by Judith Curry
The issue of separating natural from anthropogenically forced variability, particularly in context of the attribution of 20th century climate change, has been a topic of several previous threads at Climate Etc. The issue of natural vs anthropogenically forced climate variability/change has been a key issue of contention between the climate establishment and skeptics. There are some encouraging signs that the climate establishment is maturing in their consideration of this issue.
Distinguishing the Roles of Natural and Anthropogenically Forced Decadal Climate Variability: Implications for Prediction
Amy Solomon, Lisa Goddard, Arun Kumar, James Carton, Clara Deser, Ichiro Fukumori, Arthur M. Greene, Gabriele Hegerl, Ben Kirtman, Yochanan Kushnir, Matthew Newman, Doug Smith, Dan Vimont, Tom Delworth, Gerald A. Meehl, and Timothy Stockdale
Abstract. Given that over the course of the next 10–30 years the magnitude of natural decadal variations may rival that of anthropogenically forced climate change on regional scales, it is envisioned that initialized decadal predictions will provide important information for climate-related management and adaptation decisions. Such predictions are presently one of the grand challenges for the climate community. This requires identifying those physical phenomena—and their model equivalents—that may provide additional predictability on decadal time scales, including an assessment of the physical processes through which anthropogenic forcing may interact with or project upon natural variability. Such a physical framework is necessary to provide a consistent assessment (and insight into potential improvement) of the decadal prediction experiments planned to be assessed as part of the IPCC’s Fifth Assessment Report.
Citation: Solomon, Amy, and Coauthors, 2011: Distinguishing the Roles of Natural and Anthropogenically Forced Decadal Climate Variability. Bull. Amer. Meteor. Soc., 92, 141–156. doi: 10.1175/2010BAMS2962.1
Link to the complete article [here].
JC comment: The first sentence of the abstract really caught my attention: Given that over the course of the next 10–30 years the magnitude of natural decadal variations may rival that of anthropogenically forced climate change on regional scales. . . I don’t recall the climate establishment “giving” this one before. The implications of this is that the warming between 1970 or 1980 to 2000 should be operating under the same givens also.
From the Introduction:
As the science of decadal prediction is in its infancy, one would like to assess and understand the following:
- the expectations for added regional climate in-formation and skill achievable from initialized decadal predictions;
- what physical processes or modes of variability are important for the decadal predictability and prediction problem, and whether their relevance may evolve and change with time;
- what elements of the observing system are important for initializing and verifying decadal predictions; and
- in terms of attribution, to what extent are regional changes in the current climate due to natural climate variations and thus transitory, and to what extent are they due to anthropogenic forcing and thus likely to continue.
The purpose of this paper is to describe existing methodologies to separate decadal natural variability from anthropogenically forced variability, the degree to which those efforts have succeeded, and the ways in which the methods are limited or challenged by existing data. Note that the separation of decadal natural variability from anthropogenically forced variability goes beyond what has already been accomplished in previous studies that focused primarily on the detection of a long-term anthropogenic signal (Hegerl et al. 2007b) because on decadal time scales anthropogenic effects may be nonmonotonic, regionally dependent, and/or convolved with natural variability.
JC comment: the detection of the long-term signal from anthropogenically forcing was detected in the AR4 basically for the period 1970 or 1980 to 2000, without account for this: because on decadal time scales anthropogenic effects may be nonmonotonic, regionally dependent, and/or convolved with natural variability.
Observational uncertainties
Verification of the forced component of twentieth- century climate trends simulated in model experiments depends on the existence of accurate estimates of these trends in observations. Given the limited sampling in both space and time of the observations and proxy records, these verifications must be handled carefully. In particular, knowledge of the spatial patterns and magnitudes of climate trends over the oceans is hampered by the uneven and changing distribution of commercial shipping routes and other observational inputs as well as different approaches to merging analyses of the observations (Rayner et al. 2011).
An example of the impact of observational uncertainties on the interpre-tation of twentieth-century SST trends is shown in Fig. 7 based on an uninter- polated dataset [version 2 of the Hadley Centre SST dataset (HadSST2); Rayner et al. 2006] and two optimally interpolated reconstructions [the Hadley Centre Sea Ice and SST dataset (HadISST; Rayner et al. 2003) and version three of the National Oceanic and Atmospheric Administration’s (NOAA’s) extended reconstructed SST (ERSSTv3; Smith et al. 2008)]. Although trends from the three datasets share many features in common, such as a strengthening of the equatorial Pacific zonal temperature gradient (Karnauskas et al. 2009), there are also differences. Most notably, the eastern equatorial Pacific shows cooling in HadISST and warming in HadSST2 and ERSSTv3 (see also Vecchi et al. 2008). However, independently measured but related variables, such as nighttime marine air temperatures, provide some evidence that the eastern Pacific trends represented in the HadSST2 and ERSSTv3 datasets may be the more realistic ones (Deser et al. 2010b). These observational sampling issues underscore the challenge of providing a robust target for model validation of twentieth-century surface marine climate trends and perhaps the need to consider a suite of complementary measures for poorly sampled variables and/or regions.
A limitation of the instrumental record is that it spans at most a few realizations of decadal variability. Paleoclimate records—derived from tree rings, corals, lake sediments, or other “proxies”—have been used to extend this record to hundreds of years or more and are generally believed to be free of anthro- pogenic influence prior to the industrial age (Brook 2009; Jansen et al. 2007), thus constituting a poten-tial means of model verification.
JC comment: with all these uncertainties in the observations of ocean temperature, “unequivocal” and “very likely” in the AR4 seem overconfident
Modeling uncertainties.
The spatial structure and dominant time scales of natural variations differ across models (see discussion of Fig. 5). Additionally, coupled climate models produce a range of responses, in space and time, to anthropogenic radiative forcing (Fig. 8). Such differences in model estimates of internal variability and response to external forcing limit our understanding for the potential of the decadal climate predictions.
As an example, the historical changes and future response of the tropical Pacific mean state have been subjects of debate. Different proposed mechanisms disagree on the expected sign of change in the zonal SST gradient in the tropical Pacific in response to anthropogenic forcing. The observational record does little to clarify the situation, as trends in different observed SST records differ in even their sign (see Fig. 7). Models that simulate the largest El Niño–like response have the least realistic simulations of ENSO variability, while models with the most realistic simulations of ENSO project little change in the Pacific zonal SST gradient (Collins 2005). These differences in tropical Pacific interannual variability and change have implications for Pacific decadal variability through their impact on large-scale changes in the atmospheric circulation (e.g., Alexander et al. 2002; Vimont 2005).
Conclusion
The main conclusion drawn from the body of work reviewed in this paper is that distinguishing between natural and externally forced variations is a difficult problem that is nevertheless key to any assessment of decadal predictability and decadal prediction skill. Note that all the techniques are limited by some assumption intrinsic to their analysis, such as the spatial characteristics of the anthropo- genic signal, independence of noise from signal, or statistical stationarity.
JC summary: The authors of this paper are members of the climate establishment, in terms of being involved with the WCRP CLIVAR Programme and also the IPCC. This paper arguably provides more fodder for skepticism of the AR4 conclusions than anything that I have seen from the climate establishment (the authors may not realize this). The issues surrounding natural internal decadal scale variability are a huge challenge for separating out natural from forced climate change. The same issues and challenges raised for future projections remain also for the warming in the last few decades of the 20th century. Sorting this out is the key challenge. No more unequivocals or very likelys in the AR5, please.