showing that less than 66% of the models agree about whether it gets wetter or drier in the 2090's (wet/dry average over a three month season!) clearly estabilishes a level of uncertainty consistent with my claim;
the IPCC does not hide from this uncertainty (SPM, Fig 7)
not sure what you mean by "derive" an uncertainty. (?)
Perhaps calculate would be more appropriate instead of derive?
Presenting different scenarios using various models is clearly not an expression of uncertainty in statistical terminology. If the models are robust, then only one would do.
How do you define uncertainty? When I see the word it can have one of two definitions depending on the context.
- The condition of being uncertain; doubt, vagueness, ambiguity etc.
- The estimated amount or percentage by which an observed or calculated value may differ from the true value.
Definition #1 is not difficult to understand. When uncertainty is used in that context, since there is no quantified amount or percentage, whatever the subject matter is should not be assumed reliable.
Definition #2 can be quite complex, however is necessary to achieve any degree of confidence in the measurement reported or predicted. In a nutshell, if the relevant errors are not accounted for and included in the uncertainty calculation, the result is still ambiguous and not to be trusted. Uncertainty is a well established statistical methodology in science as I’d assume you know.
Therefore, until the significant errors and unknowns can be qualitatively and quantitatively determined, it is not possible to assign uncertainty in any meaningful way.
When for instance climate models were way off on precipitation this year as reported in various publications, the source of the error must be accounted for.
Another is Arctic Ice melt. Much ado was about how in the last few years it was much worse than the models predicted. Now new data may be indicating the accelerated melt was due to unusual winds and ocean circulation, not an increase in air temps. The question is then, why were the models wrong?
I contend the reason the models are not matching observations is the low level of scientific understanding of several key factors in climate; clouds (aerosols), precipitation, solar, UHI, land use change etc. and possibly that climate is too chaotic to predict. Until such time as those factors can be accurately and precisely determined, no amount of parameterizations (fudging) will improve the skillfulness of climate models at predicting future climate.
Regional "projections" are another problem for climate models.
Are climate model outputs then the result of well understood climate forcings, feedbacks, established physics etc., or more or less the numerical expression of the programmer’s assumptions?