Evaluating historical drought characteristics simulated in CORDEX East Asia against observations

Abstract

ABSTRACTThis study aims to understand how well regional climate models (RCMs) reproduce the observed characteristics of droughts based on the standardized precipitation evapotranspiration index (SPEI) over East Asia. While RCMs from CORDEX East Asia have been evaluated in terms of the mean and extreme values of precipitation and air temperature in the literature, their performance based on drought characteristics has not been revealed. Therefore, we investigate the temporal characteristics, such as the trend and stationarity, and other drought measures, such as the drought frequency, duration, severity and spatial extent, to determine how well these factors are captured and how they vary among the different models, including four RCMs, their ensemble mean and a driving general circulation model (GCM). The results are compared to a 26‐year observational dataset (1980 to 2005) from APHRODITE. Based on our analyses of the results, the specific findings are as follows. First, based on Mann–Kendall (MK) trend tests, the SPEI shows decreasing trends over the northern part of the domain and no trends in other regions, and these tendencies are captured by a limited number of RCMs. Notably, these trends were not captured by the ensemble mean of RCMs. Second, based on the Phillips–Perron stationarity test, stationarity is dominant in most regions based on both the observations and the RCMs. Third, the frequency decreases with increasing duration in all regions. This tendency is generally captured in the models to varying extents. Fourth, the drought severity decreases with increasing duration in all cases, as expected. However, the drought severity diverges markedly among the RCMs. Finally, estimates of the spatial extent using the RCMs are generally accurate in wet regions but inaccurate in dry regions. Furthermore, the spatial extents of the droughts diverge among the RCMs, and the models fail to accurately capture drought events with large spatial scales.