University of Twente Student Theses


Monitoring Spatio-temporal Dynamics of land cover changes for Bolivian land tenure reform using MODIS remote sensing images

Argandona Aramayo, Pablo (2010) Monitoring Spatio-temporal Dynamics of land cover changes for Bolivian land tenure reform using MODIS remote sensing images.

[img] PDF
Abstract:Land cover and land use changes have a direct influence over ecosystems and socioeconomic structures of regions. Some of these changes are decided and administered by local governments in a planned and controlled manner. However many of the changes is the result of one-sided and improper decision making, which directly affects local people, who start to notice important changes in land use types in their communities. In due course, various natural and cultural assets become disturbed in a non-sustainable manner. Many developing countries are grappling with how to reconcile the three objectives of increasing agricultural production, reducing poverty and using natural resources sustainably. In this context, the Bolivian government is implementing a reform of the agrarian system in Bolivia, based on the principle that everyone has the right to private property, as long as it is a “productive land” (LeyN°3545 2006). The general objective of this study is to monitor land cover patterns in Alto Parapeti by analyzing time series of MODIS imagery to determine productive and unproductive lands from the Bolivian land tenure reform perspective. The spatial and temporal MODIS NDVI patterns, in a remote sensing perspective, have been analyzed in the area of Alto Parapeti, part of the Chaco region in Bolivia. The patterns have been explored by applying a decision tree classification for land cover mapping. The land cover classes identified in the study area are dense forest, crop land, dense shrub land, mixed vegetation/shrub dominant and water bodies. The performance of the classification method was evaluated in terms of accuracy for the land cover map of 2009. Land cover maps from 2001 to 2009 were generated using the decision tree classification, but further work is needed to validate this method with more dates, and later test this approach to determine the sensitivity to inter- annual variability over the classification results of multiple years. The accuracy assessment of the land cover map 2009 was 76% overall accuracy. The performance of the decision tree classification was successfully at mapping extensive cover lands such as dense forest and mixed vegetation /shrub dominant, although was some miss-classification of dense shrub land in the transition to dense forest. The approach was far less effective at mapping smaller cover types as the small as fragmented cropland areas. The decision tree approach was compared in terms of accuracy with maximum likelihood supervised classification and isodata classification methods. The approach by decision tree provided the highest overall accuracy of three methods. The land cover classes were allocated to the “productive land” categories producing a “productive land” map Alto Parapeti from 2009. The map showed a large extension of unproductive land. This approach and can be used as a reference to prioritize the efforts of the Bolivian Agrarian Reform finding extensive unproductive land to start field verification.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page