Use of airborne and mobile terrestrial LiDAR data in the quantitative inventory of urban trees
Name: EMERSON EDUARDO OLIVEIRA DE SOUZA
Publication date: 30/07/2025
Examining board:
Name |
Role |
|---|---|
| ADRIANO RIBEIRO DE MENDONCA | Presidente |
| ANDRÉ QUINTÃO DE ALMEIDA | Examinador Externo |
| CRISTIANE COELHO DE MOURA | Examinador Interno |
| GILSON FERNANDES DA SILVA | Examinador Interno |
Summary: Urban tree planting plays an essential role in providing ecosystem services such as thermal regulation, surface runoff control, and air quality improvement. However, proper management of these trees depends on accurate, up-to-date, and efficient inventories. Given the operational limitations of conventional methods, this study evaluated the use of airborne LiDAR (ALS) and terrestrial LiDAR (TLS) data for tree detection and the estimation of biometric variables in an urban environment. The main objective was to analyze the accuracy of these technologies in estimating variables such as total height (H), diameter at breast height (DBH), and crown diameter (dc) along an urban street in the municipality of Jerônimo Monteiro, ES, Brazil. The methodology involved acquiring data with LiDAR sensors mounted on a remotely piloted aircraft (RPA) and on mobile terrestrial scanning equipment, in addition to traditional field inventory. Point clouds were pre-processed, classified, and normalized. Subsequently, digital terrain models (DTMs) were generated, individual trees were detected and segmented (ITD), structural metrics were extracted, and multiple linear regression models were fitted to estimate the variables of interest. The results showed that ALS presented higher accuracy in total height (H) estimation, with an adjusted R² of 0.95 and RMSE of 6.69%. On the other hand, TLS performed better in the estimation of DBH (adjusted R² of 0.47 and RMSE of 26.21%) and cd (adjusted R² of 0.55 and RMSE of 19.67%), providing better detail of the trees’ lateral structure. The best-performing detection algorithm was the Local Maximum Filter (LMF) with a variable linear window, especially when applied directly to the TLS point cloud. Statistical modeling using point cloud-derived variables showed robust performance, particularly with metrics such as zq95, zkurt, and ikurt. It is concluded that both ALS and TLS are effective tools for urban forest inventory, with complementary potential. The combination of ALS’s spatial coverage and TLS’s structural detail can optimize urban planning and tree management, contributing to more efficient strategies for monitoring and managing urban green areas.
Keywords: Urban afforestation, improved forest inventory, remote sensing.
