EventsThe 1st International Online Conference on Earth Science
Published
This submission belongs to the session S3. Climate Dynamics, Variability and Change of the event The 1st International Online Conference on Earth Science
Published date
31 Aug, 2026
Academic Editor
author-avatarCharles Jones
Citation
Anower Hossain Abid, Debbendu Saha, Multi-Index Drought Projections Under CMIP6 Scenarios in Northwestern Bangladesh: A Dual-Scale Model Screening and Explainable Machine Learning Approach, in Proceedings of The 1st International Online Conference on Earth Science, 2 September–4 September 2026, MDPI: Basel, Switzerland
Share
Email
Facebook
Twitter
LinkedIn

Multi-Index Drought Projections Under CMIP6 Scenarios in Northwestern Bangladesh: A Dual-Scale Model Screening and Explainable Machine Learning Approach

image
1. Department of Civil Engineering, Khulna University of Engineering and Technology, Khulna, Bangladesh
2. School of Civil Engineering and Environmental Science, The University of Oklahoma, Norman, USA
Abstract

In Bangladesh, agricultural survival basically depends on what happens in the Rajshahi Division. That area now faces a severe threat. Temperatures are climbing while rainfall gets less predictable. People have modeled South Asian climate change broadly before. But mapping exactly how hot and dry extremes overlap locally? That is still missing. Here, we track how severe heat and future drought will likely shift across Rajshahi. We utilized 13 CMIP6 Global Climate Models in a Multi-Model Ensemble. Raw climate outputs hold natural flaws. To fix those natural flaws, we pushed the daily temperature and rain readings through Empirical Quantile Mapping. We checked those adjustments directly against ERA5 reanalysis and real records kept by the Bangladesh Meteorological Department. Future hydroclimate scenarios under SSP245 and SSP585 show massive deviations from current baselines. Mean temperatures are tracking toward a 2.1°C increase by 2100. High emissions push that figure to 3.4°C. Applying the Standardized Precipitation Evapotranspiration Index reveals severe future drought conditions. Local conditions will worsen drastically. Extremely dry periods, where index values sink below negative 1.5, are modeled to happen 24 to 32 percent more often than historical baselines. Simultaneous extreme heat and drought surges are even more worrying. Consider days where heat breaches the 90th percentile while rain falls below the 10th. Historically, the area saw about 12 of those days yearly. Under high emission scenarios, they leap to over 22 days annually. Almost double. What exactly fuels these localized disasters? To find out, an XGBoost machine learning model was combined with SHapley Additive exPlanations (SHAP). The outcome was clear. Rainfall shortages used to control drought severity. But moving forward, maximum temperature fluctuations have taken over as the dominant force. Northwestern Bangladesh is basically undergoing a fundamental climate transition. Because intense heat, rather than just missing rain, will drive future agricultural droughts, officials must draft regional adaptation plans immediately to secure local water and food supplies.

Keywords
CMIP6
Compound Extremes
Explainable AI (SHAP)
SPEI Climate Change Adaptation
Northwestern Bangladesh
Integrated Hydro-Climatic Agricultural Land Suitability Assessment Using Multi-Criteria Decision Analysis and Long-Term Satellite Observations
Detection and Analysis of Extreme Temperature Events in Chuadanga, Bangladesh