Discovering Nonlinear Urban Heat Drivers Through Bayesian-Optimised Boosted Trees

DOI

10.2991/978-94-6239-703-3_11How to use a DOI?

Keywords

Urban Thermal Environment; Lightgbm; Bayesian Optimised; Additive Interpretation Algorithm; Urban Morphology

Abstract

Urban thermal environments are shaped by complex interactions between built form and ecological elements, yet many studies still rely on linear assumptions that obscure threshold and stage-dependent behaviours. This study investigates the nonlinear drivers of daytime Land Surface Temperature (LST) in Beijing’s central urban area using multi-source geospatial datasets and a four-season modelling strategy. We construct seasonal regression models with LightGBM and tune hyperparameters via Bayesian optimisation. Spatial dependence is assessed using Global Moran’s I and partially addressed by including a spatial autocorrelation term. To improve interpretability of the ensemble models, Shapley Additive Explanation (SHAP) is applied to quantify both global contributions and local nonlinear effects of urban attributes. Results show that Fractional Vegetation Cover (FVC) shows the strongest cooling association with LST, especially in summer, while Water Density (WD) provides a stable cooling effect. Building Height (BH) is generally associated with lower LST within an effective range, whereas Building Footprint (BF) tends to be associated with increasing LST as density increases. Sky View Factor (SVF) and Building Plot Ratio (BP) contribute marginally. The findings provide interpretable, data-driven guidance for season-sensitive urban climate adaptation through ecological restoration and morphology optimisation.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Xiao Zhang
AU  - Tao Wu
PY  - 2026
DA  - 2026/06/08
TI  - Discovering Nonlinear Urban Heat Drivers Through Bayesian-Optimised Boosted Trees
BT  - Proceedings of the 2026 2nd International Conference on Engineering Management and Safety Engineering (EMSE 2026)
PB  - Atlantis Press
SP  - 121
EP  - 130
SN  - 2352-5401
UR  - https://doi.org/10.2991/978-94-6239-703-3_11
DO  - 10.2991/978-94-6239-703-3_11
ID  - Zhang2026
ER  -