温跃层边界提取的组合算法*

本文提出一种用于海洋温跃层边界提取的组合算法，该算法基于调整判定阈值后的垂直梯度法和一种改进函数模型几何形态后的拟阶梯函数逼近法。为验证该算法对于温跃层边界提取的适应性、客观性，利用中国Argo实时资料中心发布的全球海洋Argo网格资料集（BOA_Argo），选取2018年夏季0°~30°N，135°E~165°E所在的太平洋区域的温度剖面数据进行处理，结果表明，该算法对低纬度海区温度剖面拟合质量较好，拟合均方根误差普遍低于1.3℃，而且提取到的温跃层上下边界在纬向更加平缓，然而对于中高纬度海区的温度剖面，组合算法的处理效果不佳，更适合直接通过垂直梯度法进行温跃层的提取。

A combinatorial algorithm for identification of the thermocline bounds

A kind of combinatorial algorithm for identifying the boundary of thermocline is proposed in this paper. This algorithm is based on the vertical gradient method and a kind of improved quasi-step function approximation method. To prove the universality and objectivity of the processing results, the dataset of BOA_Argo is used in this experiment. The temperature profiles of a specific Pacific Ocean area (0°~30°N, 135°E~165°E) are processed with this algorithm. The result reflects that this combinatorial algorithm can fit temperature profiles of low-attitude sea area with high quality and the RMSE of fitted temperatures is generally lower than 1.3℃. The boundaries of thermoclines acquired by this algorithm have more continuity and lower jitter in the direction of latitude. However, as for the temperature profiles of the middle and high-latitude sea area, the algorithm generally performs poorly. So, it is proper to take the boundaries of thermocline directly only by vertical method in the sea area of middle and high latitude.