Accuracy of grid precipitation data for Brazil: application in river discharge modelling of the Tocantins catchment

作者：Monteiro Jose A F^*; Strauch Michael; Srinivasan Raghavan; Abbaspour Karim; Guecker Bjoern

来源：HYDROLOGICAL PROCESSES, 2016, 30(9): 1419-1430.

DOI：10.1002/hyp.10708

• CFSR
• WFDEI
• precipitation
• watershed modelling
• Brazil
• tropics

Here, we compared grid precipitation data Climate Forecast System Reanalysis (CFSR) and WATCH Forcing Data methodology applied to ERA-Interim (WFDEI) data with Brazilian Weather Bureau (INMET) and Brazilian Water Agency (ANA) rain gauge data (n=2027) for the period 1980-2010 in order to evaluate which grid data set better represents precipitation, and is thus more suitable for hydrological modelling of Brazilian water resources. We found that WFDEI outperformed CFSR according to three statistical indicators. We then applied and interpolated a simple bias correction to further improve WFDEI data before we used these data to model river discharge of the Tocantins catchment with the Soil and Water Assessment Tool (SWAT). Calibration (validation in parentheses; weighted averages of all gauges) had satisfactory statistical metrics: p-factor=0.52 (0.47); r-factor=0.84 (0.99); R-2=0.78 (0.71); bR(2)=0.68 (0.47); NS=0.70 (0.66); Pbias=-4.5 (4.0). Finally, the calibrated SWAT model was used to assess the spatial distribution of the catchment's water resources. Annual green water flow (evapotranspiration) increased from the south-east (640-840mmyr(-1)) to north-west (1140-1440mm) of the Tocantins catchment, while green water storage (soil water content) increased from south (330-1070mm) to north (2180-3290mm). Blue water (water yield) had a less clear pattern, with lower values in the south and the central borders of the catchment (20-560mm) and higher values along the central axis and the north (920-1460mm). Our analysis suggested that WFDEI was an accurate representation of Brazilian precipitation. For large catchments, we therefore recommend the use of WFDEI instead of sparse and often missing rain gauge data in modelling Brazilian water resources.

• 出版日期2016-4-30