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 7
APPROACHES FOR ESTIMATING CLIMATE SENSITIVITY PETER STONE Peter Stone, from the Massachusetts Institute of Technology (MIT), provided an overview of how climate sensitivity estimates have changed over time (see Table 1). After a century of research, how has the climate community’s ability to gauge climate sensitivity advanced? Some progress has been made in reducing uncertainty, particularly with respect to distinguishing the responses to different external influences on the climate system. Regardless, many of the sources of uncertainty remain. TABLE 1 Sensitivity Estimate (°C) Source Date 5 Arrhenius 1896 2.3 Manabe and Wetherald 1967 3.0 (+/−1.5) Charney Committee report1 1979 1.5-4.5 IPCC2 1990 1.7-4.2 IPCC2 2001 1NRC, 1979. 2The latest IPCC estimate is of an effective sensitivity, while all the earlier ones are equilibrium sensitivities. In the last five years, we have become aware that climate sensitivity is not a constant, but evolves with time as the climate changes. For example, in a global warming scenario in which sea ice retreats, the ice albedo temperature feedback decreases as the ice retreats, and the temperature sensitivity decreases correspondingly. Thus, the climate sensitivity that one should use in making projections has to be matched to the time scale and scenario for which the projections apply. Statistical approaches are needed to better characterize the range of sensitivity values given in the IPCC/TAR. An important emerging approach is to use real observations to constrain sensitivity estimates, in this case Seff. However, we cannot worry about sensitivity alone, but must also look at the poorly quantified factors such as aerosol radiative forcing and ocean heat uptake, can strongly affect estimates of sensitivity. Ocean heat uptake is not well constrained by observations. We need better ocean observations, but this is difficult because of the long time scales involved. Underlying physical constraints are thus needed as well. Discussion Gregory: There was logic in the expert judgment choices to exclude high sensitivity (Seq) values, since we’ve seen that the climate system doesn’t exhibit wide swings that would be characteristic of high sensitivity values. This is because the amplitude of natural variability depends on the same feedback processes that determine the value of Seq itself. Mann: In some contexts, ocean mixing rates can be constrained by using tracer compounds (such as chlorofluorocarbons), as well as ocean heat content measurements from the past several decades. Mahlman: The finding that sensitivity evolves over time may be a good reason to use Str for careful, model intercomparisions. However, it can also prove useful to use Seff for evaluation of various model responses to specific radiative forcing scenarios over the “near term” (e.g., 10-50 years from the present).
OCR for page 8
JONATHAN GREGORY Jonathan Gregory, from the Hadley Centre for Climate Prediction and Research, explained that surface albedo, water vapor, and cloud feedbacks are among the factors that affect climate sensitivity. It is useful to know how climate will respond in time to forcings that are themselves time dependent. Because of its large heat capacity and thermal inertia, the ocean plays a major role in determining the timing of the climate system’s response to exogenous forcing. Research into the climate system’s transient response therefore almost always involves an atmospheric model coupled to a model of ocean heat uptake. Gregory described a very simple approach that he and his colleagues used to estimate a probability distribution for effective climate sensitivity of the real-world climate system (Gregory et al., 2002). It is based on the simple theory that during time-dependent climate change, the imbalance between imposed radiative forcing and radiative response is absorbed by the ocean, which contains most of the heat capacity of the system. Hence, F=Q-λΔT where F is the heat flux into the ocean, Q is the radiative forcing, and ΔT is the temperature change. One can solve the equation for the climate response parameter (λ), by estimating the other parameters as follows: Ocean heat uptake (F) was estimated from the 5 year running means of observed interior ocean temperature changes (Levitus et al., 2000). This approach was used in lieu of relying on model estimates of ocean heat uptake, which usually involve some very uncertain assumptions. However, a model still has to be used to estimate F in the late nineteenth century, since there are no observations for that period. It is not possible to obtain a true steady-state global average temperature, so instead, the temperature change (ΔT) is estimated as a difference in global average temperature between the current period and an earlier period, with the two periods as widely separated as possible, to maximize the climate change signal. Radiative forcing (Q) for greenhouse gases, sulfate aerosols, solar irradiance, and volcanic aerosols is estimated by a variety of methods, since there are no direct observations of past changes in radiative forcing. With these inputs, Gregory and colleagues calculated a PDF for sensitivity (Figure 3, pg. 29) and determined that the lower bound (5th percentile) is 1.6 K. This method of constraining the lower bound for sensitivity is objective and largely independent of climate models. Sensitivity could be further constrained if we could reduce uncertainties in the radiative forcing. Gregory reviewed other approaches that have been used to develop estimates of climate sensitivity, including: equilibrium “slab” experiments with atmospheric models coupled to mixed-layer ocean models; time-dependent atmosphere-ocean climate model experiments; simple climate models constrained by the twentieth century temperature record, which are used to develop PDF of climate sensitivity and forcing; and paleoclimate records that are used as an observational constraint. Reasons for differences among these various estimates include the fact that climate models have different feedbacks and sea surface temperature changes. Also, different studies employ different assumptions about natural and anthropogenic forcings, ocean heat uptake, and the “initial” (pre-industrial) state of the climate system. Discussion Mahlman: If we had a perfect climate observing system, including aerosols, could you use these simple model approaches to better constrain sensitivity? How long would it take? Schlesinger: It may take almost a century of observations to reduce the uncertainty, although you would learn faster if sensitivity is large (since the signal is larger). Mahlman: It wouldn’t take another century if the measured climate variables and the climate forcings were well known. In such a case, all that would be left is global natural variability, which averages out over a few decades. Stone: There are opportunities to reduce uncertainties in aerosol forcing through new observations. Mahlman: However, we don’t even know how to observe the indirect effects of aerosols. Ramaswamy: We also need to learn more about how to quantify direct radiative forcing from black carbon aerosols.
