Analysis of forest cover changes in Vietnam using hierarchical object-based classification: Challenges and opportunities

Abstract

Over the last three decades, land use/cover changes has been especially dynamics in South East Asia. Most countries continue to report the loss of forest but a few countries gain more forest area lately, including Vietnam. Different definition of forest, despite of the political nuance, has different consequences to environmental services. In quantifying changes and trajectories of forest and other land uses, we need to be clear of the forest definition we adopt and its implications, since the conclusion is critical in determining sustainable landscape management. We developed a hierarchical object-based classification scheme to produce wall-to-wall land use/cover maps of Vietnam from Landsat imageries covering twenty years period of three time steps: 1990, 2000, 2010. Spectral and spatial-based rulesets are combined in the classification algorithm. Post-classification analysis is conducted to study the land cover changes. Comparisons to the existing official maps are being made. Based on classification scheme, we tried to define the term forest into six different groups based on the amount of tree cover in the landscape. The analysis showed that with different forest definition, the trend of forest cover loss in Vietnam can also be observed differently. If forest is only defined as natural forest and logged over forest, clearly there are still loss of forest cover in the period of 2000-2010. However if the definition of forest includes forest plantation, agroforest, and monoculture plantation, forest cover re-growth in the period of 2000-2010 is obvious. Copyright© (2013) by the Asian Association on Remote Sensing