The Thermal Footprint of Urbanization: Decadal Land Cover Dynamics, Land Surface Temperature Intensification, and Machine Learning Projections in the Western Peninsula of Sierra Leone (2014–2024)

Abstract

Rapid conversion of vegetated land to impervious surfaces drives Urban Heat Island intensification, posing acute risks to human health and livelihoods across tropical Africa. This study presents the first systematic, biennial decadal analysis of Land Surface Temperature (LST) dynamics and their mechanistic linkage to Land Use/Land Cover (LULC) change across the Western Peninsula of Sierra Leone (2014–2024) — a biodiverse coastal landscape encompassing Freetown, the Western Area Peninsula National Park, and West Africa’s most ecologically significant mangrove systems. Six Landsat 8 OLI/TIRS scenes were processed through a validated single-channel emissivity-correction algorithm. Results document a mean LST increase of 2.3 °C over the decade, with areas exceeding 30 °C expanding from 18.2% to 35.8% — a near-doubling of thermally dangerous land in ten years. Stratified NDVI–LST analysis (n ≥ 500 per epoch; three zones: urban core, peri-urban transition, forested interior) confirms a persistent, statistically significant negative relationship across all epochs (R² = 0.30–0.50; slopes −20.9 to −22.2 °C/NDVI unit), quantifying a cooling benefit of 8.4 °C per 0.40 NDVI gain — an actionable planning metric. Zone-stratified modelling identifies the peri-urban transition belt as the most thermally dynamic and management-sensitive stratum. Benchmarked against Lagos, Accra, and Addis Ababa, Freetown’s warming is structurally distinct owing to peninsular confinement, a dual-ecosystem thermal buffer, and active national park forest degradation. A Random Forest framework is outlined for scenario-based LST projection under SSP2-4.5 assumptions to 2030. Findings deliver spatial guidance for Freetown’s Heat Action Plan, the FreetownTheTreeTown reforestation initiative (Adusei, 2026), and analogous programmes across climate-vulnerable tropical cities.