Conference Agenda

The urban thermal environment has become a challenge to humans in consideration of rapid urbanization and global warming. Various climate classification methods have been developed to analyze urban form and the urban heat island phenomenon. However, there is a lack of cross-comparison studies carried out to examine the accuracy of predicting land surface temperature by different climate classification methods (local climate zone, urban functional zone, and hybrid zone that integrates the strengths of local climate zone and urban functional zone), as well as their performance in statistical and machine learning models (ordinary least squares regression, geographically weighted regression, and random forest regression). Accordingly, this study focuses on comparing the performance and accuracy of predicting land surface temperature via different climate classification methods. In addition, the relative importance and marginal effect of factors on land surface temperature are discussed based on the approach with the highest accuracy. The results show that: random forest model performs best in predicting land surface temperature (average R2: 0.72); hybrid zone is the most accurate approach to predict land surface temperature (R2: 0.84); and urban functional zone (R2: 0.80) performs slightly better than local climate zone (R2: 0.76). This study helps urban planners and designers to assess which climate classification methods can more accurately predict and explain the influence of urban form on land surface temperature, and provides some insights into urban design strategies to improve the thermal environment.

Introduction

Global cities have increasingly replaced horizontal expansion with vertical growth (Frolking et al., 2024), reshaping 3D urban structures and modifying urban microclimates. Additionally, rapid urbanization leads to urban densification, which exacerbates urban warming and amplifies urban energy burdens. Vertical Greenery Systems (VGS)—integrating vegetation into building facades—effectively utilize the growing, unused facade spaces. Simultaneously, VGS provide a proven strategy to alleviate urban warming and reduce building energy consumption. 3D city models are well-suited for simulating virtual scenarios with VGS, as they act as a tool to represent 3D urban morphology of cities. Meanwhile, VGS simulation offers a new approach for expanding application domains of 3D city models. However, studies that integrate 3D city models and assess VGS’s potential remain scarce. Furthermore, due to the varying climatic conditions and dominant urban morphologies, the cooling and energy-saving potential that VGS bring are heterogeneous across cities. Exploring the correlation between climate background, urban morphology and VGS’s cooling and energy-saving potential can facilitate optimal deployment. In this study, we primarily develop a VGS simulation workflow. Next, simulations of three representative cities are conducted.

Methods

We employ a numerical model based on energy balance of VGS. Building facades from 3D city models are extracted and subsequently subdivided into 1-meter grid cells using Ladybug Tools. To obtain the corresponding fine-grained parameters as inputs for the model, we implement three major parts: microclimate simulation, solar irradiation calculation and view factors computation.

First, to simulate microclimate, buildings with similar morphological characteristics are clustered (Yang et al., 2024), with their average morphological parameters fed into the Vertical City Weather Generator (VCWG), an urban canopy model (UCM) that provides high-fidelity vertical variation of microclimate variables (Moradi et al., 2022). The energy balance model of VGS is coupled with VCWG to reflect their effects on microclimate. The cooling impact is quantified by comparing the air temperature at 1.5 meters of each homogeneous morphological unit before and after applying VGS.

Second, for solar irradiation calculation, we use the open source PVLIB-Python library to obtain the sunrise and sunset times. The Point-in-Time Grid Based simulation in Ladybug Tools is employed at grid cell level, along with Accelerad which accelerates daylighting analysis on GPU.

Third, the view factors computation consists of computing sky view factor (SVF), building view factor (BVF) and ground view factor (GVF) of each grid cell. The SVF is derived via ray-tracing. The BVF is calculated as the proportion of rays emitted from grid cells hitting buildings, using Accelerad. After fetching the values of SVF and BVF, GVF is determined by subtracting both from 1.

Finally, the numerical model is solved to derive the exterior surface temperature of each grid cell, allowing further evaluation of the reduced heat flux through the wall as a result of the shading effect of VGS (Susorova et al., 2013). Together with the differences between air temperature and indoor temperature to estimate heat removal rate, the total decreased energy consumption of a building can be assessed.

The proposed method is applied to Wuhan (Cfa), Hong Kong (Cwa), and Singapore (Af), representing distinct climates in Köppen climate classification with varying dominant urban morphologies. Building footprints, digital elevation models (DEM), and meteorological data are collected, with simulations spanning 24 hours on June 21^st, a typical hot summer day, to capture VGS’s performance under heat stress.

Results and discussion

Based on the workflow, we expect VGS to deliver cooling and energy-saving benefits in all three cities compared to baseline scenarios without VGS. In Wuhan, with its hot-humid summer, VGS is anticipated to significantly reduce air temperature at 1.5 meters through evapotranspiration and lower the energy consumption by shading during 6:00 to 20:00. In Hong Kong, marked by mountainous landscape and high-rise high-density morphology, the amount of energy conservation from VGS is potentially less pronounced than in Wuhan, but VGS also achieves a notable cooling effect under its monsoon-influenced humid subtropical climate during the daytime. In Singapore, the cooling and energy-saving impacts are expected to be substantial during the daytime, though weaker than those in Wuhan and Hong Kong, which can be attributed to its relatively low maximum daily air temperature in its tropical rainforest climate. At nighttime, the cooling impacts in all three cities are almost negligible. The exterior surface temperatures of vegetated facades are higher than bare facades, causing a slight increase in energy consumption.

