Urban Heat Island Mapping and Mitigation Planning Using Landsat and Machine Learning

Dense built-up areas with little vegetation become hotter than their surroundings - the urban heat island. This project maps the summer land-surface temperature (LST) of Chennai from Landsat 8/9 thermal data, explains it with spectral indices, land-cover fractions, the distance to the sea and elevation using a random forest and LightGBM compared with the classical NDVI + NDBI regression under honest spatial-block cross-validation, finds statistically significant hot spots with Getis-Ord Gi*, ranks the 200 Greater Chennai Corporation wards for heat mitigation and tests tree-planting and cool-roof scenarios - clearly flagged as model-based associations, not causal estimates. A web application for planners (heat map with the drivers of every pixel, ward priorities, scenario planner with a review workflow) runs on five databases. A GNU Octave cross-check, a complete MATLAB version and a QGIS project are included.

Category: School of Architecture and Planning

Architecture & Planning Project AP02 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.