中文摘要：

针对水文预报领域冰层厚度检测过程中对空气与冰层、冰层与冰下水层的分界点难以判断的难题,提出了一种基于介质温度数值分布规律的冰层厚度检测分析方法.新方法在对空气与冰层、冰层与冰下水层的分界点进行分析判断时采用了聚类分析思想,通过采用改进属性归类规则的K-means算法,对采集到的空气、冰与水不同介质的温度数值分布数据进行分类,并对各自属性的温度梯度分布数据集合进行线性拟合,从各拟合曲线交点可以获得空气与冰层、冰层与冰下水层的分界点,进而计算出冰层厚度数值.采用这一方法对2014.12-2015.3内蒙包头黄河河道冰情采集数据进行分析,证明了新方法的可行性。

英文摘要：

For the process of detecting ice thickness in hydrological forecasting file, there is a problem that is difficult to determine the demarcation among air and ice, ice and water layer. We propose a method for detection and analysis based on ice thickness distribution of the medium temperature. The new approach uses cluster thinking in the analysis and judgment of ice thickness based on numerical distribution of medium temperature. Through the use of improved property classification rules of K-means algorithm, we classified the data collections of temperate in different mediums of air, water, ice; meanwhile we fit the linear regression of respective in temperature gradient distribution of Data collection. Form the Point of intersection of each fitting curve, we obtain the demarcation points among air and ice, ice and water layer. And then calculate ice thickness values. By using this method for the Yellow River in Inner Mongolia Baotou 2014.12-2015.3 ice conditions to collect data for analysis proved feasibility of the new method.