Method to mapping the Amazonian Colombian forest disturbance as a result of selective logging base on satellite image

According to the Forest and Carbon Monitoring System of the Institute of Hydrology, Meteorology and Environmental Studies (IDEAM), 197,159 hectares were deforested in Colombia in 2018. However, small variations in canopy density are difficult to detect and quantify by traditional deforestation monitoring methods. The research sought to develop a method for mapping the disturbance of the Colombian Amazon forest due to selective logging using satellite images. For this, the precision and accuracy of the Random Forest algorithm's classification as applied to two spectral indices, ΔrNBR and the spectral mixture analysis, were evaluated. These two indices were evaluated in the Landsat 2010-2019 time series and with the help of the LandTrendr change detection algorithm, which allows for finding temporal vertices associated with the variation of the indices. The research covers an area of approximately 484,000 km² - equivalent to the size of the Colombian Amazon biome - and uses 30 Landsat images.