Enhancing Urban Climate Modeling through 3D Quantification of Greenspace using LiDAR and Generative AI

Urbanization has led to adverse microenvironment conditions, materialized through the Urban Heat Island (UHI) effect and the deterioration of air quality. This study explores the relationship between urban greenspace and these urban microclimate factors, focusing on the need to accurately quantify the three-dimensional (3D) aspects of urban greenspace to better comprehend its influence. We selected Prospect Park in Brooklyn, New York, for our investigation due to its extensive field-measured tree forestry data. Our goal was to construct a detailed 3D model of tree crowns in Prospect Park using LiDAR (Light Detection and Ranging) technology. To achieve this, geometric-based classification algorithms combined with tree segmentation algorithms were used to segment and delineate trees from the LiDAR capture. Generative A.I. was utilized to reconstruct tree canopies in order to estimate tree crown volume. This improvement enabled us to gain a more precise understanding of vegetation distribution and structure in Prospect Park. Importantly, the developed model can be extended to other urban parks, facilitating a comprehensive assessment of their ecological services. Our approach improved upon traditional satellite remote sensing, which lacks the necessary resolution to describe detailed tree crown structures. Additionally, while field surveys provide valuable data, they can be time-consuming and expensive. By leveraging LiDAR technology and advanced algorithms, our model enhanced the accuracy and efficiency of mapping tree crowns in urban green spaces. This approach provided us with a more precise understanding of the distribution and structure of vegetation within Prospect Park, and also generated a valuable new dataset that can be integrated into air purification and Urban Heat Island (UHI) models. This dataset can inform and improve decisions related to the preservation, management, and future design of urban greenspaces, ensuring their continued enhancement and the optimization.