Mominul IslamResearch Projects

Research Projects

Completed (ready to submit)
  • An Explainable AI-Enhanced Watershed Digital Twin for Streamflow Prediction under Hydroclimatic Extremes in the Transboundary Sari-Gowain Basin
  • Seasonal Water Paradox and Compound Hydroclimatic Risk Regimes in Bangladesh's Teesta River Basin: A Remote Sensing and Explainable Machine Learning Assessment
  • Hydroclimatic whiplash and cropland stress in the Sunamganj Haor, Bangladesh
  • Deep Learning-Based Monthly Rainfall Forecasting Across Bangladesh Using CHIRPS Data: A CNN–LSTM Comparison
  • Spatiotemporal Hydro-Environmental Controls on Vegetation Condition in Ratargul Swamp Forest: An Explainable Remote-Sensing and Machine-Learning Assessment
Ongoing
  • Regional Frequency Analysis of Extreme Daily Rainfall in the Surma–Kushiyara Basin, Bangladesh: Quality Control, Model Evaluation, and Return-Level Estimation
  • When Will Extreme Rainfall End? Explainable Prediction of Rainfall Persistence and Termination in the Northeastern Haor Region of Bangladesh
  • Understanding Multi-Hazard Risks in a Changing Climate: A Review of Global Patterns, Drivers, and Impacts (2015–2025)
  • Pre-Harvest Flood Risk and Vegetation Impacts on Boro Rice in Northeastern Bangladesh: A 25-Year Satellite Assessment

FloodML-Geo, a QGIS plugin

Flood susceptibility mapping inside QGIS with explainable machine learning. The flood raster goes in; spatially validated models, SHAP explanations and susceptibility maps come out. Includes a self-test and a check that flags test data overlapping the training data.

QGISRandom ForestXGBoostLightGBMSpatial cross-validationSHAP
Flood susceptibility class map produced with FloodML-Geo