Mapping fine-scale carbon sequestration benefits and landscape spatial drivers of urban parks using high-resolution UAV data

Urban parks offer dual ecological benefits by increasing carbon sinks and reducing carbon emissions and are considered an important natural strategy for cities to achieve "double carbon" goals. However, rapid, efficient, and accurate quantification of the carbon sequestration benefits of urban parks poses a significant challenge. To address this, we utilized unmanned aerial vehicle (UAV) multispectral data to construct large-area, high-resolution models of urban park carbon sequestration benefits, replacing traditional, time-consuming, and laborious large-scale field surveys. Additionally, we explored the relationship between urban park landscape patterns and the benefits of carbon sequestration. First, we used data from 12 tree species to calculate the carbon storage based on tree species and compared these results with those calculated based on forest type. Second, three prediction models were constructed using multispectral vegetation index only, texture features only, and a combination of both, in conjunction with gradient boosting decision Trees (GBDT), random forest (RF), and backpropagation (BP) neural network to generate carbon sequestration benefit maps for the entire park. These maps allowed us to determine how variations in urban park landscape structures affect carbon stocks. The results show that UAV multispectral imagery provides a fast and accurate method for measuring the carbon sequestration benefits of urban parks and offers an alternative method for generating carbon sequestration benefit maps. This research reveals the benefits of urban park carbon sequestration and explores the spatial patterns within landscapes. The findings are of great significance for guiding the estimation of urban carbon sequestration benefits and achieving carbon neutrality.