Homogeneity tests and trend analysis for rainfall data in the southern region of Peninsular Malaysia

Abstract

This study focuses on performing homogeneity tests and trend analysis for rainfall data in the southern region of Peninsular Malaysia. Twelve rainfall stations throughout the state of Johor, Melaka and Negeri Sembilan with data from 1980 to 2016 were used in this study. Four absolute homogeneity tests namely standard normal homogeneity test, Buishand range test, Pettitt test and von Neumann ratio test are carried out to determine the homogeneity of the annual rainfall data. To evaluate the performance of the methods used, three different variables were used, which are annual total amount rainfall, annual mean and annual median. The results are categorized into three classes which are useful, doubtful and suspect depending on the p-value obtained from those absolute homogeneity tests. From the results of the three variables used, only one station is considered as inhomogeneous “labelled as suspect” while the other stations are considered homogeneous. Mann-Kendall test and Sen’s slope estimator was used to study the trend analysis. Annual and seasonal rainfall indices (southwest monsoon and northeast monsoon season) with total amount of rainfall index are used to evaluate the trend analysis. From the Mann-Kendall test, p-values are obtained to determine the presence of a trend within a 5% significance level. Results for the annual period and northeast monsoon season shows that only one station has a significant increasing trend. However, during the southwest monsoon season, all of the station did not have a significant trend. To conclude the trend analysis, more than half of the stations indicated an increasing trend, although it is not statistically significant. Sen’s slope estimator is carried out to estimate the magnitude of the trend. The positive/negative value in the magnitude of the trend corresponds to the tau value in the Mann-Kendall test. Most of the rainfall stations shows a positive magnitude in slope in all criteria of the annual and seasonal rainfall indices.