Data, Code, and Trained Models for Spatiotemporal Downscaling of Terrestrial Water Storage Anomalies in the Yangtze River Basin

This repository contains the data, code, and trained model weights supporting the study “Spatiotemporal Downscaling of Terrestrial Water Storage Anomalies in the Yangtze River Basin by Integrating Multi-Source GRACE Products and Cross-Scale Physical-Consistency Constraints.” The archived data include the 0.5° BTCH-fused terrestrial water storage anomaly (TWSA) dataset and the final 0.1° downscaled TWSA dataset for the Yangtze River Basin. The repository also contains scripts for data preprocessing, spatiotemporal tensor preparation, U-ConvLSTM-CBAM model training, cross-scale physical-consistency fine-tuning, and related analyses, together with the pretrained 0.5° model and the final fine-tuned 0.1° model weights.