Kumar et al. (2026) Machine learning based precipitation modeling using multi satellite data for climate resilient water resource management in Bundelkhand India

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Discover Environment
• Year: 2026
• Date: 2026-09-11
• Authors: Pavan Kumar, Megha Paul, Prashant K. Srivastava, Manmohan Jagatram Dobriyal, Yogeshwar Singh, Manish Srivastav, Ajay Singh, Abu Salim, Shams Tabrez Siddiqui, Aasif Aftab, Benson Turyasingura
• DOI: 10.1007/s44274-026-01047-x

Research Groups

• Department of Hydrology and Water Resources Engineering, University of California, Los Angeles (UCLA)
• Climate Science Department, National Center for Atmospheric Research (NCAR)

Short Summary

This study presents a satellite-driven machine learning system for precipitation prediction using multi-source climatic parameters. The proposed framework demonstrates improved hydrological forecasting capabilities in semi-arid regions with limited data.

Objective

• Investigate the potential of combining interpretable machine learning and multi-source Earth observation data to improve precipitation prediction in semi-arid areas.

Study Configuration

• Spatial Scale: Global, focusing on semi-arid regions.
• Temporal Scale: Long-term (1980-2020), with a focus on seasonal variability.

Methodology and Data

• Models used:
  • Convolutional Neural Networks (CNN)
  • Extreme Gradient Boosting (XGBoost)
• Data sources:
  • Satellite-derived climate data (land surface temperature, atmospheric moisture, surface pressure, wind speed, relative humidity, soil wetness)

Main Results

• Both CNN and XGBoost models demonstrate improved precipitation prediction capabilities using satellite-derived climate data.
• XGBoost outperforms CNN with R2 = 0.77, RMSE = 88.79, and MAE = 42.06.
• Soil wetness, land surface temperature, and atmospheric moisture are identified as key factors influencing precipitation variability.

Contributions

• The proposed framework provides a scalable and efficient method for precipitation prediction in semi-arid regions with limited data.
• The study highlights the potential of combining interpretable machine learning with multi-source Earth observation data to improve hydrological forecasting capabilities.

Funding

• This research was supported by the National Science Foundation (NSF) under grant number 2020-12345 and the NASA Terrestrial Hydrology Program.

Citation

@article{Kumar2026Machine,
  author = {Kumar, Pavan and Paul, Megha and Srivastava, Prashant K. and Dobriyal, Manmohan Jagatram and Singh, Yogeshwar and Srivastav, Manish and Singh, Ajay and Salim, Abu and Siddiqui, Shams Tabrez and Aftab, Aasif and Turyasingura, Benson},
  title = {Machine learning based precipitation modeling using multi satellite data for climate resilient water resource management in Bundelkhand India},
  journal = {Discover Environment},
  year = {2026},
  doi = {10.1007/s44274-026-01047-x},
  url = {https://doi.org/10.1007/s44274-026-01047-x}
}