Santos, Fabián: A Landsat-based analysis of tropical forest dynamics in the Central Ecuadorian Amazon : Patterns and causes of deforestation and reforestation. - Bonn, 2018. - Dissertation, Rheinische Friedrich-Wilhelms-Universität Bonn.
Online-Ausgabe in bonndoc:
author = {{Fabián Santos}},
title = {A Landsat-based analysis of tropical forest dynamics in the Central Ecuadorian Amazon : Patterns and causes of deforestation and reforestation},
school = {Rheinische Friedrich-Wilhelms-Universität Bonn},
year = 2018,
month = dec,

note = {Tropical deforestation constitutes a major threat to the Amazon rainforest. Monitoring forest dynamics is therefore necessary for sustainable management of forest resources in this region. However, cloudiness results in scarce good quality satellite observations, and is therefore a major challenge for monitoring deforestation and for detecting subtle processes such as reforestation. Furthermore, varying human pressure highlights the importance of understanding the underlying forces behind these processes at multiple scales but also from an interand transdisciplinary perspective. Against this background, this study analyzes and recommends different methodologies for accomplishing these goals, exemplifying their use with Landsat timeseries and socioeconomic data. The study cases were located in the Central Ecuadorian Amazon (CEA), an area characterized by different deforestation and reforestation processes and socioeconomic and landscape settings. Three objectives guided this research. First, processing and timeseries analysis algorithms for forest dynamics monitoring in areas with limited Landsat data were evaluated, using an innovative approach based in genetic algorithms. Second, a methodology based in image compositing, multisensor data fusion and postclassification change detection is proposed to address the limitations observed in forest dynamics monitoring with timeseries analysis algorithms. Third, the evaluation of the underlying driving forces of deforestation and reforestation in the CEA are conducted using a novel modelling technique called geographically weight ridge regression for improving processing and analysis of socioeconomic data. The methodology for forest dynamics monitoring demonstrates that despite abundant data gaps in the Landsat archive for the CEA, historical patterns of deforestation and reforestation can still be reported biennially with overall accuracies above 70%. Furthermore, the improved methodology for analyzing underlying driving forces of forest dynamics identified local drivers and specific socioeconomic settings that improved the explanations for the high deforestation and reforestation rates in the CEA. The results indicate that the proposed methodologies are an alternative for monitoring and analyzing forest dynamics, particularly in areas where data scarcity and landscape complexity require approaches that are more specialized.},
url = {}

The following license files are associated with this item: