An alternative approach to characterize time series data: Case study on Malaysian rainfall data

This paper focuses on the characterization of time series rainfall data to understand the behavior in the Malaysian rainfall data. An analysis of the rainfall behavior of different time periods is also conducted. The rainfall data is rounded. A bar with a width of 12 pixels is drawn for each day in the data. The length of the bars drawn is equal to the corresponding rainfall value. Each bar is separated by a space of 2 pixels. Bars for data from different year are stored in different rows. Morphological opening is performed on the barcode image obtained, using line kernels of increasing length. Connected component-labeling algorithm is implemented on the resulting images to identify the individual bar codes and to compute the number of bars remaining after the opening process. A daily rainfall record for the duration of 30 years, obtained from the Melaka Meteorological Station, is analyzed. The results provide characterization of behavior in daily rainfall data.