The use of global datasets in the SWAT model for tropical watershed with limited ground data: A case study in serayu upper catchment

Global datasets, such as SOILGRIDS and global weather are more accessible and precise. Recently, the use of global datasets has been increasingly common in hydrological research. This study was designed to evaluate the use of global datasets (global weather and SOILGRIDS) as an alternative input to support the lack of ground measurement station and ground data in a tropical watershed, Upper Serayu Watershed. A hydrological model, SWAT, was used to run a daily hydrological model. The results showed that the combination of Global Weather and SOILGRIDS as the model input produced RRMSE = 0.70, R2 = 0.7, and PBIAS = 15.5. These figures suggest that this combination generates a model with excellent performance, except for NSE = 0.45 that indicates nearly satisfactory performance. The model created using ground measurement data showed a better performance, with NSE = 0.80, RRMSE = 0.42, R2 = 0.82, and PBIAS = 9.7. In conclusion, global datasets create a hydrological model with slightly lower performance than the one generated by ground measurement. Nevertheless, the former has the potential as an alternative to the latter, especially in the circumstance of missing ground data. Based on the 2009-2013 simulation, the modeled flow discharges were close to the measured ones, suggesting that datasets from Global Weather and SOILGRIDS can adequately substitute for ground measurement data in estimating the streamflow of a tropical watershed, that is, Upper Serayu Watershed. © 2020, Taiwan Joint Irrigation Associations.