Evaluating CMIP Simulations of Historical Continental Climate using Koeppen Bioclimatic Metrics

Tuesday, May 13, 2014 - 07:00
Add to Calendar


The classic Koeppen bioclimatic classification scheme associates generic vegetation types (e.g. grassland, tundra, broadleaf or evergreen forests, etc.) with regional zones that are defined by characteristics of the annual cycles of continental temperature (T) and precipitation (P), two climate variables whose interplay is critically important for living organisms. The locations or areas of the Koeppen vegetation types derived from observational data thus can provide concise metrical standards for simultaneously evaluating CMIP historical climate simulations of T and P in naturally defined regions. For this study, 14 Koeppen vegetation types are derived from annual-cycle climatologies of T and P in some 40 CMIP3 and CMIP5 simulations of the 1980-1999 period. Metrics for evaluating the ability of the CMIP3/5 models to simulate the correct locations and areas of each vegetation type, as well as measures of overall model performance, also are developed. It is found that the CMIP models generally are most deficient in simulating: 1) climates of drier zones (e.g. desert, savanna, grassland, steppe vegetation types) located in the southwestern U.S. and Mexico, eastern Europe, southern Africa, and central Australia; 2) climates of regions such as central Asia and western South America where topography plays a key role. Nevertheless, CMIP5 models tend to show incremental improvements in simulation performance relative to their CMIP3 counterparts. A deconstruction of selected regional T or P biases that exemplify general model performance problems also will be presented. Acknowledgments: 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.

Presentation file(s):