Application of Mahalanobis-Taguchi system in Rainfall Distribution

Abstract

The rainfall time series is often nonlinear and multi-time scale because of hydrology, meteorological, and human activity. Weather stations gather information on a diverse set of parameters on order to monitor and analyses patterns of rainfall. Nevertheless, not all parameters are created equal in terms of its significance or effectiveness in carrying out classification and optimization actions. The objective is to classify rainfall occurrences by the RT method and optimize the parameter selection process by the T method using Mahalanobis-Taguchi system (MTS). The data was collected using Vantage Pro2 weather station at UMPSA Gambang campus and it consists of 16 various parameters. As a results, RT method can classify the data samples in terms of MD for the months of June, October and December by utilizing the, while simultaneously the number of parameters is reduced to only those that substantially contribute to the classification. This brings the total number of parameters decrease from 16 to 8 when compared to the T method. So, this research methods offer a simplified and effective way for analyzing rainfall patterns and optimizing the data gathering processes at weather stations.