Methodology for Monitoring Natural Vegetation Loss in the Pantanal Biome Using Medium Spatial Resolution Imagery

The Pantanal features complex vegetation in terms of physiognomy; its original cover is increasingly replaced by agricultural and livestock land use. Given its importance and the resulting impacts on biodiversity and the physical environment, the biome was included in the Brazilian deforestation monitoring project. Accordingly, the objective is to develop a methodology and map native vegetation loss within the biome to generate strategic information for monitoring purposes. Medium-spatial-resolution imagery from the Landsat satellite series, specifically the TM (Landsat 5), ETM+ (Landsat 7), and OLI (Landsat 8) sensors were used, covering the period from 2000 to 2021. Using Terra Amazon software, the images were enhanced, and a region-based segmentation algorithm was applied. The hybrid interpretation (combining digital and visual methods) classified only instances of natural vegetation loss, regardless of subsequent land use. A specific interpretation key was developed based on standard image interpretation elements. Accuracy assessment employed a stratified random sampling approach based on the adopted categories (Natural Vegetation and Vegetation Loss by year). The study produced a time series of Vegetation Loss from 2000 to 2021, achieving an overall accuracy of 94.5% (±1.21%). The producer’s and user’s accuracy for natural vegetation are 98.6% (±0.38%) and 97.3% (±1.40%), respectively. The methodological distinction lies in the use and integration of modern technologies (drones, satellite imagery, GIS, WebGIS, and geospatial data platforms), the identification and analysis of regional dynamics, the hybrid interpretation method (manual + digital), and field verifications adapting traditional PRODES deforestation monitoring protocols to overcome the unique hydrological, seasonal, and structural idiosyncrasies of the Pantanal. Regarding regional dynamics, the analysis considers landscape changes driven by wet and dry seasons, land management practices (such as native pasture clearing, cattle grazing, and the use of fire), the replacement of native grasslands with exotic pastures, and the classification of wetlands (temporary ponds, floodplains, and drainage channels) within the mapped Vegetation Loss polygons.