Disturbance Detection and Assessment with MODIS Imagery in Boreal Forest Ecosystem

摘要

The forests in eastern Siberia and Northeastern China have been undergoing dramatic changes during the last several decades due to forest fire, insect infestation, massive logging, agricultural conversion, and afforestation. Forest disturbances change the forest species composition, spatial structure, biomass level and productivity. These changes and the following recovery processes leave these disturbed areas in various successional stages. Identification of disturbed areas and assessment of their successional condition are the most important tasks in forest dynamic monitoring. In this paper, the land cover changes caused by fire or logging in forest areas were investigated using Moderate Resolution Imaging Spectroradiometer (MODIS) Vegetation Indices products (MOD13A1) data. The basic assumption is that the disturbances will cause significant changes in MODIS signature between two years during the summer. The data quality flags for NDVI and EVI were used to exclude pixels affected by clouds. We compared the spectral variations between disturbed and non-disturbed regions and the best spectral variables to describe the forest canopy damages and its response direction were determined. ISODATA classifier method was used to discriminate the degree of changes.