CLASSIFICATION SYSTEM DRIVES DISAGREEMENT AMONG BRAZILIAN VEGETATION MAPS AT A SAMPLE AREA OF THE SEMIARID CAATINGA

Our work addresses a key challenge in earth system modelling: how to optimally exploit the information contained in satellite remote sensing observations in the calibration of such models. For this we thoroughly test a number of measures that quantify the fit between an observed and a simulated spatial pattern. We acknowledge the difficulties associated with such a comparison and suggest using measures that regard multiple aspects of spatial information, i.e. magnitude and variability.

Satellite data offer great opportunities to improve spatial model predictions by means of spatially oriented model evaluations. In this study, satellite images are used to observe spatial patterns of evapotranspiration at the land surface. These spatial patterns are utilized in combination with streamflow observations in a model calibration framework including a novel spatial performance metric tailored to target the spatial pattern performance of a catchment-scale hydrological model.

This paper investigates the skill of 90-day low-flow forecasts using three models. From the results, it appears that all models are prone to over-predict runoff during low-flow periods using ensemble seasonal meteorological forcing. The largest range for 90-day low-flow forecasts is found for the GR4J model. Overall, the uncertainty from ensemble P forecasts has a larger effect on seasonal low-flow forecasts than the uncertainty from ensemble PET forecasts and initial model conditions.