURE 3.2 Sensitivity of aerosol forcing for an aerosol layer above lowest cloud layer.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change explicit in this analysis, the method clearly requires quantitative descriptions of governing processes (in this case, via Mie theory of scattering). In essence, the method evaluates partial derivatives (i.e., evaluates overall changes caused by "parameter" changes). In contrast to this illustrated case for direct effects, if quantitative, realistic descriptions of processes are unavailable (as is the case for indirect effects), then sensitivity/uncertainty analyses cannot be performed quantitatively. True, some estimates of indirect effects are becoming available, and, as reviewed in Chapter 1, emerging consensus seems to be that indirect effect are "probably comparable" to direct effects (Jones et al., 1994; Pincus and Baker, 1994; Boucher and Lohmann, 1995). From the perspective of sensitivity/uncertainty analysis, however, the reality is that the greatest uncertainty about indirect effects arises not from varying parameters within developing models but from differences among the emerging models themselves. As a result, classical sensitivity analysis is inapplicable or, at least, premature. Therefore, research priorities must be defined more heuristically, as described in the next section. FRAMEWORKS FOR RESEARCH AND FUNDING PRIORITIES From the previous two sections, four central features of a priority framework are apparent: (1) numerical models must be central to the ICARUS program; (2) via these models, sensitivity/uncertainty analyses can define priorities; (3) to improve the quantitative base of these analyses, an immediate ICARUS thrust must be to improve quantitative understanding of indirect effects; (4) in the meantime, until understanding of indirect effects increases substantially, priorities can be defined only on a heuristic basis. For what follows, this heuristic base is our assessment that direct and indirect effects are "probably comparable." In this section, therefore, we present only qualitative analyses of research and funding priorities. We describe the results as priority frameworks to emphasize that complete construction currently seems impossible. To begin to outline these frameworks, consider Figure 3.3, which shows—hypothetically—relative uncertainties for the four identified aspects of the total problem: indirect effects for remote marine and continental locations, and direct effects for organic and inorganic aerosols. Also shown (again, only qualitatively) is a lower value for the uncertainties, identified as an ICARUS Phase 1 goal.1 1   Although we have identified model development as the prime goal of ICARUS, it has not been identified separately in Figure 3.4 (and subsequent figures). The reason follows from our firm conviction that modeling must be an integral component of all aspects of ICARUS research.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.3 Qualitative indications of current radiative forcing uncertainties for indirect effects (separately for marine and continental clouds) and for direct effects (separately for organic and inorganic aerosols) and a qualitative indication of the uncertainty goal (to be defined by USGCRP) for the first phase of ICARUS research. Given the uncertainty estimates shown in Figure 3.3 (presumed to be known), research priorities could be set, for example, as indicated qualitatively in Figure 3.4. Relative differences between corresponding bars in Figures 3.3 and 3.4 reflect, qualitatively, the weighting of the uncertainties of Figure 3.3. Typically the goal of applying these weighting factors is to progress toward solving a specific aspect of the overall problem more rapidly. In reality, the task of "suitably weighting" other factors is nontrivial and is particularly difficult when there are conflicting views on any particular issue. Consequently, Figure 3.4 is presented only to show the method for constructing this research priority framework. Nonetheless, it might be useful to mention the following two factors that we expect would be included in such a weighting, thereby explaining the evident differences between Figures 3.3 and 3.4: Because sulfur emissions especially are expected to change rapidly during the next decade (with substantially greater increases from some rapidly developing economies, such as China's, and substantially smaller releases associated with acid rain regulations in other countries), it might be supposed that, during the next decade, changes in direct radiative effects from inorganic particles (both increases and decreases) will become more

