Integrating explainable machine learning and LULC analysis to assess seasonal drivers of urban heat stress and thermal vulnerability

Rapid urbanization in tropical coastal cities has intensified urban heat island effects; however, seasonal dynamics of urban thermal ecological quality and their conditioning mechanisms remain insufficiently explored in South Asian metropolitan environments. This study develops a seasonal explainable framework to predict urban thermal field variance index (UTFVI), identify dominant conditioning factors, and evaluate LULC-specific thermal vulnerability in Chattogram Metropolitan Area (CTGM), Bangladesh, during 2019, 2022 and 2025. Sixteen factors from multi-source geospatial datasets were analyzed using MLR, RF and XGBoost with spatial block cross-validation. SHAP interpreted predictor contributions across six season–year datasets. The results revealed that XGBoost outperformed RF and MLR (R 2 = 0.778–0.812). The Worst UTFVI class covered about one-third of CTGM in every dataset, increasing by 2.51 and 2.63 percentage points in summer and in winter between 2019 and 2025, respectively. NDBI was identified as the dominant predictor in all datasets (mean |SHAP| values: 0.0214–0.0259, direction: +0.965 to +0.980). Nighttime light intensity also positively associated with UTFVI, while precipitation showed negative association. Vegetation revealed the most favorable conditions (80.60–92.16% Excellent), followed by waterbodies (66.96–75.15%). Conversely, Barren land showed the highest worst shares in 2019 and 2022 (56.33–63.14%), whereas built-up areas ranked highest in 2025 (approximately 59%). Within vegetation and agricultural areas, higher EVI was associated with lower UTFVI, while proximity to water had effect. These findings demonstrate how spatially validated, explainable modeling can distinguish persistent thermal drivers from seasonally varying contributions and inform targeted mitigation through vegetation conservation, impervious-surface management and climate-responsive planning in rapidly urbanizing coastal cities.

Authors

Institutions

Publication Details

Journal
Sustainable Futures
Published
2026-11-21
DOI
https://doi.org/10.1016/j.sftr.2026.102314
Primary Topic
Urban Heat Island Mitigation
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Integrating explainable machine learning and LULC analysis to assess seasonal drivers of urban heat stress and thermal vulnerability

Mohammed Abdus Salam, Md. Mahmudul Hasan, Sajid Ibne Jamalfaisal, Sayed Abu Johany
Sustainable Futures
Urban Heat Island Mitigation
article

Integrating explainable machine learning and LULC analysis to assess seasonal drivers of urban heat stress and thermal vulnerability

Mohammed Abdus Salam, Md. Mahmudul Hasan, Sajid Ibne Jamalfaisal, Sayed Abu Johany
article en

Abstract

Rapid urbanization in tropical coastal cities has intensified urban heat island effects; however, seasonal dynamics of urban thermal ecological quality and their conditioning mechanisms remain insufficiently explored in South Asian metropolitan environments. This study develops a seasonal explainable framework to predict urban thermal field variance index (UTFVI), identify dominant conditioning factors, and evaluate LULC-specific thermal vulnerability in Chattogram Metropolitan Area (CTGM), Bangladesh, during 2019, 2022 and 2025. Sixteen factors from multi-source geospatial datasets were analyzed using MLR, RF and XGBoost with spatial block cross-validation. SHAP interpreted predictor contributions across six season–year datasets. The results revealed that XGBoost outperformed RF and MLR (R 2 = 0.778–0.812). The Worst UTFVI class covered about one-third of CTGM in every dataset, increasing by 2.51 and 2.63 percentage points in summer and in winter between 2019 and 2025, respectively. NDBI was identified as the dominant predictor in all datasets (mean |SHAP| values: 0.0214–0.0259, direction: +0.965 to +0.980). Nighttime light intensity also positively associated with UTFVI, while precipitation showed negative association. Vegetation revealed the most favorable conditions (80.60–92.16% Excellent), followed by waterbodies (66.96–75.15%). Conversely, Barren land showed the highest worst shares in 2019 and 2022 (56.33–63.14%), whereas built-up areas ranked highest in 2025 (approximately 59%). Within vegetation and agricultural areas, higher EVI was associated with lower UTFVI, while proximity to water had effect. These findings demonstrate how spatially validated, explainable modeling can distinguish persistent thermal drivers from seasonally varying contributions and inform targeted mitigation through vegetation conservation, impervious-surface management and climate-responsive planning in rapidly urbanizing coastal cities.

Sustainable FuturesVol. 12
Noakhali Science and Technology University (BD), Jagannath University (BD)
Urban Heat Island Mitigation
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon