Performance of Principal Component Analysis to Classify Precipitation Type from Raindrop Size Distribution Data at Kototabang, Indonesia

Abstract

This study examines the use of principal component analysis (PCA) to classify the RDSD data at Kototabang, Indonesia. In addition to PCA with 6 attributes (hereinafter called PCA6) that had been developed by a previous researcher, this study also examines PCA with 7 attributes (PCA7) by adding radar reflectivity factor. The PCA is applied to the RDSD that had been classified by a wind profiler into Stratiform (S), deep convective (DC), shallow convective (SHC) and mixed stratiform/convective (MSC). The number of unclassified data is much smaller than that reported by previous study in which it is about 33-47% with PCA6 and 29-44% with PCA7. While the PCA classifies the same group for different rain type from wind profiler, especially for Group I (moderate D_o and large N _W ) and II (small D_o and N _W ), some differences are observed. Each rain type classified by wind profiler has different dominant group in which Group II is dominant for S, Group I and V (large D_o and low N _w ) are dominant for DC, Group I, IV (small D_o and large N _w ) and VI (small D_o and very large N _w ) are dominant for SHC and Group I is also dominant for MSC type.