Variability analysis of Kuroshio intrusion through Luzon Strait using growing hierarchical self-organizing map

Abstract

An advanced artificial neural network classification algorithm is applied to 18 years of gridded mean geostrophic velocity multi-satellite data to study the Kuroshio intrusion into the South China Sea through the Luzon Strait. The results suggest that the Kuroshio intrusion may occur year round. However, intrusion is not the major characteristic of the region. The intrusion mode occurs only 25.8 % of the time. Winter intrusion events are more frequent than summer events. Both stronger intrusion (which is related to wind speed) and weaker intrusion (which may be related to the upstream Kuroshio transport) may occur during winter, but stronger intrusion is dominant. In summer, the Kuroshio intrusion is almost the weaker type. The Kuroshio intrusion through the Luzon Strait usually occurs when the Pacific decadal oscillation index is positive (72.1 % of the time). This study shows that growing hierarchical self-organizing map is a useful tool for analyzing Kuroshio intrusion through the Luzon Strait.