Evaluation of gridded meteorological datasets and their potential hydrological application to a humid area with scarce data for Pirapama River basin, northeastern Brazil

This work evaluated the simulation of streamflow using observed and estimated gridded meteorological datasets and the Soil and Water Assessment Tool (SWAT) model for a humid area with scarce data in northeastern Brazil. The coefficient of determination (R-2), Nash-Sutcliffe efficiency (NS), root mean square error (RMSE), normalized root mean square error (NRMSE), and percent bias (PBIAS) were used to assess the SWAT results yielded by estimated and observed rainfall data. The hydrological modeling data from three streamflow stations were used (2000 to 2006 for calibration and 2007 to 2010 for validation). The results show that at daily scale, the estimated rainfall data show a poor agreement (R-2 ranging from 0.22 to 0.04) with the observed rainfall but good agreement at monthly (R-2 = 0.85) and annual scales (R-2 = 0.80). The results showed that estimated accumulated precipitation overestimated the observed data. The results showed that R-2 ranged from 0.51 to 0.55 at monthly scale and 0.44-0.52 at annual scale. However, the global data can represent well the variability of rainfall within the region. The results indicated a good correlation in the seasonal variability (R-2 ranged from 0.72 to 0.60). The modeling results using monthly TRMM data and observed rainfall data showed good values of NS and R-2 during calibration and validation, but PBIAS was unsatisfactory for the three streamflow gauges. The streamflow estimates from the SWAT model using data from the TRMM satellite showed that such data are capable of generating satisfactory results after calibration, although measured rainfall data presented better results; the data could support areas with scarce rainfall data and be applied to other river basins, for example, to analyze the hydrological potential of other basins in the coastal region of northeastern Brazil. Over the past three decades, considerable advances have been made in remote sensing with environmental satellites, increasing the amount of information available, including rainfall estimates. In this context, the use of TRMM data to estimate rainfall has ultimately been shown to be an interesting alternative for areas with scarce rainfall data.