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.4 Qualitative indication of relative ICARUS research priorities for different topics, with the differences from Figure 3.3 resulting from weighting the uncertainties of Figure 3.3 with USGCRP strategic and integrating priorities; here, the weighting has been by assumed amounts. important than those of organic particles. Consequently, for Figure 3.5, we have shown, qualitatively, an enhancement of the priority for research on the direct effects of inorganic particles as opposed to organic particles. However, we must remember that organic particles are particularly poorly understood (Penner, 1995). Hence, the research priorities shown in Figure 3.4 may indeed be reversed as more is learned about organic aerosols. Because major, disruptive social actions to alleviate or anticipate global change are not likely to be undertaken until temperature data show an unambiguous warming trend; because available temperature data are primarily from continental sites; because these data currently show a decrease in the daily temperature range (DTR), caused mainly by nights warming more than days (Karl et al., 1995); because the magnitude of this DTR decrease is inconsistent with climate models that account only for increases in "greenhouse gases"; and because there are suggestions (Hansen et al., 1995) that this DTR decrease is caused by increases in low-level cloud cover (an expected indirect consequence of increasing anthropogenic aerosols)—therefore, for Figure 3.4, we have shown, qualitatively, an enhancement of the priority for research on indirect effects of anthropogenic aerosols on continental as opposed to marine clouds.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.5 Qualitative indication of relative funding priorities (resource allocations) for the indicated broad research topics, with the differences from Figure 3.4 (research priorities) resulting from weighting these priorities with costs to perform the research; here, the weighting has been by assumed amounts. We repeat, however, that the purpose of presenting these details (immediately above) has been only to illustrate how a research priority framework can be constructed by incorporating scientific uncertainties with other priorities, typically dictated by more comprehensive systems [USGCRP, Committee on Environment and Natural Resources (CENR), National Science and Technology Council (NSTC), et al.]. To proceed from research priorities (indicated qualitatively in Figure 3.4) to funding priorities (indicated qualitatively in Figure 3.5), it is of course necessary to estimate costs for specific research efforts within each broad research category (i.e., the four broad categories shown in Figures 3.4 and 3.5). Some of these details are indicated later in this section; first, however, to emphasize the differences between research priorities (Figure 3.4) and funding priorities (Figure 3.5), the latter figure shows a hypothetical result from such a detailed analyses. For the hypothetical example shown in Figure 3.5, we are suggesting (only to illustrate the framework) (1) that costs to perform studies of indirect effects as they are presently done (typically performed by relatively small research groups) would normally be lower than those for studies of direct effects (and lower for continental than for marine studies, based on

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change costs of aircraft flights and shipboard sampling); (2) that costs to study direct effects of organic aerosols would be rather substantial because of the added expense of sampling and chemical analyses (relative to well-developed techniques for inorganic aerosols); and (3) that costs to study direct effects of inorganic aerosols can be high if the decision is made to take advantage of the many benefits of remote sensing from satellites—as well as other weighting factors. AN EXAMPLE OF SENSITIVITY AND UNCERTAINTY ANALYSES In this section, we illustrate the sensitivity/uncertainty analysis method further by presenting both sensitivity and uncertainty analyses for direct radiative effects of sulfate particles. The goal here is to examine the framework more closely and, to this end, consider how only the rightmost bar in Figure 3.5 (i.e., costs to reduce uncertainties about direct effects of inorganic aerosols) might be derived. For this case, in contrast to the case of indirect radiative effects, considerable knowledge is already available. Also, in contrast to the case for indirect effects, there are currently about a half dozen more-or-less comprehensive numerical models that attempt to simulate these direct effects. A partial analysis from one such model has been demonstrated above. There, results from only a sensitivity analysis were shown—for only some meteorological and aerosol property variables, and based on only a one-dimensional radiation code. A more thorough analysis would use a global atmospheric chemistry model to predict sulfate concentrations. Then, estimates could be made for the sensitivities of direct radiative forcing to uncertainties in both intensive and extensive aerosol properties and to some assumptions and approximations of the model. (If a model omits a particular process such as cloud venting, the sensitivity of the resulting radiative calculation to this omission obviously cannot be estimated.) To date (to our knowledge), such an analysis has not been performed, and because it is essential for establishing ICARUS research priorities, it will be identified as a task for immediate initiation in the ICARUS program. In the meantime, to examine the priority framework in more detail here, we illustrate with an uncertainty analysis based on a simple, zero-dimensional model. Table 3.3 gives estimates for uncertainty factors for terms in the following equation for the areal-mean shortwave radiative forcing by sulfate and biomass burning aerosol <ΔFR>:

