| Popis: |
The river discharge data is one of the most important pieces of information to regulate various water resources, including flood frequency analysis, drought and flood prediction, etc. The missing observer discharge data, even a short gap, influences the whole analysis and gives a totally different result. Filling data gaps in streamflow data is thus a critical step in any hydrological study. Interpolation, regression-based analysis, artificial neural networks, and modeling are all methods for generating missing data. While using the hydrological model to generate the data, we first need to calibrate the hydrological model. The single-site calibration of the hydrological model has its own limitations, due to which it does not correctly predict the streamflow at intermediate gauge locations. This is because, while calibrating the model for the final outlet, we tune the parameters that affect the results for the final outlet only and neglect the intermediate sites' output. In this study, we demonstrate the importance of multi-site calibration and use the calibrated hydrological model to generate the missing data at intermediate sites.For this study, we selected the Godavari River basin and calibrated it at the final outlet (single-site calibration) and at 18 + 1 outlets (multi-site calibration). The whole basin is divided into 103 subbasins, and the Soil and Water Assessment Tool (SWAT) hydrological model is used for this study. After the successful multi-site calibration, we generated the missing data at 25 different gauging locations. The initial results from single-site calibration (NSE (0.57) and R2 (0.61)) show good agreement between observed and simulated discharge for the final outlet. The multi-site calibration analysis is in progress, and full results will be presented at the conference. |
