Machine Learning Framework for High-Resolution Air Temperature Downscaling Using LiDAR-Derived Urban Morphological Features

Aug 4, 2026

Urban climate models often lack the spatial resolution needed for detailed city-scale air temperature studies, making high-resolution estimation computationally demanding. A data-driven framework is proposed to downscale air temperature using publicly available outputs from urban climate models, especially UrbClim datasets. The approach extracts urban morphological features from LiDAR by building 3D models with deep learning, then combines them with meteorological parameters to train machine-learning models. Results show LightGBM achieves the best accuracy (RMSE 0.352 K, MAE 0.215 K) and enables street-level temperature pattern analysis.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages42

Reading Time1 h 46 min

AccessFree to download

Introduction

Urban microclimate and urban climate modeling

Downscaling need for mismatch of spatial scales

arXiv:2409.02120v1 [[physics. ao-ph](physics. ao-ph)] 31 Aug 2024 Machine Learning Framework for High-Resolution Air Temperature Downscaling Using LiDAR-Derived Urban Morphological Features Fatemeh Chajaei, Hossein Bagheri Faculty of Civil Engineering and Transportation, University of Isfahan, Isfahan, Iran, [h.bagheri@cet.ui.ac.ir](h.bagheri@cet.ui.ac.ir) Abstract This is the pre-acceptance version, to read the final version, please go to Urban Climate on ScienceDirect, [https://www.sciencedirect.com/science/article/abs/pii/S2212095524002992](https://www.sciencedirect.com/science/article/abs/pii/S2212095524002992. Climate)[. Climate](https://www.sciencedirect.com/science/article/abs/pii/S2212095524002992. Climate)[ ](https://www.sciencedirect.com/science/article/abs/pii/S2212095524002992. Climate)models lack the necessary resolution for urban climate studies, requiring computationally intensive processes to estimate high resolution air temperatures. In contrast, Data-driven approaches offer faster and more accurate air temperature downscaling. This study presents a data-driven framework for downscaling air temperature using publicly available outputs from urban climate models, specifically datasets generated by UrbClim. The proposed framework utilized morphological features extracted from LiDAR data. To extract urban morphological features, first a three-dimensional building model was created using LiDAR data and deep learning models. Then, these features were integrated with meteorological parameters such as wind, humidity, etc., to downscale air temperature using machine learning algorithms. The results demonstrated that the developed framework effectively extracted urban morphological features from LiDAR data. Deep learning algorithms played a crucial role in generating three-dimensional models for extracting the aforementioned features. Also, the evaluation of air temperature downscaling results using various machine learning models indicated that the LightGBM model had the best performance with an RMSE of 0.352°K and MAE of 0.215°K. Furthermore, the examination of final air temperature maps derived from downscaling showed that the developed framework successfully estimated air temperatures at higher resolutions, enabling the identification of local air temperature patterns at street level. The corresponding source codes are available on GitHub: [https://github.com/FatemehCh97/Air-Temperature-Downscaling](https://github.com/FatemehCh97/Air-Temperature-Downscaling). Keywords: Urban microclimate, Air temperature downscaling, 3D building model, Machine learning, Deep learning, LiDAR, Urban morphology 1. Introduction With the growing trend of urbanization worldwide, urban areas are undergoing rapid development, leading to the emergence of unique climatic conditions known as urban climate [1, 2] . Urban climate is a complex phenomenon formed by a combination of factors such as regional climate patterns, human activities, industrial and commercial activities, land-use patterns, etc. [3, 4 , 5 , 6] . As urbanization continues, changes in urban climate are increasing globally. Urban climate change is one of the challenges of the city, which has extensive impacts on health, livelihoods, economy, infrastructure, services, and ecosystems [7, 8] . Therefore, understanding urban climate is essential for efficient urban planning, sustainable development, and ensuring the well-being of urban residents [9] . Urban climate models enable the simulation, prediction, and assessment of various climate scenarios by analyzing different meteorological variables, providing valuable insights for urban planners and decision-makers [10] . Preprint submitted to Elsevier September 5, 2024 Research on climate modeling is constantly evolving thanks to the development of new sensors and measurement techniques, increased computational power, and advanced knowledge. Among the various climate parameters, air temperature is one of the key

FAQs

Why is air temperature downscaling needed in urban climate studies?

Urban climate models typically output data at resolutions too coarse for planners and designers, so downscaling is required to capture urban detail and complexity for local analyses.

How does the framework derive urban morphological features?

It creates 3D building models from LiDAR data using deep learning, then extracts morphological features and integrates them with meteorological variables for downstream temperature prediction.

Which machine learning model performs best and what are the reported errors?

LightGBM delivers the best performance, with RMSE of 0.352 K and MAE of 0.215 K in the evaluation of air temperature downscaling.

Aug 4, 2026

Upload by Genevieve