Statistical downscaling of regional climate model datasets under heterogeneous terrains | |
| Author | Hosen, Md Latif |
| Call Number | AIT Thesis no.WM-21-07 |
| Subject(s) | Climatic changes--Statistical methods Heterogeneous computing |
| 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 |
| Abstract | Climate change affecting the whole world. Future climate projection is essential for managing its effects appropriately. GCMs able to capture the climate at the course level. However, fine-resolution climate data is required for impact assessment studies. RCMs data is higher resolution than GCMs but limited. Researchers are trying to develop finer-scale climate data for local-level impact assessment studies. Thus, downscaling is a process that can resize course data to fine data. This study developed an improved statistical downscaling method incorporating lapse rate. The lapse rate is the change of variables with elevation. We consider surrounding climatic factors through lapse rate. So, we termed it spatio-temporal multiple linear regression statistical downscaling (MLR-Lapse rate) model. We interpolated Aphrodite 25 km data to 5 km and verified that with two meteorological stations data. Results found good matched (both R2 greater than 0.8). With this Aphrodite, we downscaled 25 km RCM data to 5 km. We worked with the two variables, monthly rainfall, and monthly mean temperature. Our study area is Mae Nam Wang basin in Thailand, baseline period was 1970 to 2005, and the projected future was 2021 to 2050. We evaluated the model performance. MAE was 25 mm/month, RMSE 50 mm/month, and R2 above 0.75 in most grids in the baseline for monthly temperature. For monthly mean temperature, MAE and RMSE 0.75 ˚C/month and R2 above 0.75 in most grids in the baseline period. Besides, we compared the improved model with the temporal statistical downscaling (LReg) model. We found MAE and RMSE improvement by around 10%, and R 2 improved slightly in most grids for both monthly variables. For extreme indices, MLR-Lapse rate monthly rainfall matched better with Aphrodite in the baseline. However, the LReg monthly mean temperature match with better Aphrodite than the MLR-Lapse rate model. Projected monthly rainfall for both RCP 2.6 and RCP 8.5 resulted from May to July decrement most areas around 10% in the future from the baseline Aphrodite. August to October Northern part increased by around 2.5%, while other parts remain primarily unchanged or slightly decreased ( up to 2.5%) for both scenarios. From November to April, some of the northern grids reduced rainfall up to 20%, while other parts have insignificant variation. |
| Year | 2021 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Civil and Infrastucture Engineering (DCIE) |
| Academic Program/FoS | Water Engineering and Management (WM) |
| Chairperson(s) | Shanmugam, Mohana Sundaram |
| Examination Committee(s) | Babel, Mukand;Shrestha, Sangam;Virdis, Salvatore G. P. |
| Scholarship Donor(s) | Government of Bangladesh |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2021 |