Our work presents an innovative workflow to evaluate the potential of applying VGS on building facades through 3D city models across diverse cities, providing a basis for design guidelines. The anticipated results reveal that the cooling effect of VGS is amplified under higher air temperatures, while the shading effect might be impaired by mutual shading in high-rise, high-density cities during the daytime. Moreover, at nighttime, almost no cooling impact was found with the weakened evaporation rate, and the increase of building energy consumption can be explained by VGS’s insulation effect. These findings suggest that cooling and energy-saving benefits strongly correlate with climatic background and urban morphology, which indicates the importance of tailored schemes for different cities. It should be noted that since climatic background matters, the cross-seasonal influences of VGS deserve further exploration. In future research, we will incorporate multi-seasonal VGS simulation.

Conclusion

The study presents a novel workflow for assessing the cooling and energy-saving potential of VGS in Wuhan, Hong Kong, and Singapore, demonstrating both the capability and suitability of 3D city models for VGS simulation. The expected results indicate that VGS can mitigate urban warming and reduce energy consumption, with performance varying across different climatic conditions, urban morphologies and time of day. These findings highlight a promising strategy to utilize building facades and offer insights for urban planners to optimize the real-world deployment of VGS. Future research should extend the temporal scope through multi-seasonal VGS simulation.

References

Frolking, S., Mahtta, R., Milliman, T., Esch, T., & Seto, K. C., 2024. Global urban structural growth shows a profound shift from spreading out to building up. Nature Cities, 1(9), 555-566.

Moradi, M., Krayenhoff, E. S., & Aliabadi, A. A., 2022. A comprehensive indoor–outdoor urban climate model with hydrology: The Vertical City Weather Generator (VCWG v2. 0.0). Building and Environment, 207, 108406.

Susorova, I., Angulo, M., Bahrami, P., & Stephens, B., 2013. A model of vegetated exterior facades for evaluation of wall thermal performance. Building and Environment, 67, 1-13.

Yang, L., Chen, Y., Li, Y., Zhu, H., Yang, X., Li, S., & Tang, G., 2024. Is 3D building morphology really related to land surface temperature? Insights from a new homogeneous unit. Building and Environment, 266, 112101.

Urban energy planning relies on digital technologies to support the transition toward more sustainable cities. In this context, Urban Digital Twins (UDTs) are emerging as key tools to manage, visualize, and analyse complex urban data. Many scholars have highlighted the potential of UDTs in energy simulation fields, namely positive energy districts (Coors and Padsala, 2024), household consumption (Padsala et al., 2024), heating demand (Würstle et al., 2020), and greenhouse gas emissions (Alva et al., 2024), across various scales.

This work aligns with the current discourse around developing Energy UDTs by ensuring a direct coupling between an UDT platform (City2Twin) and a Building Energy Simulation (BES) model. This integration generates detailed energy data and enriches 3D city models with domain-specific attributes such as heating demand, supporting decisions related to urban heating and retrofitting planning. The approach uses the geometric data from the UDT to feed an automated, parametric energy model and reinjects the simulation results into the UDT for visualization and analysis.

Most urban-scale energy assessment tools rely on archetypes, degree-day methods, or benchmark data, which often overlook individual building characteristics such as geometry, thermal inertia, or usage patterns. In contrast, this method applies a more detailed, data-based simulation while remaining computationally efficient. This coupling offers a scalable, standardized, and data-driven solution that bridges the gap between digital urban models and building-level thermal simulations. Reintegration into the UDT improves the structuring, accessibility, and interpretation of energy data for planning and policymaking. The workflow is tested on a real urban district to assess its feasibility and potential for broader application.

Thermal imaging provides valuable insights into building performance and urban environmental dynamics, yet integrating thermal data with 3D models remains challenging, particularly with limited equipment. This paper presents a comprehensive workflow from point cloud acquisition to 4D thermal model creation using a novel low-cost mobile triple-camera system in Rabat (Morocco). The methodology leverages a unique configuration with a central thermal camera positioned between two RGB cameras. First, terrestrial laser scanning data is processed to create 3D models suitable for thermal analysis and future microclimate simulations. The exterior orientation of RGB images is determined using Structure from Motion, then transferred to thermal imagery through the fixed geometric relationship between cameras, achieving high spatial accuracy with average deviations of control points measuring only 2.9 cm. Temperature data extracted via the FLIR SDK is projected onto the 3D model using ray-casting, with a weighted integration method resolving overlapping data based on view angle quality. The resulting spatiotemporal model enables analysis of dynamic thermal behavior, revealing how vegetation and architectural elements influence facade temperatures throughout the day. Initial results demonstrate clear temperature distributions corresponding to shading patterns and solar exposure. This framework creates spatially continuous ground truth data for upcoming comparisons with LASER/F (LAtent, SEnsible, Radiation / Fluxes) simulation outputs, establishing a robust validation methodology for urban microclimate modelling.