Study on Singapore's rainfall intensity-duration-frequency (IDF) relationships

The purpose of this research study is to analyze and derive the spatial rainfall Intensity- duration-frequency (IDF) relationship patterns in Singapore, specifically in the Eastern, Northeastern and Western regions in Singapore. The Intensity-duration-frequency (IDF) relationships derived from rainfall measurements are commonly used in engineering, specifically in water resource and hydraulic engineering for the designing, outlining and planning of projects related to water resources (Elsebaie, 2012). Recognizing and establishing the rainfall patterns and IDF rainfall relationships is basic to the design and construction of urban drainage works such as storm sewers, culverts and other hydraulic structures. This study describes the establishment of the IDF rainfall relationship in the Eastern, Northeastern and Western regions of Singapore using the Gumbel distribution and the Log Pearson Type III distribution (LPT III). The IDF curves and trends for each region and station are tabulated. Based on the Gumbel distribution, the results suggest that there is no increasing trend in rainfall intensity in the Eastern, Northeastern and Western regions in Singapore, with the highest rainfall intensity correlated with the lowest return period of rainfall. It is also observed that the highest rainfall intensity occurred in the Northeastern region while the lowest rainfall intensity occurred in the Eastern region. Based on Log Pearson Type III distribution, the peak rainfall intensity is also observed in the North-East region. Some discrepancies can be observed between the results of Gumbel distribution and LPT III distribution. The results obtained using Gumbel distribution are slightly lower for return periods of 5 years and below, compared to the results obtained using the LPT III distribution. The rainfall intensities obtained from these two methods showed good agreement with other studies on rainfall pattern of some regions in Singapore, such as the study conducted by Oyebande (1982). Goodness of fit tests using the chi-square method was also applied to both distributions. It was found that the Gumbel distribution has the best match (Izinyon, & Igbinoba, 2013) for the data used in this report. A further analysis on the Gumbel distribution using the regression test showed that the results from Gumbel were most accurate for lower return periods. More methods such as Log-normal, Pearson Type III, Simple scaling, Two- parameter-Log-Normal together with Gumbel distribution method and LPT III can be employed for frequency analysis and the averaged observations used to strengthen the accuracy of this study. Various other limitations of the results and study are discussed.Bachelor of Engineering (Civil