Remote Sensing, Machine Learning, and Image Processing for Environmental Conservation
摘要
In this project, the carbon dioxide retained in wooded areas dominated by Pinus halepensis Mill. in the Valencian Community has been calculated. These areas represent 62% of the wooded masses in the territory of the Valencian Community, and the calculation of carbon dioxide retained has been made possible through satellite image processing, remote sensing, and the use of automatic learning. Starting from data from a forest inventory and Sentinel-2 images of a small area, a training database has been developed to calculate the amount of biomass for each pixel. From this data, the conversion from Biomass to Carbon has been performed, and by applying the stoichiometric relationship between Carbon and carbon dioxide, the CO2 retained in these masses can be obtained. Thanks to the process, the biomass of the studied areas (21 million tons), the Carbon Stock (10 million tons), and the retained CO2 (39 million tons) have been quantified, as well as the creation of biomass and Carbon Stock maps that provide a tool not only visual but also helpful for environmental management.