Urban Tree Mapping using airborne LiDAR: Analysing vegetation changes between 2010, 2017, and 2022 A Gothenburg case study

Abstract

In the years ahead, the escalation of global climate change underscores the crucial need for greater understanding and initiative-taking measures to address its far-reaching effects. In urban settings, climate change is already reshaping daily life, with urban vegetation becoming a vital tool in reducing its adverse effects. In urban environments, the struggle between vegetated and built-up areas intensifies, resulting in concerns about the scarcity of green infrastructure and the reduction of essential ecosystem services. This thesis investigates the changes in high vegetation in the municipality of Gothenburg, focusing on vegetation taller than 2.5 meters, using high-resolution airborne Light Detection and Ranging (LiDAR) data. The main purpose was to detect these changes and evaluate the effectiveness of airborne LiDAR datasets and methods in mapping vegetation changes over time. The method of this thesis utilizes Python to process high-resolution 3D LiDAR data from the years 2010, 2017, and 2022 to map vegetation in Gothenburg. The findings indicate a net increase in high vegetation covering an area of +14.02 km2 (179 743 trees with a crown diameter of 10 m) over the specified period. There has been a loss in -9.17 km2 or 117 178 trees, where a significant portion of this loss has occurred in the outskirts of the municipality. This decline can be attributed to changes in urban development, characterized by the expansion of the city centre and the consequent increase in housing requirements, which leads to loss of vegetation. The increase in vegetation can be partly attributed to longer growing seasons due to climate change. However, it is also significantly affected by the complex classification process, leading to numerous errors. Attempts to mitigate these errors through various filtering processes have been unsuccessful due to the dynamic and intricate urban environment. The method employed requires additional optimization for application in urban areas. The primary objective was to develop an automated filtering system for vegetation classification in urban areas. However, extensive manual filtering was necessary, which proved inefficient for such a large study area.