Hits:

Indexed by:期刊论文

Date of Publication:2016-07-01

Journal:International Conference on Applications and Mathematical Analysis in Engineering and Science (AMAES)

Included Journals:SCIE、CPCI-S

Volume:12

Issue:3

Page Number:623-633

ISSN No.:1348-9151

Key Words:bi-level programming; clustering method; sea ice surface; pressure ridges; morphology feature

Abstract:As the most important morphological feature of the sea ice surface, pressure ridges are generally analyzed by clustering methods owing to the significant variations of the morphology caused by the geographical location and the environment in which they form. To seek an effective method for clustering the pressure ridges, we establish a bi-level programming model in which the cluster centers and the cluster number k are the parameters and the distance between the sample and the cluster center is the performance function. Based on the measured data set of the sea ice surface elevation in northwestern Weddell Sea of Antarctic, we propose a corresponding algorithm to search for the optimal solution of this bi-level programming model. The results indicate that the optimal cluster number of the proposed algorithm is 3 for the measured data set, the boundary between any two adjacent clusters obtained by this proposed algorithm is clearer and the performance indicators, such as the performance function of the bi-level programming model, the maximum distance within a cluster and the minimum distance between any two clusters are all better than those obtained by the traditional k-means clustering algorithm. The influences of the geographical location and the environment on the morphological variations of the pressure ridges can also be perfectly reflected by the cluster results obtained by the developed algorithm.