Long-term reconstruction of satellite-based precipitation, soil moisture, and snow water equivalent in China

Abstract. A long-term high-resolution national dataset of precipitation (P), soil moisture (SM), and snow water equivalent (SWE) is necessary for predicting floods droughts assessing the impacts climate change on streamflow in China. Current daily or sub-daily datasets P, SM, SWE are limited by a coarse spatial resolution lack local correction. Although SM data derived from hydrological simulations at scale have fine resolutions take advantage forcing data, models not directly calibrated with data. In this study, we produced 0.1? 1981–2017 across China, using global background on-site as input satellite-based reconstruction benchmarks. Global 0.25?P merged to reconstruct historical P China Merged Precipitation Analysis (CMPA) available 2008–2017 stacking machine learning model. The reconstructed used drive HBV model simulate 1981–2017. simulation Soil Moisture Active Passive Level 4 (SMAP-L4) depth (Che Dai, 2015) Moderate Resolution Imaging Spectroradiometer (MODIS) cover Cross-validated temporal splitting CMPA median Kling–Gupta efficiency (KGE) 0.68 all grids scale. KGE calibration 0.61 For two snow-rich regions, KGEs 0.55 ?2.41 Songhua Liaohe basins northwest continental basin respectively Generally, performs better southern eastern than northern western northeast other regions SWE. As first that combines information observations benchmarks, product valuable future investigations processes.