I recently read an interesting discussion on ‘uncertainty in climate modeling’ by Tebaldi, Schmidt, Murphy and Smith in the Bulletin of Atomic Scientists, But the authors ignore some of the central problems that plague climate models that try to predict the development of future climate. I am referring here to three major issues:
1) Uncertainties of the scenarios that determine the emission of greenhouse gases, principally economic growth, which is closely tied to the use of energy. Economic growth in turn, is a function of population and economic development and may be roughly approximated by GDP growth. The IPCC lists a wide spectrum of what they consider to be plausible scenarios and calculates global temperatures for the year 2100 with an uncertainty spread of an order of magnitude [IPCC 2007, Fig. SPM.5, p.14].
2) Structural uncertainties. I include here uncertainties in climate forcing, both anthropogenic and natural; in climate feedbacks; and in the hundred or so parameters that go into constructing a model, mainly concerned with clouds. While the IPCC uses fairly precise numbers for the various greenhouse gases, it omits the most important one, namely water vapor. Its contribution is encompassed within the models in terms of a positive feedback that amplifies the forcing of anthropogenic greenhouse gases by a factor of about 3.
The uncertainties listed for aerosols are quite large, particularly for the indirect effects of aerosols in providing condensation centers for cloud formation. [IPCC-AR4 2007, Fig. TS-5, p.32]. In addition, aerosols come in different flavors, ranging from reflecting sulfates to absorbing soot particles. Unlike well-mixed GH gases, like CO2, aerosols show particular geographic and temporal distributions, which also affect climate projections significantly. Given the realistic range of aerosol compositions used here, it is not possible for global models to correctly calculate the cloud albedo effect if composition is ignored [Roesler and Penner 2010].
James Hansen, a leading climate modeler, called attention to our inadequate knowledge of radiative forcing from aerosols when he stated, “the forcings that drive long-term climate change are not known with an accuracy sufficient to define future climate change” [Hansen 1998].
Parameterization is a vexing issue for climate modelers. James Murphy [Nature 2004] lists some 100 or more parameters that must be chosen, using the modelers.“best judgment.” Varying just six of these parameters related to clouds can change the climate sensitivity from 1.5 up to 11.5 degC [Stainforth et al 2005].
Even more important, the feedbacks (from WV and from clouds) may actually be negative rather than positive (as assumed in all climate models). This possibility follows from the analyses of satellite data [by Lindzen and Choi 2010 and by Spencer and Braswell 2010].
3) Chaotic Uncertainty. It is well understood that climate is a chaotic object and climate models reflect that property. The outcome of a particular model run (“simulation”) depends sensitively on the initial conditions; even minute changes can lead to greatly differing outcomes. For example, the five runs of a Japanese MRI model show temperature trends that differ by almost a factor of 10, an order of magnitude. (If more runs had been performed, the spread would have been even greater.) One can show [Singer and Monckton 2011] that taking the mean of an ensemble of more than 10 runs leads to an asymptotic value for the trend. However, most modelers face constraints on time and money and are not able to carry out so many runs. For example, of the 22 models in the IPCC compilation of “20 CEN” [an IPCC term for a group of climate models] there are 5 single run models, 5 two-run models, and only 7 models with four or more runs.
Clearly, models cannot be used to predict future global temperatures reliably. (Note that variability and uncertainty of models is even greater for regional temperatures and for quantities other than temperature, such as precipitation.) The chief value of models, I believe, derives from their use to test sensitivity of outcome to variations in specific forcings or input parameters.
Hansen, J.E., et al. 1998. Climate forcings in the industrial era. Proc. Natl. Acad. Sci. USA 95: 12753-12758.
IPCC-AR4 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press.
Murphy, J.M., et al. 2004. Quantification of modeling uncertainties in a large ensemble of climate change simulations. Nature 429: 768-772.
Roesler, E.L. and J.E. Penner. 2010. Can global models ignore the chemical composition of aerosols? GRL 37: doi:10.1029/2010GL044282
Singer, S.F. and C.W. Monckton. 2011. Chaotic behavior of climate models. (Submitted)
Stainforth, D.A., et al. 2005. Uncertainty in predictions of the climate response to rising levels of greenhouse gases. Nature 433: 403-406.
Items of notes and interest from the web.
Pursuant to Title 17 U.S.C. 107, other copyrighted work is provided for educational purposes, research, critical comment, or debate without profit or payment. If you wish to use copyrighted material from this site for your own purposes beyond the 'fair use' exception, you must obtain permission from the copyright owner. Views are those of authors and not necessarily those of Canada Free Press. Content is Copyright 1997-2015 the individual authors. Site Copyright 1997-2015 Canada Free Press.Com Privacy Statement
Close Window ExpressionEngine v2.10.1 - Build Date: 20150501 - © Copyright Notice 2003 - 2017 EllisLab, Inc. License Number: 1946-0972-4464-7746 - [email protected] Script executed in 0.0501 seconds - 19 SQL queries used