Hydrological modeling in an agricultural basin in the Brazilian Cerrado using satellite precipitation data

Abstract

Precipitation data are one of the main inputs of a hydrological model and, therefore, the low density of rain gauges and large amounts of missing data in certain regions are obstacles to the development of hydrological research. In this sense, the quality of satellite precipitation products (SPP) was evaluated by comparing them with conventional data obtained from rain gauges (OP); subsequently the Soil and Water Assessment Tool (SWAT) model was calibrated and validated for three distinct periods according to the change in land use. Periods 1 and 2 were calibrated considering only the observed precipitation data, while period 3 included the addition of precipitation data from the SPP - Tropical Rainfall Measuring Mission - satellite due to the many years with information missing. Finally, the generated flows were examined, discussing the feasibility of using alternative precipitation data for hydrological simulation in a basin with a scarcity of data. The SPP showed a strong correlation with the OP for the hydrographic basin under study according to the statistical metrics analyzed, despite overestimation of figures for wet periods. The OPs satisfactorily simulated the flow rates of the basin under study for the two periods of analysis, as well as period 3, when the SPP was added. However, periods 1 and 2, which only contained information from rain gauge observations, presented better statistics than period 3. The values of the statistical coefficients attested to the potential of using the satellite data, in addition to observed data, to enable the calibration and validation processes of the SWAT model in regions with low station density or missing data.