Composite Heat Wave Risk (CHWR) framework using Principal Component Analysis (PCA) and machine learning for assessing...

Area of Science:

• Climate Change Adaptation
• Agricultural Science
• Geospatial Risk Assessment

Background:

• Rising temperatures and extreme heat events threaten food security, especially in climate-sensitive regions like South Asia.
• West Bengal, India, faces significant socio-climatic stress due to diverse agroecology and climate-vulnerable farming systems.
• District-level heat wave risk assessments integrating multiple dimensions are limited, hindering effective agricultural planning.

Purpose of the Study:

• To develop a composite, spatially explicit heat wave risk assessment for agricultural communities in West Bengal.
• To integrate climatic hazards, exposure, and socio-economic vulnerability using the IPCC AR6 framework.
• To predict district-level heat wave risk classes using advanced modeling for climate resilience.

Main Methods:

• Developed a Composite Heat Wave Risk Index (CHWRI) using Principal Component Analysis (PCA).
• Employed a Long Short-Term Memory (LSTM) model optimized with the Hippopotamus Optimization Algorithm (HOA) for spatio-temporal risk prediction.
• Integrated climatic, agricultural, and socio-economic data for a comprehensive risk framework.

Main Results:

• Identified significant spatial heterogeneity in heat wave risk across West Bengal's agroecological zones.
• The PCA-LSTM-HOA model demonstrated strong agreement between observed and predicted risk classes (72.73% accuracy).
• Purulia district was identified as a major heat wave risk hotspot, with significant spatial clustering of risk.

Conclusions:

• The PCA-LSTM-HOA framework provides a novel, interpretable tool for prioritizing heat wave risk in agriculture.
• The findings support targeted early warning systems and climate-resilient agricultural planning in vulnerable regions.
• This research contributes to building resilience against accelerating global warming impacts on food security.