He et al. (2026) Coupling data assimilation and machine learning to improve land surface conditions and near-surface temperature and humidity forecasts

Identification

• Journal: Agricultural and Forest Meteorology
• Year: 2026
• Date: 2026-02-10
• Authors: Xinlei He, Shaomin Liu, Tongren Xu, Fei Chen, Zhitao Wu, Ziwei Xu, Xiang Li, Rui Liu
• DOI: 10.1016/j.agrformet.2026.111063

Research Groups

• Institute of Loess Plateau, Shanxi University, Taiyuan, China
• State Key Laboratory of Earth Surface Processes and Disaster Risk Reduction, Faculty of Geographical Science, Beijing Normal University, Beijing, China
• Division of Environment and Sustainability, The Hong Kong University of Science and Technology, Hong Kong, China
• College of Geography and Remote Sensing Sciences, Xinjiang University, Urumqi, China
• Institute of Urban Study, School of Environmental and Geographical Sciences, Shanghai Normal University, Shanghai, China

Short Summary

This study coupled a hybrid data assimilation-machine learning framework (DL) with the Weather Research and Forecasting (WRF) model to quantify the impacts of incorporating soil moisture (SM) and vegetation data on land surface initialization and near-surface weather forecast accuracy. The results indicate that optimizing leaf area index (LAI) and SM significantly improves the simulation of water table depth, evapotranspiration, air temperature, and humidity, and refines land surface initial conditions for improved near-surface weather forecasts.

Objective

• To quantify the impacts of incorporating soil moisture (SM) and vegetation data on land surface initialization and near-surface weather forecast accuracy by coupling a hybrid data assimilation-machine learning framework (DL) with the Weather Research and Forecasting (WRF) model.

Study Configuration

• Spatial Scale: Southern Great Plains (SGP) of the United States.
• Temporal Scale: Across dry and wet years; forecast periods extending up to one month.

Methodology and Data

• Models used: Hybrid data assimilation-machine learning framework (DL), Weather Research and Forecasting (WRF) model.
• Data sources: Satellite-based leaf area index (LAI), multi-source soil moisture (SM) data.

Main Results

• Optimizing LAI and SM significantly improves the simulation of water table depth, evapotranspiration (ET), air temperature, and humidity within the WRF model.
• LAI optimization provides additional benefits to the WRF model, particularly during dry years.
• The DL method effectively refines land surface initial conditions at the beginning of the forecast period.
• This refinement improves the estimation of near-surface atmospheric conditions (e.g., air temperature and humidity) and alters precipitation patterns during the forecast period.
• The integration of LAI and SM is more effective in improving forecasts during wet/normal years compared to dry years.
• The DL method can optimize initial conditions and improve near-surface weather forecasts for up to one month.

Contributions

• Development and application of a novel coupled hybrid data assimilation-machine learning framework (DL) with the WRF model for land surface initialization.
• Quantification of the specific impacts of incorporating satellite-based LAI and multi-source SM data on land surface conditions and near-surface weather forecast accuracy.
• Demonstration of the significant improvements in simulating key hydrological and atmospheric variables (water table depth, ET, air temperature, humidity) through LAI and SM optimization.
• Identification of the differential benefits of LAI optimization in dry years and the varying effectiveness of LAI and SM integration across different climatic conditions (wet/normal vs. dry years).
• Validation of the DL method's capability to refine initial conditions and extend the accuracy of near-surface weather forecasts for up to one month.

Funding

• Not specified in the provided text.

Citation

@article{He2026Coupling,
  author = {He, Xinlei and Liu, Shaomin and Xu, Tongren and Chen, Fei and Wu, Zhitao and Xu, Ziwei and Li, Xiang and Liu, Rui},
  title = {Coupling data assimilation and machine learning to improve land surface conditions and near-surface temperature and humidity forecasts},
  journal = {Agricultural and Forest Meteorology},
  year = {2026},
  doi = {10.1016/j.agrformet.2026.111063},
  url = {https://doi.org/10.1016/j.agrformet.2026.111063}
}