Distributions of decadal means of temperature and precipitation change under global warming

There remains uncertainty in the projected climate change over the 21st century, in part because of the range of responses forced by rising greenhouse gas concentrations among global climate models. This paper applies a method of estimating distributions and probability density functions (PDFs) for forced change, based on the pattern scaling technique and previously used for Australia, to generate changes in temperature and precipitation at locations over the globe, from simulations of 23 CMIP3 models. Changes for 2030 and 2100, under the A1B scenario for concentrations, for both seasonal and annual cases are presented. The PDFs for temperature have a standard deviation that averages 31% of the mean change, and they tend to be positively skewed. The standard deviation for precipitation averages 15% of the base climate mean, leading to five and 95 percentile estimates that are of opposite sign for most of the globe. A further source of uncertainty of change for a particular period of time, such as a decadal average, is the unforced or internal variability of climate. A joint probability distribution approach is used to produce PDFs for decadal means by adding in an estimate of internal variability. In the decade centered on 2030, this broadens the PDFs substantially. The results are related to time series of observations and projections over 1900-2100 for the agricultural regions of Iowa and the Murray-Darling Basin. For most land areas, warming becomes clearly discernable, allowing for both uncertainties, in the next few decades. Data files of the key results are provided.