De et al. (2026) Integrating Machine Learning with Geo-Spatial Temporal Satellite Data for Improved Flood Susceptibility Assessment

Identification

• Journal: Lecture notes in networks and systems
• Year: 2026
• Date: 2026-01-01
• Authors: Roshni De, Debatosh Chakraborty, Dwijen Rudrapal, B.B. Bhattacharya
• DOI: 10.1007/978-3-032-06700-5_33

Research Groups

• Department of Mathematics, NIT Agartala, Jirania, India
• Department of Computer Science and Engineering, NIT Agartala, Jirania, India

Short Summary

This study develops a machine learning framework for improved flood susceptibility mapping in the Cachar district, Assam, India, by integrating various geo-spatial and temporal satellite-derived features to enhance predictive accuracy in data-constrained environments.

Objective

• To develop a data-driven framework for flood susceptibility mapping using machine learning and geo-spatial temporal satellite data.
• To examine the influence of multiple satellite-derived geo-spatial and temporal features on flood susceptibility.
• To enhance predictive accuracy in flood-prone, data-constrained environments.

Study Configuration

• Spatial Scale: Cachar district, Assam, India.
• Temporal Scale: Focus on dynamic processes influenced by continuous land use changes and climate change, utilizing recent satellite data for geo-spatial and temporal features, particularly monsoon precipitation.

Methodology and Data

• Models used: Machine Learning classifiers (optimized through rigorous feature selection using IGR and VIF score and comparative evaluation).
• Data sources:
  • NDVI (Normalized Difference Vegetation Index) from Landsat 8
  • LULC (Land Use Land Cover) from Sentinel-2
  • Topographic variables from SRTM DEM (Shuttle Radar Topography Mission Digital Elevation Model)
  • Soil texture from OpenLandMap
  • Monsoon precipitation from CHIRPS (Climate Hazards Group InfraRed Precipitation with Station data)
  • Flood extent information from NDWI (Normalized Difference Water Index)
  • Flood extent information from Sentinel-1 SAR (Synthetic Aperture Radar) data

Main Results

• The developed machine learning framework effectively integrates diverse geo-spatial and temporal satellite data for flood susceptibility mapping.
• A rigorous feature selection process, utilizing IGR (Information Gain Ratio) and VIF (Variance Inflation Factor) scores, along with comparative evaluation across various classifiers, successfully optimized the flood susceptibility model.
• The study underscores the critical importance of integrating machine learning with remote sensing data to construct precise flood risk models.
• The resulting model provides a valuable tool for disaster management teams to accurately identify vulnerable regions.

Contributions

• Development of a novel data-driven framework for flood susceptibility mapping specifically designed for flood-prone, data-constrained environments.
• Demonstration of enhanced predictive accuracy in flood susceptibility assessment through the integration of a comprehensive set of diverse satellite-derived geo-spatial and temporal features with machine learning.
• Application of a rigorous and systematic feature selection process (IGR and VIF score) for optimizing machine learning models in flood susceptibility mapping.
• Providing a practical and precise flood risk model to aid disaster management teams in identifying vulnerable regions, thereby contributing to improved disaster preparedness and mitigation.

Funding

Not specified in the provided text.

Citation

@article{De2026Integrating,
  author = {De, Roshni and Chakraborty, Debatosh and Rudrapal, Dwijen and Bhattacharya, B.B.},
  title = {Integrating Machine Learning with Geo-Spatial Temporal Satellite Data for Improved Flood Susceptibility Assessment},
  journal = {Lecture notes in networks and systems},
  year = {2026},
  doi = {10.1007/978-3-032-06700-5_33},
  url = {https://doi.org/10.1007/978-3-032-06700-5_33}
}