Open Collections

Kadel, Abhinab

University of British Columbia

This thesis examines the quality of GESS streamflow forecasts over the mountainous terrains of Nepal for two years, 2014 and 2015, focusing on run-of-river (ROR) hydropower operation. A reforecast dataset is used for the forecasts and is compared with streamflow observations at five different sites with existing hydropower facilities. The forecasts are verified using verification metrics such as bias, flow variability, correlation, Kling-Gupta efficiency, the Nash-Sutcliffe efficiency, and the Continuous Ranked Probability Score. The verification is performed across two flow seasons: wet and dry, distinguished by the 70th percentile of climatological flow. First, the raw forecasts are verified. The results show an overall poor performance of the forecasts. Second, a simple moving window multiplicative bias correction approach called the Degree of Mass Balance (DMB) is tested. The 2014 year is set aside as the calibration year to calculate the best bias correction approach, such as the window length and the best DMB formulation. The best DMB configuration for each site and forecast horizon is then tested in the independent verification year of 2015. The bias-corrected forecasts show much-improved performance in all metrics. Finally, the bias-corrected GESS forecasts are evaluated for two use-cases commonly faced by ROR hydropower operators in Nepal: flood forecasting in the wet season and energy generation forecasting in the dry season. The GESS forecasts raised more false alarms and would not have predicted at least half of the flood events in the sites studied. Furthermore, the forecasts did not yield more revenue than a simple persistence forecast. Thus, there is a need to improve the forecasts before they can add real value to the ROR operators.

Text

2022-04-14

Attribution-NoDerivatives 4.0 International

http://hdl.handle.net/2429/81217

Atmospheric Science

University of British Columbia

UBCV

http://creativecommons.org/licenses/by-nd/4.0/