Functional Data Analysis Technique on Daily Rainfall Data: A Case Study at North Region of Peninsular Malaysia

Abstract

The study of rainfall features and patterns are very useful for water management systems, water resources engineering and also in agricultural planning. It can be beneficial in order to reduce the risks and losses. Functional data analysis technique is one of the method can be used to explore and display the pattern and variation of the rainfall data. This technique displays the pattern in the form of curves. The first and second derivatives of the curves represent the rate of change and the acceleration of the curves. The objective of the study is to model two rainfall features; rainfall amount and rainfall occurrence by using functional data analysis technique at eight rainfall stations from the north part of Peninsular Malaysia. Markov chain model has been used to model the rainfall occurrence and Fourier basis to smoothing the data. The results show that both of the rainfall features have similar bimodal pattern. Although the mean curves are slightly similar, the first peak of variance curve for rainfall occurrence is higher than the second peak which is difference with variance curve for rainfall amount. The relationship between rainfall amount and rainfall occurrence for both observed and estimated curve is also discussed. Keywords: Functional Data Analysis; Markov Chain; Rainfall Amount; Probability of Rainfall Occurrence. 2010 Mathematics Subject Classification: 62P12