以中分辨率成像光谱仪(moderate—resolutionimagingspectroradiometer,MODIs)影像为数据源,提出了一种基于数据挖掘的海冰检测方法。根据海冰的反射率选择特征属性,并基于数据挖掘提取海冰检测规则。试验结果表明,在无云覆盖情况下,应用本文挖掘出的海冰检测规则能够得到较高精度的检测结果,与目前常用的归一化积雪指数(normalizeddifferencesnowindex,NDSI)海冰检测方法相比,本文提出的海冰检测方法不仅能够检测出固定冰,而且能检测出厚度较薄的浮冰。
In this paper, with MODIS(moderate-resolution imaging spectroradiometer) images as the data source, a method of sea ice detection is probed from the perspective of data mining. First attributes are se- lected based on the reflectivity;then common rules are digged out for detection and identification of sea ice after selecting samples. The result of the experiment shows that in the case of cloud-free coverage, high accuracy can be obtained, comparied with NDSI, not only for marginal ice zone, but also pack ice zone.