The Predictability of Annual Evapotranspiration and Runoff in Humid and Nonhumid Catchments over China: Comparison and Quantification

Climate change and its potential threats on water security call for reliable predictions of evapotranspiration (ET) and runoff Q at different time scales, but current knowledge of the differences in their predictability between humid and nonhumid regions is limited. Based on spatially distributed catchments in China, the authors characterized their predictability and provided plausible explanations. Using the Budyko framework, it was confirmed that annual ET is predictable in nonhumid regions but less predictable in humid regions, and annual Q is predictable in humid regions but less reliable in nonhumid regions. The main cause of the varied predictability lies in the variation of water storage change ΔS in the water balance equation. It affects both the estimation and the variability of Q in nonhumid catchments more than that in humid catchments, which increases the challenge of predicting annual Q in nonhumid regions, while the opposite effect occurs in annual ET prediction between humid and nonhumid catchments. Moreover, the differences between the controlling factors of ET variability in different regions add more differences in their predictability. The dominant control of precipitation makes it easy to predict annual ET in nonhumid regions. By contrast, precipitation, potential evaporation, and their covariance take considerable effort to determine annual ET variations, which leads to less reliable ET estimation and predictability in humid catchments. Therefore, one can accurately predict annual ET in nonhumid catchments and Q in humid catchments based on commonly used hydrological models. With proper consideration of ΔS, the predictability of annual ET and Q in both humid and nonhumid catchments can be improved.