Forest disturbance assessment with Landsat data in Middle Volga region

摘要

The study presents the results of multistage of classification using supervised algorithm of Maximum Likelihood to assessment the forest using of multitemporal Landsat (summer-autumn) images for mapping cover disturbance in Middle Volga region. Analysis of the degree of forest cover disturbance was determined based on by biotic (fires, thematic map of the state of forest cover) and complex factors (forest disturbance map using the Disturbance Index) for 1985-2014. Calculation of the influence degree ("weight index") of each factor on the disturbance of forest cover was made. The integrated disturbance map was obtained based on a bivariate statistical analysis of raster data. The algorithm of geospatial analysis of thematic raster data allowed determining the degree of correlation between the indicators of the revealed violations of forest cover in the Middle Volga region and various factors affecting these violations. For classification and validation of thematic maps of forest cover disturbance was used test sites with independent field data for each satellite Landsat image of middle spatial resolution. Cross-validated overall accuracies, ranged between 81-87%, showed high degree of agreement between the forest cover disturbance map and the ground truth data. Research results can be applied for the long term forest monitoring and development of forest management plan for the investigated region.