Precipitation trend analysis using discrete wavelet transform at the Langat River Basin, Selangor, Malaysia
Narges Palizdan,Yashar Falamarzi,Yuk Feng Huang,Teang Shui Lee
DOI: https://doi.org/10.1007/s00477-016-1261-3
IF: 3.821
2016-05-11
Stochastic Environmental Research and Risk Assessment
Abstract:The main purpose of this study was to determine the most dominant periodic components that affect the annual and seasonal precipitation trends in each homogenous rainfall region in the Langat River Basin, Malaysia for the period 1982–2011. Performing this research could be essential because in the previous studies on detection of trend in Malaysia, the details of variations of different time scales and the periodic responsible for the observed trends were not investigated. Using discrete wavelet transform (DWT) coupled with Mann–Kendall at the regional scale for the first time particularly in the context of Malaysia is the contribution of this study. In order to form the homogenous rainfall regions, first the total annual and seasonal precipitation in each year was spatialized into 5 km × 5 km grids using the inverse distance weighting method. The obtained precipitation series for the grids were then grouped applying the Ward’s clustering method based on the similarity of precipitation time series. After allocating a cluster number to each grid, the boundary of the regions was formed in ArcGIS software. Following which, in each homogenous region the areal precipitation series were computed by the Thiessen polygon method. The Mann–Kendall (MK) test was used to detect trend and the DWT coupled with the MK test and the sequential MK analysis were then utilized in order to find out the time scale which affected the observed trend in each homogenous region. On annual scale, it was found that D1 (plus approximation) component in regions Annual Cluster1 (AC1) and AC2 was the periodic mode responsible for trends. On seasonal scale, in regions Northeast monsoon Cluster 1 (NC1), NC3, SC1 and Southwest monsoon Cluster 2 (SC2), D1 (with approximation), in regions NC4, Inter monsoon 1 Cluster 1 (I1C1), I1C2, Inter monsoon 2 Cluster 1 I2C1 and I2C2, Detail 2 (D2) (plus approximation) and in region NC2, Detail 3 (D3) (with approximation added) component were the most influential periodicity for trends.
environmental sciences,engineering, environmental,water resources, civil,statistics & probability