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change where FT is the incident solar flux (solar constant times the cosine of the zenith angle (watts per square meter), T is the fraction of the incident light transmitted by the atmosphere above the aerosol, A is the area of the geographical region over which the flux is averaged (square meters), AC is the fractional cloud cover, RS is the albedo of the underlying surface, b is the upward fraction of the radiation scattered by the aerosol, is the scattering efficiency of fine-particle sulfate at a reference low relative humidity (square meters per gram), f(RH) accounts for increase in scattering with increasing relative humidity, is the source strength of SO2 (grams per second), is the fraction of emitted SO2 that yields sulfate aerosol, and is the sulfate lifetime in the atmosphere (seconds). Figure 3.6 graphically shows the uncertainties estimated in Table 3.3A (for sulfate aerosols), along with a hypothetical ''ICARUS Phase 1 Goal" for an acceptable uncertainty level. At this point, it is quite understandable that disagreements would exist about the central values and ranges assigned to the quantities listed in Table 3.3 and plotted in Figure 3.6. For example, the overall uncertainty for the areal mean shortwave radiative forcing attributable to sulfate aerosol [ΔFR] [estimated by the Intergovernmental Panel on Climate Change (IPCC, 1995a)] is 2.2, as opposed to the value 1.89 given in Table 3.3. There can be disagreements, also, about choices of entries in such tables (e.g., inclusion or omission of assumptions about atmospheric processes such as cloud venting, wet and dry deposition). More generally, there can be disagreements about the adequacies of the model(s) used to generate these uncertainty estimates.2 Also, there is no doubt that defining an acceptable uncertainty will be difficult. Setting these difficulties aside temporarily, however, we now suggest how to proceed from Figure 3.6. 2    We want to reemphasize the desirability that model diversity be maintained—and even enhanced. Given the approximations and knowledge deficiencies in so many components of existing models, this diversity is one of the few sources of strengths in existing programs. As a consequence of this diversity (and these deficiencies), resulting model predictions can differ dramatically, thereby providing a measure of appropriate confidence to be placed in predictions from any one model. Stated differently, at present there is little that would cause us greater concern about estimates of climate cooling by aerosols than community acceptance of a "standard model." Perhaps it will be reasonable to accept such a standard a decade or so from now, but we are certain that such a time has neither arrived nor is fast upon us.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change TABLE 3.3 Factors Contributing to Estimates of the Direct Forcing by Anthropogenic Sulfate (A) and Biomass Burning (B) Airborne Particles, Estimated Ranges, and Resulting Uncertainty Factors (for estimates of changes in reflected solar radiation) Quantity Central Value Estimated Range Uncertainty Factor, ufi A. Anthropogenic Sulfate Aerosol mass scattering efficiency (m2 g-1) 5 3.6-7 1.400 Average atmospheric lifetime (days) 6 4-8 1.375 Aerosol hemispheric backscatter fraction 0.15 0.12-0.22 1.267 Fraction of SO2 oxidized to sulfate aerosol 0.5 0.4-0.6 1.200 Square of atmospheric transmittance above aerosol layer 0.714 0.594-0.86 1.204 Fractional increase in scattering efficiency from hygroscopicity 1.7 1.4-2.0 1.176 Source strength of anthropogenic S (Tg per year) 71 62-81 1.141 Fraction of Earth not covered by clouds 0.39 0.35-0.44 1.128 Square of surface co-albedo 0.72 0.65-0.80 1.111 Total uncertainty factor = exp {[Σ (In ufi)2]1/2 = 1.89a Result: If central value is -0.6 W m-2, then the range is from -0.3 to -1.1 W m-2

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change B. Biomass Burning Aerosols Emission factor (g/kg C in fuel) 32 18-65 1.750 Aerosol mass scattering efficiency (m2 g-1) 5 3.6-7 1.400 Average atmospheric lifetime of smoke (days) 5 3.6-7 1.400 Aerosol hemispheric backscatter fraction 0.15 0.11-0.20 1.333 Aerosol mass absorption efficiency (m2 g-1) 0.7 0.5-0.9 1.286 Square of atmospheric transmittance above aerosol layer 0.714 0.594-0.86 1.204 Amount of biomass burned (Tg of C per year) 3800 3200-4500 1.197 Fractional increase in scattering efficiency from hygroscopicity 1.7 1.4-2.0 1.176 Fraction of Earth not covered by clouds 0.39 0.35-0.44 1.128 Square of surface co-albedo 0.72 0.65-0.80 1.111 Total uncertainty factor = exp {[ ∑ (In ufi)2]1/2 = 2.45 a Result: If central value is -0.8 W m-2, then the range is from -0.3 to -2.0 W m-2 a This analysis assumes that these factors are independent, whereas several are not. For example, the aerosol hemispheric backscatter fraction and aerosol mass scattering efficiency both depend on the assumed refractive index through Mie theory. Also, the fractional increase in scattering efficiency from hygroscopicity is related to the atmospheric transmittance because both depend on relative humidity. Also, more complete models find substantial aerosol scattering even in areas covered by clouds and would therefore include consideration of the codependence of aerosol scattering in these regions on relative humidity as well. SOURCE: Penner et al. (1994).

