优化K-means算法在中国近海气候区划中的应用

Application of Optimized K-means Clustering Algorithm in China Offshore Climate Regionalization

利用我国近海多年的风、浪及温湿资料,采用DB指标和最大距离法相结合的优化K-means聚类算法,对冬季我国近海进行气候区划,避免了传统K-means算法中确定聚类数目和初始聚类中心的主观性。结果表明,冬季我国近海可被划分为3个区域:1区主要包括28°N以北的海域以及我国东南沿海的一条狭长区域,2区主要包括台湾周边海域、台湾海峡、巴士海峡以及自越南东南部起由西南部向东北部的南海大部分海域,3区则主要包括台湾岛西南部海域、北部湾、广东沿海以及南海东南部。根据给出的区划指标,分析各分区的气候特征及其对海上舰船航行和出海人员的影响,判定1区为基本适宜区,2区为不适宜区,3区为适宜区。

Based on wind,wave height,temperature and humidity data,the optimized K-means algorithm which combines DB criterion and the method of maximum distance is used in China＇s offshore climate regionalization in winter.The optimized algorithm avoids the drawbacks of subjectivity of determining cluster number and initial cluster centers in the traditional K-means algorithm.The results show that three regions are obtained：Region 1 mainly includes the sea area north of 28°N and the long narrow area in the southeast coastal of China;Region 2 mainly includes the circumjacent seas of the Taiwan Island,Taiwan Strait,Bashi Channel and most of the South China Sea from the southwest to the northeast;Region 3 mainly includes the sea area in the southwest of the Taiwan Island,Beibu Gulf,coastal areas of Guangdong Province and the southeast of the South China Sea.By analyzing the climate characteristics of each region and their influences on ships and people on the seas,according to the regionalization criteria,we conclude that Region 1 is basically suitable for sailing,Region 2 is not suitable,and Region 3 is suitable.