OCR for page 9
NATALIA ANDRONOVA Natalia Andronova, from the University of Illinois, identified the different tools that are used for studies of climate sensitivity, including both comprehensive and simple climate models and simple statistical models such as a first-order autoregression model (e.g., Miles and Gildersleeves, 1977; Tol and Vellinga, 1998). She recommended that statistical models not be used for climate sensitivity estimation because their equilibration time to doubling CO2 is two orders of magnitude smaller than that of simple climate models and coupled of atmosphere-ocean general circulation models. Some estimates of climate sensitivity that have been published in recent years are shown in Table 2. Two general approaches that can be used for climate sensitivity estimation are “mapping” and “optimal estimation.” One can use mapping with constrained input, wherein the observed temperature record and a defined PDF for radiative forcing and ocean heat uptake are used as input to calculate a PDF for climate sensitivity (e.g., Gregory et al., 2002). Alternately, one can run the model with constrained output, wherein subjective PDFs for climate sensitivity, radiative forcing, and ocean heat uptake are used as model input, and the output (constrained by the observed data record) updates the input PDFs for climate sensitivity and model parameters (e.g., Forest et al., 2002; Harvey and Kaufman, 2002). TABLE 2 Sensitivity Estimate (°C) Source 2.3±0.9 Hoffert and Covey (1992) 1.0-9.3 Andronova and Schlesinger (2001)2 1.0-5.0 Harvey and Kaufman (2002)3 1.4-7.7 Forest et al. (2002)2 1.6-∞, (median 6.1) Gregory et al. (2002)2 1The Hoffert and Covey study presented the mean and standard deviation for climate sensitivity derived from paleoclimate data. 2The Andronova and Schlesinger, Forest et al., and Gregory et al. studies present 90 percent confidence intervals. 3The Harvey and Kaufman study used prescribed climate sensitivity values ranging from 1.0-5.0°C in their model. There are two methods for the optimal estimation: the single-value optimal estimation (see, e.g., Andronova and Schlesinger, 2000) and multiple-value optimal estimation techniques (Andronova and Schlesinger, 2001). The single-value optimal estimation traces the value of the climate sensitivity in a domain of the model parameters that gives the best fit between the observed and simulated temperature changes. The multiple-value optimal estimation technique is based on multiple realizations of the single-value optimal estimation for the set of “surrogate observations.” The surrogate observations are the sum of the simulated temperature departure and “surrogate residuals.” The surrogate residuals are obtained by randomly mixing (bootstapping) the single observed residuals—the difference between the observed and simulated temperature departures. Running the model for many surrogate observations yields a PDF for climate sensitivity, as well as for other parameters. In general, the resulting PDFs for climate sensitivity obtained from these different techniques should coincide. The main difference between existing estimations obtained by these two methods comes from description of the “climate noise” and uncertainties in forcing models. Discussion Broccoli: Perhaps they could get more constraints by further exploiting observational data (for instance, going beyond just global and hemispheric mean temperatures and looking at zonal mean temperature differences). Prather: What about uncertainty related to ozone forcing? Andronova: Some forcing models include tropospheric ozone radiative forcing, while others do not. When this forcing is included, its hemispheric distribution is similar to the aerosol radiative forcing distribution.
OCR for page 10
Stone: Our group at MIT is currently working on including other important forcings, such as solar irradiance and land-use changes, in simple climate models. Mahlman: The low inferred paleo-estimation of climate sensitivity (presumably Seq) needs further explaining. The original Milankovich orbital climate forcing that preceded the last ice age was originally -1 Wm−2. Roughly 10,000 years later, the inferred effective total forcing was roughly -4 to 6 Wm−2. The total positive feedbacks that led to this remarkably large climate cooling came mainly from ice albedo, water vapor, and CO2 feedbacks. However, these inferred large positive feedbacks do not necessarily imply higher-than-expected positive feedbacks in the human-caused climate warming problem that we currently face.
Representative terms from entire chapter: