REMOTE SENSING IMAGERY DETECTS HYDROMORPHIC SOILS HIDDEN UNDER AGRICULTURE SYSTEM

The pressure for food production has expanded agriculture
frontiers, posing a threat to water resources. For instance,
placing crop systems over hydromorphic soils (HS), have a
direct impact on groundwater and influence the recharge of
riverine ecosystems. We applied a temporal remote sensing
strategy (Synthetic Soil Image - SySI) associated with random
forest (RF) to map HS in an 735,953.8 km² area in Brazil.
Slope and SySI contributed the most for the prediction model
using RF with cross validation (accuracy of 0.92). The assessments showed that 14.5% of the study area represented
HS, mostly located inside agricultural areas. Soybean and
pasture areas had up to 14.9% while sugar cane had just 3%.
Here we present an advanced remote sensing technique that
may improve the identification of HS under agriculture and
assist public policies for their conservation.