Cluster and Trend Analysis of Rainfall Time Series in the Nakdong River Basin

Abstract

In Korea, it is difficult to efficiently manage water resources, due to the variability of rainfall in addition to the increasing outflow due to climate change in recent years. Therefore, if the trends and characteristics of rainfall at each station can be identified in advance, the problems caused by the variability of rainfall can be effectively dealt with. In this study, the data on rainfall characteristics of 64 rainfall stations in the Nakdong river basin were collected for the period 2000 to 2019. The data were analyzed according to the elevation of each station using K-means cluster analysis, and the rainfall trends at each station were identified by homogeneity test and modified Mann-Kendall test. The analysis showed an increasing trend in March, April, November, December, spring and autumn, and a decreasing trend in January, May, September, summer and year. Also, based on the cluster analysis, it was confirmed that when the number of clusters was set to three, the rainfall characteristics were different depending on the elevation of each station. It is believed that linking the characteristics of rainfall by cluster and the results of trend analysis by station derived from the study can be used to come up with a water resource management plan that takes into account the variability of rainfall.