Zhang et al. (2026) Global-regional nested forecasting of soil moisture

Identification

• Journal: npj Climate and Atmospheric Science
• Year: 2026
• Date: 2026-08-25
• Authors: Quan. Zhang, Yuze. Sun, Wenbin. Liu, Yichuan. Zhang, Yanyan. Huang, Jiaolong Ying, Yanfei. Xiang, Dongxiao. Xu, Shuo. Wang, Le. Yu, Xiaomeng. Huang
• DOI: 10.1038/s41612-026-01522-5

Research Groups

• Department of Earth System Science, Ministry of Education Key Laboratory for Earth System Modelling, Institute for Global Change Studies, Tsinghua University, Beijing, China.
• State KeyLaboratory of Climate System Prediction and Risk Management, Nanjing University of Information Science & Technology, Nanjing, China.
• Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Science and Natural Resources Research, Chinese Academy of Sciences, Beijing, China.
• Hebei Province Collaborative innovation center for sustainable utilization of water resources and optimization of industrial structure, School of Water Resources and Environment, Hebei GEO University, Shijiazhuang, China.
• Joint International Research Laboratory of Catastrophe Simulation and Systemic Risk Governance, School of National Safety and Emergency Management, Beijing Normal University, Zhuhai, China.
• State Key Laboratory of Climate Resilience for Coastal Cities (SKL-CRCC), The Hong Kong Polytechnic University, Hong Kong, China.

Short Summary

This study proposes an Artificial Soil Moisture Forecasting Model (ASM) that integrates atmospheric, oceanic, and land surface variables within an autoregressive forecasting framework to predict soil moisture evolution at the synoptic-to-early-subseasonal time scale. ASM consistently outperforms representative deep learning models across forecast lead times.

Objective

• Develop a data-driven, autoregressive forecasting system for predicting soil moisture evolution.
• Improve soil moisture forecast skill and spatial resolution compared to existing approaches.

Study Configuration

• Spatial Scale: Global and regional (0.1°) scales.
• Temporal Scale: Synoptic-to-early-subseasonal time scale (1–14 days).

Methodology and Data

• Models used: Artificial Soil Moisture Forecasting Model (ASM), Transformer, ConvLSTM, ResNet, U-Net.
• Data sources: ERA5 reanalysis dataset, ECMWF forecasts.

Main Results

• ASM consistently outperforms other models across forecast lead times in terms of ACC and RMSE.
• ASM achieves a higher ACC of 0.612 and lower RMSE compared to ECMWF at a 14-day lead time.
• Regional high-resolution forecasts using ASM achieve higher forecast skill than ECMWF-driven regional model.

Contributions

• This study provides an improved data-driven approach for soil moisture forecasting, which complements traditional numerical models.
• The proposed autoregressive framework effectively captures the persistence and low-frequency variability of soil moisture.

Funding

• Not specified in the provided text.

Citation

@article{Zhang2026Globalregional,
  author = {Zhang, Quan. and Zhang, Quan. and Sun, Yuze. and Liu, Wenbin. and Zhang, Yichuan. and Huang, Yanyan. and Ying, Jiaolong and Huang, Yanyan. and Huang, Yanyan. and Xiang, Yanfei. and Xu, Dongxiao. and Wang, Shuo. and Yu, Le. and Huang, Xiaomeng. and Huang, Xiaomeng.},
  title = {Global-regional nested forecasting of soil moisture},
  journal = {npj Climate and Atmospheric Science},
  year = {2026},
  doi = {10.1038/s41612-026-01522-5},
  url = {https://doi.org/10.1038/s41612-026-01522-5}
}