Comparative analysis of statistical reduction and classification methods for Philippine rainfall

Authors

Ella R. Castillo ⋅ PH Manila Observatory
Charlotte Kendra Castillo ⋅ PH Manila Observatory
Jose T. Villarin, S.J. ⋅ PH Manila Observatory

Abstract

Changing climatic conditions necessitate the updating of traditional climate classification schemes in the Philippines, which are based primarily on the Coronas Climate Atlas created in 1920. This study aims to establish an objective method for reducing rainfall data that is varying both in space and time and classify areas with similar climatic rainfall patterns. Precipitation data with a 0.5° x 0.5° resolution is extracted from the global decadal and climatological monthly means of the Intergovernmental Panel on Climate Change (IPCC) Data Distribution Center. Two statistical methods of clustering are evaluated: Positive Matrix Factorization (PMF) and Empirical Orthogonal Functions (EOF). Both methods identify four dominant climate types with similar profiles. Comparative analysis and validity tests show that the PMF produces a better representation of the country’s climatological rainfall patterns, although results are inconclusive without high-resolution data. It is recommended that more applicable indicators of fit be developed to better compare clustering techniques.

Downloads

Issue

2005: Proceedings of the 23rd Samahang Pisika ng Pilipinas Physics Congress

Article ID

SPP-2005-1G-04

Section

Atmospheric Studies

Published

2005-10-26

How to Cite

[1]

ER Castillo, CK Castillo, and JT Villarin, Comparative analysis of statistical reduction and classification methods for Philippine rainfall, Proceedings of the Samahang Pisika ng Pilipinas 23, SPP-2005-1G-04 (2005). URL: https://proceedings.spp-online.org/article/view/SPP-2005-1G-04.