1
Flood forecasting model for Pasak river basin, Thailand | |
Author | Gautam, Mahesh Raj |
Call Number | AIT Thesis no.WM-96-12 |
Subject(s) | Flood forecasting--Pasak River Basin |
Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering. School of Civil Engineering |
Publisher | Asian Institute of Technology |
Series Statement | Thesis ; no. WM-96-12 |
Abstract | Flood forecasting is one of the important non-structural measure of flood controlling activities. For the flood forecasting purpose, the use of observed water level or discharges at real time is very helpful in making the forecast result more accurate. This study is mainly focused in developing suitable models with options of making several time step ahead forecast with adaptive error correction approach. In addition to such model development, the performance evaluation of developed models is also sought for the purpose of finding their suitability for the flood forecasting at prominent streamgauging stations in the Pasak river basin. Considering the present status of available data and the catchment characteristics, appropriate type of models were selected for the flood forecasting purpose at the three major streamgauging stations in the Pasak river, Thailand. Popular lumped conceptual models such as NAM and Tank were used for rainfall-runoff transformation. The output of these models was modeled with suitable ARMA stochastic model for better performance in the real time flood forecasting purpose. Use of relatively new tool of forecasting- Neural Net (NN) model was made with different architectures. A more objective approach of selection of architecture of such NN models was adopted rather than relying solely on prevalent trial and error approach. In addition to the comparison of performances of the models applied, comparison study of NAM and Tank model structures was also made in the present study. The use of Mike 11 Flood forecasting tool was also made in the study with the aim of finding shortcomings that needs to be addressed properly in order to make such hydrodynamic model based study more appropriate. Some general issues and considerations for model selection was also discussed with. The Neural Net based model was found most appropriate choice for the forecasting purpose at locations (such as Lomsak streamgauging station in the present study) where other methods have been found unsuitable. The adaptive error correction approach for real time forecasting was found simple yet functionally efficient. At the Wichianburi streamgauging station both Tank+AR(3) and NAM+ AR(3) combinations were found suitable as compared to NN based models and the model presently on use on the operational flood forecasting purpose by the Meteorological Department of Thailand (MED). However the Tank+ AR(3) combination was slightly preferred than NAM + AR(3) combination. At Buachum streamgauging station, the combination of 3-parameter Muskingum model and AR (1) was found more suited as compared to other models namely neural net models and the model used by MED. Although the performance ofNN based models in overall was found not as good as their counterpart models used in the study, the result of the present study reflects the potential of NN based models as viable alternative for rainfall-runoff transformation and flood forecasting purpose. |
Year | 1996 |
Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. WM-96-12 |
Type | Thesis |
School | School of Civil Engineering |
Department | Department of Civil and Infrastucture Engineering (DCIE) |
Academic Program/FoS | Water Engineering and Management (WM) |
Chairperson(s) | Tawatchai Tingsanchali; |
Examination Committee(s) | Gupta, Ashim Das;Kubo, Naritaka; |
Scholarship Donor(s) | DAAD(Germany); |
Degree | Thesis (M.Eng.) - Asian Institute of Technology, 1996 |