Land degradation mapping in the MATOPIBA region (Brazil) using remote sensing data and decision-tree analysis

The expansion of agricultural activities in Brazil resulted in a loss of approximately 50% of the native vegetation of the Cerrado biome, mainly to the north of the biome in a region known as MATOPIBA. Besides contributing to accelerate climate change, the removal of native vegetation is associated with increase soil degradation. In this study, we used the decision tree method to predict degradation using as indicators Normalized Difference Vegetation Index (NDVI) derived indices over the period 1985–2018. Results showed that approximately 4677 km2 (0.63%) of MATOPIBA region is under strong signs of degradation, mostly on pasture and grasslands areas, commonly used for grazing. Although the percentage of degraded areas is relatively small, the mapping indicated that the majority of degradation patches are concentrated not only in areas of low resilience, e.g., areas dominated by weakly structured sandy soils, but also where highly stable latosols dominate. More importantly, these areas were frequently affected by fire over the study period. This finding suggests that the combination of unsustainable traditional land management practices, such as fire, the replacement of forest by pasture, and anthropic activities such as inadequate road planning, are the main factor that trigger degradation processes in this Brazilian region. The study results are important for decision makers for defining investment priorities for conservation and for the recovery of degraded areas. [Display omitted] •Land physical characteristics do not necessarily imply soil degradation.•Environmental conditions associated with land management trigger the process.•86% of the areas with high susceptibility are composed of grassland vegetation.•In the study region, 4677 km2 (0.63%) is under degradation.