- 101 - SAITM Research Symposium on Engineering Advancements 2014 (SAITM – RSEA 2014) RAINFALL INTENSITY-DURATION-FREQUENCY RELATIONSHIP FOR COLOMBO REGION IN SRI LANKA R.Suthakaran 1* , K.Perera 2 and N.Wikramanayake 3 1 Department of Mathematics and Computer Science, The Open University of Sri Lanka, Sri Lanka, Email: rzuthakaran650@gmail.com 2 Department of Engineering Mathematics, Faculty of Engineering, University of Peradeniya, Sri Lanka, Email: kanthip@pdn.ac.lk 3 Department of Civil Engineering, The Open University of Sri Lanka, Sri Lanka, Email: tomwiks@yahoo.com ABSTRACT This paper presents a rainfall intensity-duration-frequency curve for 2, 5, 10, 20, 50, 100 and 200 years return periods for Colombo region. Annual peak rainfall values for different durations such as 1, 2, 4, 6, 12 and 24 hours were extracted for the purpose of this analysis. Then these values were fitted to Log Pearson type III (LP3) and Gumbel Extreme Value (EV1) distribution. Subsequently, the best fitted distribution for a particular duration was selected based on the goodness-of-tests approach. LP3 distribution was found to be the best fitted distribution for 1, 4, 6 and 24 hour’s duration of annual peak precipitation while, EV1 was the most appropriate distribution for other durations such as 2 and 12 hour’s. Further, the IDF curve was developed for Colombo region and it is recommended for the predication of maximum rainfall intensities. Keywords: Return Period, Intensity-Duration-Frequency (IDF) curve, Probability Distribution, Goodness-of-Tests 1. INTRODUCTION A rainfall intensity-duration-frequency (IDF) relationship is the most commonly used method for designing and planning of various water resource projects stated by Chow and Maidment [3]. The IDF is a kind of mathematical relationships between the rainfall intensity, duration and the return period. This relationship can be determined through statistical analysis of rainfall data collected from the corresponding stations. The major objective of this study is to find the most appropriate probability distribution for annual maximum rainfall at different level of duration. And also, estimate the maximum rainfall intensity for a particular duration and their corresponding return period. In addition to this, develop IDF curve for Colombo region. F. Y. Logah [4] analyzed the rainfall IDF relationship for Accra in Ghana by assuming that the data followed a Gumbel distribution. Furthermore, Garcia-Bartual and Schneider [5] used several statistical distributions and found the Gumbel distribution fitted to data well. Moreover, Jebreen M.Al-jebreen (1985) [6] calculated the floods magnitude for different return periods 5,10,25,50 and 100 years, using extreme value Type I distribution and Log-Pearson Type III distribution. Further, they stated extreme value type I distribution is the most suitable distribution for this region. 2. METHODOLOGY The maximum annual rainfall values for periods of 1, 2, 4, 6, 12 and 24 hour’s were extracted from a data set of 11 years of precipitation data at 15 minute intervals which was obtained from the pluviograph charts recorded at the Colombo Meteorological Station by the Department of Meteorology. Then, two popular probability distributions such as EV1 and LP3 were considered in order to find the most appropriate distribution at different duration. Subsequently, the best fitted distribution was chosen based on the goodness-of- fit test. And also, mean    and standard deviation  of the peak annual values of all durations were estimated. 2.1 Goodness of Fit Test 2.1.1 Akaike Information Criteria (AIC) AIC is a method of selecting a best model from a set of models. In general, the preferred model is the one with the minimum AIC value. AIC is defined as,       Where k is the number of parameters in the statistical model and L is the maximized value of the likelihood function for the estimated model. 2.1.2 Bayesian Information Criteria (BIC) The Bayesian Information Criterion (BIC) is       