1 AIT Asian Institute of Technology

Comparative performance assessment of selected models for hydrological forecasting : a case study of Karnali River Basin, Nepal

AuthorGhimire, Suwas
Call NumberAIT Thesis no.WM-19-03
Subject(s)Hydrological forecasting--Nepal--Karnali River Basin
Flood forecasting--Nepal--Karnali River Basin

NoteA thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Water Engineering and Management
PublisherAsian Institute of Technology
AbstractHydrological model is one of the major components of flood forecasting and warning system. Selection of a suitable model has always been one of the challenges in developing a reliable flood forecasting system. The tradeoff between model performance and the complexity of the model has remained as an unexplored area of research. In such scenario insights on the performance of models and its variation in accordance with model complexity is felt. This study focuses on the comparative performance assessment of selected models for hydrological forecasting with a case study of the Karnali river basin, Nepal. During the study, thirteen models have been reviewed so far. They were arranged and classified into four groups based upon complexity. The complexity of the model was defined based upon a number of parameters, spatial resolution, data requirement and type of model. One representative model was selected from each group based upon model selection criteria. IFAS, HEC-HMS, HBV and GR4J were selected models in the order of complex to simple. They were calibrated (16yrs, 1999-2014) and validated (5yrs, 1994-1998) with a warm-up period of two years (1992-1993). Thus, calibrated models were used for hydrological forecasting using forecasted rainfall data up to 3 days lead time. Results from the hydrological forecast were evaluated based on their distribution and performance metrics at different flow levels. Finally, models were compared along with their ensemble to find out the best alternative. HEC-HMS produced the best result at all lead time in terms of all performance metrics except AUC. HBV, IFAS and GR4J followed HEC-HMS in terms of their performance for 1 day lead time. At no time, result from the ensemble was better than the best model. Results show that increasing model complexity doesn’t necessarily improve the performance of the model. Similarly, an ensemble of hydrological models doesn’t necessarily improve the performance of hydrological forecasting. However, it could improve the distribution of the forecasted flow.
Year2019
TypeThesis
SchoolSchool of Engineering and Technology (SET)
DepartmentDepartment of Civil and Infrastucture Engineering (DCIE)
Academic Program/FoSWater Engineering and Management (WM)
Chairperson(s)Babel, Mukand Singh
Examination Committee(s)Shrestha, Sangam;Sarawut Ninsawat;Agarwal, Anshul;
Scholarship Donor(s)Water Resources University Capacity Building (WRU CB) Project;Asian Institute of Technology Fellowship ;
DegreeThesis (M.Eng.) - Asian Institute of Technology, 2019


Usage Metrics
View Detail0
Read PDF0
Download PDF0