Spatiotemporal prediction of shallow water table depths in continental China

Macro-scale spatiotemporal distribution of shallow water table depths is important for terrestrial-ecosystem and climate research, and management of water resources. In this paper, an approach is presented to predict the water table depths from climate forcing at a large scale for a region such as continental China. This is achieved by adopting transfer function-noise (TFN) models and parameter regionalization methods. The parameters of the TFN models, which use precipitation surplus (precipitation minus potential evapotranspiration) as the input and water table depth as the output, are calibrated in gauged areas by the Kalman filter method coupled with the global optimization algorithm SCE-UA (shuffled complex evolution method developed at The University of Arizona), and the calibrated parameters are then regionalized to ungauged areas for the region within the same classified zones based on Gaussian Mixture Model (GMM) clustering method. The region such as continental China is classified into several zones by the GMM clustering method according to soil and climate characteristics. Verification and cross-validation show that the proposed methods for calibration and regionalization of the TFN-models for estimation of the water table depths are effective. The spatial and temporal variations of water table depths at a macro-scale in continental China are predicted by TFN models with the calibrated parameters for gauged areas and the regionalized parameters for ungauged areas.