Trends in extreme daily rainfall in Peninsular Malaysia : 1948 - 2010
Loading...
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Teknologi Malaysia
Abstract
Climatologists have been scrutinizing changes in the climate of the past by analyzing climatic time series. Climate variability and change on rainfall have not been adequately investigated in Malaysia. The objectives of this study are: to introduce a new quality-controlled daily rainfall dataset for Peninsular Malaysia, to examine trends in extreme daily rainfall indices, and to introduce a new forecasting method for monthly rainfall series on a decadal base. Available daily rainfall dataset from the Department of Irrigation and Drainage Malaysia were analyzed. Homogeneity of the daily series was assessed by means of Bivariate test and Standard Normal Homogeneity Test considering correlation, distance and altitude of neighboring stations. Inhomogeneous series were then adjusted based on defined criteria. In addition, seven well-established missing value estimation techniques were compared and missing values were estimated using the best performance method. As a result, a new generation of rainfall dataset with at least 40-year length accompanied by complementary information was introduced for the region. Trends in daily extreme rainfall indices from 1948 to 2010 were then investigated. In the second phase of the study, after a comparison of existing metaheuristic algorithms, centripetal accelerated particle swarm optimization was introduced to forecast average monthly rainfall series for a decade. The results of this study show that two discernable patterns of missingness are prevailing in the dataset. The pair wise comparison of the stations suggested that the highest correlation coefficients were observed in less than 50km distances and the average in this radius was 0.43. The homogenization process identified that 445 stations (72.24%) are inhomogeneous. A significant decreasing trend was observed in annual total precipitation, while simple intensity index, highest 1 day precipitation amount and the contribution of very wet days to the total annual precipitation reveled positive trends. The introduced hybrid Singular Spectrum Analysis-CAPSO algorithm for the monthly rainfall forecasting presented the best results compared to the other metaheuristic algorithms.
Description
Thesis (PhD (Civil Engineering))
Keywords
Rain and rainfall—Malaysia, Rainfall intensity duration frequencies, Rainfall reliability