Abstract Author: Xiao-Wei Quan, Martin Hoerling, Jon Eischeid
Abstract Title: 20th Century U.S. Drought in Observations and CMIP Simulations
Abstract: Greenhouse gas forced climate simulations (CMIP) and observational data are diagnosed to understand the temporal variability of large-scale, severe intensity drought over the contiguous U.S. during 1901-2000. We show that when drought measures are properly defined from monthly data, their statistical properties such as area-coverage of severe drought is similar among models, and furthermore that the statistical distribution of the drought metrics in models are in close agreement with observed estimates. Suitable drought indices include the Standardized Precipitation Index (SPI) and the self-calibrated Palmer Drought Severity Index (SC-PDSI).
These robust measures for the description of drought variability are then used to examine the change in the statistics of large-scale severe drought in U.S. during the 20th Century. The frequency of large-scale severe drought in U.S. is found to increase substantially during the post-1950 period in the CMIP simulations. This is particularly evident in drought estimated from a PDSI metric (which estimates the combined effect of precipitation and temperature on surface moisture balances), though there is only a modest change in precipitation in the models. In contrast, the observed PDSI shows a slight decrease in the frequency of large-scale severe drought in U.S. post-1950. We argue that the reason for this difference between the model and observed changes in U.S. drought is due to a substantial increase in observed precipitation over U.S. in recent decades, a feature not simulated in the CMIP runs. The increased precipitation has acted to overwhelm the impact of rising land surface temperatures that decreases available surface moisture as suggested by the CMIP runs.