Bioclimatic Evaluation of CMIP Historical Climate Simulations
Köppen bioclimatic classification derives generic vegetation types from various characteristics of the interactive annual-cycles of continental regional temperature (T) and precipitation (P). Thus, by comparing a climate model’s Köppen vegetation types with those implied by observations of T and P, it is easy to pinpoint biases in the model’s simulations of regional continental climate that have significant biological consequences.
We conducted such a Köppen evaluation of simulations of 1980-1999 historical climate by 27 current-vintage models participating in phase 5 of the Coupled Model Intercomparison Project (CMIP5), and by 19 CMIP3 predecessor models. In addition to investigating qualitative model discrepancies in regional vegetation type, we also developed quantitative measures of model bioclimatic performance relative to observationally based reference vegetation types. These measures included 1) a vegetation “hits” metric h(v) that recorded the percentage of matches of model and reference vegetation type v at each grid box, and 2) a vegetation area metric a(v) that quantified model agreement in global percentage area of each vegetation type v.
The collective bioclimatic performance of the CMIP5 models is found to be generally superior to that of the CMIP3 collection, but with much of the improvement occurring in sub-polar and middle latitudes. In contrast, the collective CMIP5 simulation of moist tropical and arid subtropical vegetation types is little better than that of the CMIP3 models. Since no CMIP5 model showed more than about 70 percent overall agreement with the reference vegetation types, there clearly is room for improving the simulation of regional T and P. This is a necessary precondition for representing vegetation that realistically interacts with both the combined global climate and carbon cycle, which are simulated by comprehensive Earth Systems Models.
This work was funded by the U.S. Department of Energy Office of Science and was performed at the Lawrence Livermore National Laboratory under Contract DE-AC52-07Na27344. We also acknowledge the World Climate Research Programme's Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the modeling groups (listed in Table 2 of this paper) for producing and making available their simulation outputs. For CMIP, the U.S. Department of Energy's Program for Climate Model Diagnosis and Intercomparison (PCMDI) provides coordinating support and leads the development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. Finally, we gratefully acknowledge the generous computational assistance provided by Charles Doutriaux.