Forest degradation is a significant yet underestimated source of carbon emissions. Traditionally, monitoring forest degradation has been difficult due to a lack of sufficiently frequent satellite observations and reliable analysis methods. Recent advancements in satellite remote sensing provide new opportunities to monitor, attribute and analyze forest degradation. This dissertation develops methods to monitor and attribute forest degradation and analyzes the spatial-temporal patterns of forest degradation and associated carbon emissions. A new method, Continuous Change Detection and Classification - Spectral Mixture Analysis (CCDC-SMA), was developed on Google Earth Engine (GEE) to monitor abrupt and gradual forest degradation in temperate climate zones using Landsat time series. CCDC-SMA was applied to the Republic of Georgia from 1987-2019. Results show that forest degradation affected a much larger area than deforestation. In addition, CCDC-SMA was extended to monitor forest degradation in the tropics and applied in Laos. Attribution of the drivers of forest degradation was based on a combination of CCDC-SMA results, post-disturbance land cover classification and object-based image analysis. Shifting cultivation is the largest kind of forest disturbance in Laos, affecting 32.9% ± 1.9% of Laos during 1991-2020. The results show that shifting cultivation has been expanding and intensifying in Laos, especially in the last five years. Furthermore, the length of fallow periods has been continuously declining, which indicates that shifting cultivation is becoming increasingly unsustainable. Combining biomass estimates from the Global Ecosystem Dynamics Investigation (GEDI) and area estimates of shifting cultivation, the net carbon emissions from shifting cultivation during 1991-2020 in Laos are 1.28 ± 0.12 petagrams of CO2 equivalent (Pg CO2 eq). Tree canopy height and aboveground biomass density are strongly correlated with the years of regrowth since the latest year of slash-and-burn activities, which can be expressed using logarithmic models. It takes 131 years for the biomass to recover to pre-disturbed levels based on the logarithmic models. In addition to advancements in remote sensing of forest degradation, the results of this dissertation provide valuable information for policy related to forest management and reduction of carbon emissions.
Identifer | oai:union.ndltd.org:bu.edu/oai:open.bu.edu:2144/46368 |
Date | 16 June 2023 |
Creators | Chen, Shijuan |
Contributors | Woodcock, Curtis, Olofsson, Pontus |
Source Sets | Boston University |
Language | en_US |
Detected Language | English |
Type | Thesis/Dissertation |
Rights | Attribution 4.0 International, http://creativecommons.org/licenses/by/4.0/ |
Page generated in 0.0171 seconds