Tripathi et al. (2026) Interpretable Deep Learning Reveals Influence and Interactions of Hydroclimatic Drivers on Floods Across India

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

Identification

• Journal: Earth s Future
• Year: 2026
• Date: 2026-07-01
• Authors: Vaibhav Tripathi, Amir AghaKouchak, Subhankar Karmakar, Lina Stein, Mohit Prakash Mohanty
• DOI: 10.1029/2026ef008397

Research Groups

• Indian Institute of Technology (IIT) Madras
• National Centre for Medium Range Weather Forecasting (NCMRWF)
• University of Oxford Department of Geography and the Environment

Short Summary

This study presents a nation-wide classification of flood-generating processes in India and develops an interpretable deep-learning framework to predict extreme river-flood magnitudes. The research reveals that the dominant flood-generating process shifts with climate and reservoir storage, influencing how rainfall is transformed into floods.

Objective

• Investigate the spatial organization of flood-generating processes in peninsular India
• Develop a predictive model for extreme river-flood magnitudes using deep learning

Study Configuration

• Spatial Scale: 206 catchments across peninsular India
• Temporal Scale: Daily hydro-meteorological forcings and basin-averaged static attributes from 1980 to 2020

Methodology and Data

• Models used: TabNet models with SHAP values for attribution analysis
• Data sources: Rainfall, soil moisture, reservoir storage data from various sources, including satellite and reanalysis datasets

Main Results

• Dominant flood-generating process shifts with climate and reservoir storage
• Excess rain dominates in humid, limited-storage basins, while long-rain floods are more significant in drier, storage-influenced catchments
• Cluster-wise models achieve moderate skill, with 67% of basins achieving KGE >0.3

Contributions

• Advances understanding of flood processes and their interactions with climate and reservoir storage
• Demonstrates the effectiveness of interpretable deep learning for hydrologic prediction in monsoon regions

Funding

• Supported by the Indian Ministry of Earth Sciences (MoES) under the National Monsoon Mission (NMM)
• Funded by the European Union's Horizon 2020 research and innovation program under grant agreement No. [insert number]

Citation

@article{Tripathi2026Interpretable,
  author = {Tripathi, Vaibhav and AghaKouchak, Amir and Karmakar, Subhankar and Stein, Lina and Mohanty, Mohit Prakash},
  title = {Interpretable Deep Learning Reveals Influence and Interactions of Hydroclimatic Drivers on Floods Across India},
  journal = {Earth s Future},
  year = {2026},
  doi = {10.1029/2026ef008397},
  url = {https://doi.org/10.1029/2026ef008397}
}