Evaporation variability under climate warming in five reanalyses and its association with pan evaporation over China

Abstract

AbstractWith the motivation to identify actual evapotranspiration (AE) variability under climate warming over China, an assessment is made from five sets of reanalysis data sets [National Centers for Environmental Prediction‐National Center for Atmospheric Research (NCEP‐NCAR), NCEP‐Department of Energy (NCEP‐DOE), Modern‐Era Retrospective Analysis for Research and Applications (MERRA), Interim Reanalysis, and Japanese 55‐year Reanalysis (JRA‐55)]. Based on comparison with AE estimates calculated using the Budyko equation, all five reanalysis data sets reasonably reproduce the spatial patterns of AE over China, with a clearly southeast‐northwest gradient. Overall, JRA‐55 (NCEP‐DOE) gives the lowest (highest) reanalysis evaporation (RE) values. From 1979 to 2013, dominant modes of RE among five reanalyses are extracted using multivariate empirical orthogonal function analysis. Accordingly, the interdecadal variation of RE is likely driven by the change of temperature, and the interannual variation is constrained by the water supply conditions. Under climate warming, RE increase in the Northwest China, Yangtze‐Huaihe river basin, and South China, while they decrease in Qinghai‐Tibet Plateau, and northern and Northeast China. Moreover, the relationship between RE and pan evaporation (PE) are comprehensively evaluated in space‐time. Negative correlations are generally confirmed in nonhumid environments, while positive correlations exist in the humid regions. Our analysis supports the interpretation that the relationship between PE and AE was complementary with water control and proportional with energy control. In view of data availability, important differences in spatial variability and the amount of RE can be found in Northwest China, the Qinghai‐Tibet Plateau, and the Yangtze River Basin. Generally speaking, NCEP‐NCAR and MERRA have substantial problems on describing the long‐term change of RE; however, there are some inaccuracies in the JRA‐55 estimates when focusing on the year‐to‐year variation.