First-order verification of hydrologic model predictions for the Brazilian Cerrado

The InVEST hydrologic models were designed to represent the dominant processes of sediment/nutrient delivery and retention by land use/ land cover (LULC). Since they are not standard rainfall-runoff models, classical calibration and testing methods are not always applicable. Testing and calibration were particularly difficult in our study region due to: i) the absence of local data (e.g. in-stream sediment or nutrient concentration time series, contributions from point sources) to calibrate the model or a more sophisticated model, and ii) the existing uncertainties in nutrient/sediment dynamics in this region of Brazil (Thomas et al., 2004; Moreno-seceña et al., 2011; Filoso et al., 2006; Maillard & Santos, 2008) (e.g. magnitude of nutrient leaching from sugar cane fields, effect of cattle ranching).