Probabilistic real-time hydrological forecasting in the Upper Chao Phraya River, Thailand | |
| Author | Acharya, Suwash Chandra |
| Call Number | AIT Thesis no.WM-15-07 |
| Subject(s) | Hydrological forecasting--Thailand--Upper Chao Phraya River Hydrologic models--Thailand--Upper Chao Phraya River |
| Note | A thesis submitted in partial fulfillment of the requirements for the Degree of Master of Engineering in Water Engineering and Management |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. WM-15-07 |
| Abstract | The estimation of predictive uncertainty in hydrological forecasting is essential for risk based decision making in flood warning. The application of un certainty as a post - processor of hydrological model output, such as water level, can provide additional information useful for short - term hydrological forecasting. Quantile regression and its different configurations as well as multivariate Gaussian distribution with various predictors has been tested to quantify the uncertainty in hydrological forecasting . This study first analyses three different configurations of quantile regression and a bivariate probability model which share s imilarity in datasets used. The forecast error and forecast water level were used as predictand and predictor respectively to dev elop the following probabilistic models: (i) Linear Quantile Regression, (ii) Linear Quantile Regression in Gaussian space using N ormal Q uantile T ransformation , (iii) Weighted Linear Quantile Regression, and (iv)Bivariate Normal Distribution. These four di fferent models were developed and tested for the operational flood forecasting system in Chao Phraya River Basin, Thailand. The quality of forecasts in terms of reliability, sharpness and overall skill were assessed using various graphical and numerical ve rification metrics. The results from the verification of forecasts did not show the improvement of forecast in both reliability and sharpness simultaneously. The estimation uncertainty in rainfall forecasts, which has major influence on hydrological forecasts, can help in generating probabilistic hydrological forecasts. Further, the study focuses on the uncertainty associated with the rainfall forecasts from WRF forecasting. A combination of log - normal, logistic regression and bivariate normal distrib ution was used to describe the relationship between WRF forecasts and observed rainfall. The conditional distribution was generated and the ensembles were generated using Latin Hypercube Sampling method. The verification of ensemble forecasts were done by reliability plot and ROC curve which shows positive skill for the generated ensembles. The use of selected quantiles from rainfall ensembles showed the overestimation when higher quantiles were used. |
| Year | 2015 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. WM-15-07 |
| Type | Thesis |
| School | School of Engineering and Technology |
| Department | Department of Civil and Infrastucture Engineering (DCIE) |
| Academic Program/FoS | Water Engineering and Management (WM) |
| Chairperson(s) | Babel, Mukand Singh |
| Examination Committee(s) | Shrestha, Sangam;Madsen, Henrik;Piyamarn Sisomphon |
| Scholarship Donor(s) | HM King’s Scholarship |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2015 |