Comparison of parametric and nonparametric estimation methods for annual precipitation in Kuala Lumpur

Flash floods are becoming a critical issue as they occur more frequently in recent years. Managing watersheds and water resources, researching floods and droughts, and monitoring climate change are all connected to annual precipitation. Therefore, discovering the most accurate method for calculating annual precipitation is crucial. This study compares two basic approaches to estimating annual precipitation parameters: parametric and nonparametric. The research focuses on fitting the distribution of annual precipitation for fifteen strategically located rain gauge stations scattered around Kuala Lumpur. These stations play a crucial role in providing comprehensive data for the study. The Generalized Extreme Value (GEV) distribution is utilized for parametric approaches with Maximum Likelihood Estimation (MLE) as the parameter estimator. Meanwhile, the kernel function using the Gaussian distribution is applied for the nonparametric method. Two approaches are used to compute the smoothing parameter: Silverman’s Rule of Thumb (ROT) and the Adamowski Criterion (AC). The goodness-of-fit of the proposed models is assessed using the Mean Relative Deviation (MRD) and Mean Squared Relative Deviation (MSRD) statistics to evaluate nonparametric and parametric models. The results show that ROT was the best method compared to AC and MLE in fitting the distribution for the fifteen rainfall stations in Kuala Lumpur. According to the study, nonparametric approaches can be an alternative for estimating the annual precipitation in Kuala Lumpur.

Text 123913.pdf - Published Version Restricted to Repository staff only Download (339kB) Official URL or Download Paper: https://science.lpnu.ua/mmc/all-volumes-and-issues...

Actions (login required)

View Item