Generating high-resolution rainfall data using statistical disaggregation techniques

• Main Author: Tan, Sheryl Zi Hui
• Other Authors: Qin Xiaosheng
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Civil engineering::Water resources
• Online Access: https://hdl.handle.net/10356/145397
• Institution: Nanyang Technological University

The rapid increase in rate and magnitude of climate change has been a global concern across the world in recent years. Apart from global warming being one of the contributing factors, global precipitation changes are also affecting the occurrence of extreme events such as droughts and floods, causing flood events to happen with more intense precipitation and accelerating the occurrence of drought events. With the increased risk in such extreme events, it is thus of high importance for designers to conduct water resource planning and management to ensure the abundance of water supply. To do so, knowledge from extreme rainfall events such as droughts and floods are needed, which can be achieved by analyzing rainfall data at finer timescales. However, the availability of data at timescales as fine as hourly is of a bigger challenge than coarser timescales such as daily due to factors such as complicated geographical conditions for various locations and costly data procurement. Therefore, the common approach is to convert the available daily data to hourly precipitation that will aggregate up to the daily totals. Disaggregation is thus one of the rainfall prediction methods adopted to generate the rainfall data at finer scales. In this project, the historical rainfall data obtained from the Changi Climatic Station will be used to perform disaggregation, with a data period from 1991-2010. There are a total of six Bartlett-Lewis Rectangular Pulse Model (BLRPM) parameters required to perform disaggregation in the Hyetos software: parameter lambda λ, parameter phi Φ, parameter kappa ĸ, parameter mean μx, shape factor α and scale factor v. These parameter values will be obtained by performing the Solver Function in the Excel template provided for the Hyetos software. These values will then be included as part of the input coding that will be inserted into the Hyetos software to perform disaggregation. Upon completing the disaggregation in the Hyetos program, the results will be discussed and compared to deduce the feasibility and accuracy of the rainfall disaggregation done by the software. Lastly, the report will end with a conclusion summarizing the key takeaways from the disaggregation conducted, as well as proposing recommendations that can be considered to improve the efficiency and reliability of the disaggregation conducted by the software.