CAJ | 학술논문

基于1992年10月至2009年11月卫星观测的海表高度(SSH)时间序列数据，应用增长型分级自组织映射(GHSOM)人工神经网络方法研究南海北部和西太平洋 SSH和中尺度涡旋的变化，识别出该海域 SSH的季节和年际变化信号。分析表明，流经吕宋海峡的黑潮分支在冷季入侵南海北部，同时在吕宋岛西北海域出现一个强烈的气旋式涡旋，表层黑潮的入侵与跨过吕宋海峡南北的经向压力梯度密切相关。黑潮的非入侵事件主要出现在暖季。春秋季节作为两个事件的过渡期，环流结构复杂，由 GHSOM的第2层特征图进一步进行分类识别。黑潮入侵事件和非入侵事件发生的百分比分别为24.57%和27.53%，过渡模态的百分比为47.87%。当入侵南海事件发生时，南海北部表层环流流态相对简单，主要为气旋环流控制南海北部，吕宋海峡表层海流是否入侵南海，与南海北部中尺度涡旋特别是吕宋岛西北的气旋式涡的变化关系密切；反之，在非入侵事件发生时，南海北部出现多涡结构，环流流态复杂，表明吕宋海峡海流入侵南海对南海北部环流也有重要调整作用。除季节尺度变化外，年际时间尺度变化信号也十分显著。在1994-1995、1997-1998和2002-2003年期间，表层黑潮入侵南海北部的事件要显著多于其他年份，然而入侵事件在1998-2001年和2006-2009年时间段明显减少，非入侵事件增加。应用欧氏距离定义的模态2的时间发展序列与Ni?o3.4指数序列延迟相关。
기우1992년10월지2009년11월위성관측적해표고도(SSH)시간서렬수거，응용증장형분급자조직영사(GHSOM)인공신경망락방법연구남해북부화서태평양 SSH화중척도와선적변화，식별출해해역 SSH적계절화년제변화신호。분석표명，류경려송해협적흑조분지재랭계입침남해북부，동시재려송도서북해역출현일개강렬적기선식와선，표층흑조적입침여과과려송해협남북적경향압력제도밀절상관。흑조적비입침사건주요출현재난계。춘추계절작위량개사건적과도기，배류결구복잡，유 GHSOM적제2층특정도진일보진행분류식별。흑조입침사건화비입침사건발생적백분비분별위24.57%화27.53%，과도모태적백분비위47.87%。당입침남해사건발생시，남해북부표층배류류태상대간단，주요위기선배류공제남해북부，려송해협표층해류시부입침남해，여남해북부중척도와선특별시려송도서북적기선식와적변화관계밀절；반지，재비입침사건발생시，남해북부출현다와결구，배류류태복잡，표명려송해협해류입침남해대남해북부배류야유중요조정작용。제계절척도변화외，년제시간척도변화신호야십분현저。재1994-1995、1997-1998화2002-2003년기간，표층흑조입침남해북부적사건요현저다우기타년빈，연이입침사건재1998-2001년화2006-2009년시간단명현감소，비입침사건증가。응용구씨거리정의적모태2적시간발전서렬여Ni?o3.4지수서렬연지상관。
Sea surface height (SSH) variations and eddies on both sides of the Luzon Strait are examined from the merged satellite altimeter data from October 1992 to November 2009. The neural network analyses based on the growing hierarchical self-organizing map (GHSOM) are used to extract feature patterns of the circulation variability. The evolution of the characteristic circulation patterns with time is investigated. The seasonal and inter-annual variations of the sea-surface circulation are identified. The analysis indicates a branch of the Kuroshio intrudes into the northern South China Sea (NSCS) during the cold season, accompanied by a stronger cyclonic eddy northwest of Luzon Island. That is closely related to the meridional pressure gradient cross the Luzon Strait. The Kuroshio does not intrude remarkably in the warm season. The spring and autumn, as transitional periods, have relatively complicated circulation structures, which can be explained by more GHSOM arrays in the second layer. The prominent Kuroshio intrusion and non-intrusion patterns denote 24.57%and 27.53%of all SSH variation, respectively. The transitional pattern sums up to 47.87%. The seasonal variation is modulated by the inter-annual variation. The events that the surface Kuroshio intruded into the NSCS in 1994?1995, 1997?1998 and 2002?2003 are much more frequent than the other periods, and the frequency of the events decreased remarkably from 1998 to 2001 and from 2006 to 2009, with no intrusion events were prominent in these two periods. The time series of non-intrusion pattern is best correlated with the Ni?o3.4 index, when it lags the Ni?o3.4 index by one month.