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.6 Plot of the uncertainties listed in Table 3.3A for sulfate aerosols, with a qualitative indication of the level to which the uncertainty could be set as a goal for the first phase of ICARUS research. From Figure 3.6 (showing uncertainties), Figure 3.7 (showing research priorities) can be derived by applying various weighting factors such as those mentioned earlier in this section. In this case, we are suggesting that higher priorities would be given to atmospheric lifetimes (because there is serious doubt that the value used in Table 3.3 is accurate) and to SO2 oxidation [because this oxidation is critically important also for estimating indirect effects, i.e., defining what fraction of the emitted SO2 becomes new cloud condensation nuclei (CCN)]. For some other research topics identified in Figure 3.6, a relative decrease in research priority has been suggested, because although the associated uncertainty may be roughly as indicated in Figure 3.5, the necessary data (e.g., on backscattered radiation, cloud cover, and surface albedo) are available, and what is needed is "only" additional analysis of existing data.3 These comments, however, are meant to be merely qualitative. In order to define funding priorities (i.e., resource allocations), cost estimates to perform each research task are applied to the research priorities shown in Figure 3.7. The result is shown qualitatively in Figure 3.8, which completes this qualitative description of how only the rightmost bar in Figure 3.4 is derived. Consider how one of the eight bars shown in Figure 3.8 (funding priorities) would be derived, which in turn provides some details for only one of the four bars in Figure 3.4. In particular, consider details of how to reduce current 3    Whereas data on backscattered radiation are available, few data are available on the backscattered fraction.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.7 Qualitative indication of research priorities for direct radiative effects of sulfate aerosols, derived from Figure 3.6 (uncertainties) by weighting with such factors as mentioned in the text. FIGURE 3.8 Qualitative indication of the relative costs to reduce the uncertainties shown in Figure 3.6, consistent with the research priorities shown in Figure 3.7, accounting for the cost of performing the research (e.g., a prorated portion of satellite costs to measure backscattered radiation).

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.9 Qualitative indication of the relative contributions from different processes to current uncertainty in the atmospheric lifetime of aerosol sulfate, with a qualitative indication of the level to which the uncertainty could be set as a goal for the first phase of ICARUS research. uncertainties in the atmospheric lifetime of sulfate aerosols (second bar from the left in Figure 3.8). Figure 3.9 shows (only qualitatively) the relative contributions from different processes to current uncertainties in this lifetime. Given knowledge about relative uncertainties as shown qualitatively in Figure 3.9, and given still another set of weighting factors (e.g., increase in priority for precipitation efficiencies because they are also a source of major uncertainty in estimates of precipitation within climate models), research priorities can be deduced. Then to these research priorities can be applied the costs of performing the research, leading to funding priorities, as indicated qualitatively in Figure 3.10, which completes this indication of how only one bar in Figure 3.8 could be derived (which in turn suggests the derivation of only one bar in Figure 3.4). Summary In this chapter we have outlined a formal framework, based on sensitivity/uncertainty analysis of aerosol-climate models, to select research

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change FIGURE 3.10 Qualitative indication of funding priorities to reduce the uncertainties shown in Figure 3.9. priorities. Even in view of this formal procedure, it is clear that, in many respects, it is premature to apply such an analysis to the aerosol-climate problem. Yet, on a qualitative basis it is possible to identify a number of the key research areas that will need to be addressed to reduce uncertainties by using this procedure. Already there are high-priority tasks (Chapter 2) that (we are certain) any future formal sensitivity/uncertainty analysis will demonstrate to be critically important. We therefore conclude that "to get off the ground," ICARUS must plan on some initial "seat-of-the-pants" flying, without the aid of formal sensitivity/uncertainty analysis. The extent to which the ICARUS program incorporates a formal sensitivity/uncertainty analysis approach in setting priorities will have to evolve in time. Chapter 4 presents a possible structure for the ICARUS program and the recommended research.

OCR for page 89
A Plan for a Research Program on Aerosol Radiative Forcing and Climate Change This page in the original is blank.