Precipitation trend analysis for The Langat River Basin, Selangor, Malaysia

Almost all the hydrological and meteorological variables such as rainfall and temperature etc are affected by global climate change. Precipitation is one significant climate factor that affects the state of the environment and contributes significantly towards the natural hydrological process. Its temporal fluctuation is significant both in terms of the scientific knowledge and practical applications, such as in water resources planning and management. It is necessary to evaluate the regional precipitation trend to improve water management strategies for a given region. A study was conducted to analyse the annual and seasonal precipitation trends of 03 stations in the Langat River Basin, located in the state of Selangor, Malaysia, at both the stationary and on regional scales. On an at site scale the rainfall trends were detected using the Mann-Kendall (MK) test, the Sen’s Slope Estimator and the Linear Regression analysis for the period 1928-2011. The lag-1 approach was utilized at the 95% significance level to test the serial correlation of the data series. Then the autocorrelated time series were pre-whitened. On the annual scale, it was found that most of the stations in the basin were characterized with insignificant precipitation trends. The significant trends were found only at the four stations, namely stations 44301, 44305, 44320 and 2719001. The results of the seasonal precipitation trend analysis showed that most of the stations, during the North East Monsoon (NEM) and Inter Monsoon 1 (INT1) seasons, and half of the stations in South West Monsoon (SWM) season, experienced insignificant positive trends. To the contrary, for the Inter Monsoon 2 (INT2) season, the majority of the stations showed negative trends. It was found that during NEM season the station 44301, for INT1 the stations 44301, 2719001 and 3118069 were established as having significant changes, while in SWM, the station 2917001 and during INT2 the stations 2615131and 44301 showed significant trends. In the regional study, the regional annual and seasonal precipitation trends for the period 1982-2011 were examined at the 95% level of significance, using the Regional Average Mann-Kendall (RAMK) test and the Regional Average MannKendall coupled with bootstrap (RAMK-bootstrap) method. In order to identify the homogeneous regions respectively for the annual and seasonal scales, firstly, at- site mean total annual and separately at site mean total seasonal precipitation were spatialized into 5 km × 5 km grids using the inverse distance weighting (IDW) algorithm. Next, the optimum number of homogeneous regions (clusters) is computed using the Silhouette Coefficient approach. Next, the homogeneous regions were then formed using the K-mean Clustering method. The applied homogenous region analysis method in the present study is a new approach. From the annual scale perspective, all three regions showed positive trends. However, the application of the two methods at this scale showed a significant trend only in the region AC1. The region AC2 experienced a significant positive trend using only the RAMK test. On a seasonal scale, all regions showed insignificant trends, except the regions I1C1 and I1C2 in the Inter Monsoon 1 (INT1) season which experienced significant upward trends. In addition, it was proved that the significance of trends has been affected by the existence of serial and spatial correlations. We also aimed to determine the most dominant periodic components that affect the trends in each homogeneous region for the period 1982-2011. The combination of wavelet analysis and Mann-Kendall test at regional scale was used for the first time in this study. In each region the areal precipitation series were computed using the Thiessen polygon method. In the method using wavelet transformation, the mother wavelet type and the number of decomposition levels were assessed based on two criteria. The first one is the Mean Relative Error (MRE) between the wavelet approximation component and the original data set and the second one is the relative error of the MK Z-value ( ) between the approximation series and the original data series. The Discrete Wavelet Transform (DWT) coupled with the Mann-Kendall (MK) test and the sequential MK analysis were utilized in order to find out which time scale are affecting the trends observed in each homogenous region. On annual scale it was found that DWT1 in region AC1 and AC2 and DWT2 in region AC3 were the periodic modes responsible for trends. On seasonal scale, in regions NC1, I1C1, I1C2, I1C3, SC1, SC2, I2C1 and I2C2, DWT1, in region NC2, DWT2 and DWT3, in region I2C3, DWT2 were the most influential periodicity for trends, respectively.

Preview Text FK 2014 75IR.pdf Download (1MB) | Preview

Actions (login required)

View Item