Downscaling precipitation using an artificial neural network in the Ping River Basin, Thailand | |
| Author | Roy, Juthika |
| Call Number | AIT Thesis no.WM-16-34 |
| Subject(s) | Climatic changes--Thailand--Ping River Basin Neural networks (Computer science) |
| 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-16-34 |
| Abstract | Thailand’s rainfall is mainly controlled by the monsoons and the variations in the local rainfall pattern are linked with climate condition. According to IPCC (2007), the increasing emission of GHG s brings changes in the surface temperature affecting atmospheric circulation pattern and influences precipitation. This continued emission will extend the changes of global climate in future by the same token. Due to significant changes in the precipitation pattern in past decades, it is necessary to analyze the impact of global climate change on water resource system by downscaling coarse scale climate variables into regional scale. In this present study, six downscaling models for Ping River basin are developed using Artificial Neural Network which can capture the non - linear relationship between predictand and predictors. In this study, 30 large scale atmospheric variables are taken from NCEP and GCM predictors and screened by Pearson correlation analysis, cross - correlation and standard deviation under three approaches. Based on performance indices, the best ANN models are selected. Six different ANN models are developed with six different input dataset. Multiple linear regression approach is used to build six other downscaling models to compare the performance of the ANN models. Those models are evaluated by different statistical indices and extreme climate indices. In this study MPI - ESM - MR CMIP5 GCM is used to collect predictors for present (1966 - 1995) and future time series (2011 - 2040; 2041 - 2070; 2071 - 2100). Effect of climate change on downscaled future precipitation under RCP 4.5 is analyzed for six ANN downscaling models. In the comparative analysis, ANN models perform better regarding maximum correlation with observed precipitation, the ability to capture maximum precipitation, the computation of short & medium term drought event analysis and CDF’s of maximum/minimum consecutive 5 month rainfall. After downscaling future precipitation through ANN models, mix results are exhibited in trend analysis of ANN models. But the projections with developed downscaling models reveal a significant changes in future monthly precipitation and number of wet events in the future. Results obtained from the study pro vide understanding of suitability of different input dataset in downscaling precipitation through ANN. This study exhibits the efficiency of ANN downscaled precipitation in capturing extreme weather events. This study reveals the condition of extreme climate indices in future precipitation projections under climate change scenario which will help the policy makers to take strategic adaptation measures to reduce the impact on water resource systems. |
| Year | 2016 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. WM-16-34 |
| 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;Duc Hoang Nguyen;Manukid Parnichkun |
| Scholarship Donor(s) | NICHE - BGD 081 Project, The Netherlands |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2016 |