This study presents a model integrating attention mechanisms and physical constraints to improve flood prediction. It forecasts floods up to 6 hours in advance. The model enhances accuracy by focusing on critical input features and historical patterns. Results demonstrate its superior performance compared to other models, offering improved flood prediction with greater interpretability and alignment with physical laws to aid community preparedness.

Flash drought (FD) in the North China Plain (NCP) from 1981 to 2022 was identified by three methods. It occurs more frequently in central and northern NCP than southern, with longer duration in southern and eastern NCP. FD shows seasonal variations. NCP is prone to severe FDs. There is a decreasing trend in FD, with three hotspots in northwestern, eastern and southwestern NCP. The FD frequency is affected by thresholds in the identification methods. This study may be valuable for FD mitigation.

Reanalysis soil moisture products are a vital basis for hydrological and environmental research. Previous product evaluation is limited by the scale difference (point and grid scale). This paper adopts cosmic ray neutron sensor observations, a novel technique that provides root-zone soil moisture at field scale. In this paper, global harmonized CRNS observations were used to assess products. ERA5-Land, SMAPL4, CFSv2, CRA40 and GLEAM show better performance than MERRA2, GLDAS-Noah and JRA55.