There is a very informative presentation regarding IPCC model skill in a powerpoint presentation by Graeme Stephens of Colorado State University. He is lead scientist on the NASA CloudSat Mission and has new data from this study with which to compare with the ability (actually lack thereof) to accurately simulate the climate system [thanks to Marcel Crok for alerting us to this important new information!].
He includes a “disclaimer” at the beginning of his talk which reads
“Disclaimer – while the validity of some of the findings of the IPCC must be seriously questioned on strict scientific grounds, I do think they represent the most reasonable expectation given our qualitative state of understanding of the moist processes of the climate system.”
Despite the caveats he wrote in the disclaimer, his serious questions include the following (with further details in the slides)
On slide 4:
“Analysis of climate models together with constraints from observations enables an assessed likely range to be given for climate sensitivity for the first time ……….. It is likely to be in the range 2 to 4.5°C with a best estimate of about 3°C (IPCC, 2007).”
WHOA – not so fast:
The climate sensitivity is curiously inversely correlated with aerosol (direct and indirect) forcing (Kiehl, 2007). That is the climate sensitivity is conditioned to fall in a given ‘comfort’ range. Clearly, better quantification of direct and indirect forcing will provide an important constraint on model projections.
On slide 9:
Model low, warm cloud optical and radiative properties are significantly different (biased) compared to those observed – two factors contribute to this extreme (bright) bias ‐ the LWP [liquid water path] is one, particle size is another.
On slide 12:
1. Aerosol and aerosol‐cloud effects are a huge lever ‘constraining’ the climate sensitivity to a ‘range of comfort’.
2. Observational inferences on indirect radiative forcing do not support the large values of forcings being applied in models. I would recommend model assessments be done with/without IRF [indirect radiative forcing]
3. Models contain grave biases in low cloud radiative properties that bring into question the fidelity of feedbacks in models.
4. The presence of drizzle in low clouds is ubiquitous and significant enough to influence the radiative properties of these clouds and must play some role in any feedbacks.
On slides 24-25:
Models produce rain 2‐4 times too frequently regardless of resolution…..…. and 2‐3 times too light.
On slide 26:
While it is expected that ‘ heavy precipitation events will continue to become more frequent’, our predictive tools (either climate or NWP models) contain major biases that are symptomatic of unrealistic rain physics.
While I believe the changes that are likely to occur are primarily driven by changes in the large scale atmospheric flows, we have to conclude our models have little or no ability to make credible projections about the changing character of rain and cannot conclusively test this hypothesis.
This model bias isn’t merely solved by higher resolution of models – to the contrary, there are fundamental flaws in the way rain is triggered in models on all scales. The consequence to other aspects of the Earth system model is profound.
This presentation by Graeme Stephens highlights the inability of the IPCC models to make skillful predictions of climate decades into the future on the global scale, much less regional spatial scales.
Regional assessments based on the IPCC model results in such reports as the CCSP series (as well as the set of talks moderated by Tom Karl, Director of the National Climate Data Center and current President of the American Meteorological Society (e.g. see) are flawed, scientifically unsupported reports and are misleading policymakers.