Multi-scale Data Framework for the Monitoring of Carbon Stock Changes in the 21st Century
Abstract
Climate change mitigation through the reduction of emissions from deforestation and degradation (REDD+) has presented a great potential for many countries with large areas of forests. However, estimates of carbon stocks and emissions that can be used for implementing climate mitigation at local to jurisdictional scales are missing. Starting from early 2000s, the number of satellite observations of forest cover and structure and in situ forest inventory data have increased exponentially in anticipation of global climate mitigation and the Paris Agreement. A unified framework is therefore needed to synthesize the information from various sources and to deliver accurate estimations of carbon stocks and changes across global vegetation. Here we integrate the ground inventory, airborne, and satellite lidar (GLAS, ICESAT-2, and GEDI) data to construct the multi-scale training data of live vegetation carbon globally from 2001 to 2020. Using multi-temporal series of satellite data in the optical and microwave domains, we build wall-to-wall carbon stocks maps annually based on the multi-scale deep learning model that takes input environmental/satellite layers at multiple spatial and temporal resolutions and output carbon estimates for multiple scales. We combine carbon stocks with satellite observations of forest cover change from land use activities (deforestation, degradation, fire) and provide estimates of the carbon loss and gain spatially and temporally across the global vegetation. Results from our study show emerging patterns of vegetation carbon changes in the 21th century. This work has been performed at the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFM.B25G1540Y