Clustering rainfall pattern in Malaysia using functional data analysis

• Main Authors: Hamdan, Muhammad Fauzee, Suhaila, Jamaludin, Jemain, Abdul Aziz
• Format: Conference or Workshop Item
• Published: 2015
• Subjects: QA Mathematics
• Online Access: http://eprints.utm.my/id/eprint/59186/; http://dx.doi.org/10.1063/1.4907466
• Institution: Universiti Teknologi Malaysia

Understanding rainfall pattern is important for planning and prediction in hydrology, meteorology, water planning and agriculture. There are two important features of rainfall: the rainfall amount and the probability of rainfall occurrence. The discrete raw data of rainfall precipitation was reconstructed into rainfall amount curves by using functional data analysis method. Hierarchical clustering method with complete-linkage method was used to search for natural similar groupings of rainfall amount curves. The functional clustering illustrated the four dominant patterns for rainfall amount curves. In additional, adaptive Neyman test showed that each clusters are significantly different with from each others.