散射计海面非气旋风场块状模糊去除方法

Ocean Surface Non-cyclone Wind Block Ambiguity Removal Algorithm for Scatterometer

  • 摘要: 利用最大似然法 (MLE) 对散射计数据反演得出的风矢量,一般存在多个模糊解,故需采用圆中数滤波法进行模糊去除。但传统圆中数滤波法难以解决风场反演中块状模糊问题。该文根据散射计非气旋性第1风场的空间分布特性,归纳出一套适合散射计的加强型圆中数滤波块状模糊去除方法,并探讨了其适应性。该方法定义简单,计算量小,且易收敛。利用欧洲EUMETSAT提供的部分ASCAT (Advanced SCATterometer) L2原始数据对该方法进行验证结果表明,对于非气旋风场分布情况,该方法在利用其他工具去除台风气旋覆盖区域后,能有效解决非气旋区域风场块状模糊问题。

     

    Abstract: The Maximum Likelihood Estimation (MLE) algorithm for scatterometer wind vector retrieval generates several wind vector ambiguities, so a circle median filter is needed to perform the ambiguity removal. But the traditional circle median filtering method can hardly solve the block ambiguity problem. According to the spatial distribution characteristic of the most likely ambiguities in each non-cyclone wind vector cell, a new enhanced circle median filtering method for block ambiguity removal is derived and discussed theoretically, with experiments carried out to check its adaptability. This method features simple definition, low computation and easiness to converge. Using some L2 raw data from EUMETSAT to validate the method, the results indicate that under non-cyclone wind distribution condition the new method is effective in resolving the problem of block ambiguity after eliminating the cyclone wind field with other reference data.The core of the enhanced circular median filter algorithm is to initialize the non-cyclone characteristics first, which can effectively solve the problem of massive fuzzy. Thus the data which may cause circular median filtering failure and the data that affect the neighborhood will be corrected. Then divide the two-dimensional space into M rows and N columns, then calculate the wind field in the open window. The wind vector in the center of a window is solved by selecting an alternative from corresponding fuzzy solutions. And then do this with the next location iteratively, until the wind field does not change or until the times of iterations reaches a preset maximum number. Finally, defective value in the wind field is smoothed.Compared with traditional circle median filtering method, this approach is better in several ways. First, by initializing the first wind field, the fuzziness of the second wind field is reduced; while with the traditional method the block fuzzy cannot be removed. Second, the calculation process is simple and need no statistical circular histogram, nor do they need to calculate the mean. Third, the definition of circular median is only one rather than getting multiple solutions. And last, the calculation will not be interrupted by the narrow wind element values in the boundary region.The enhanced circular median filtering method based on non-cyclone wind field vector distribution characteristics can overcome the harsh conditions of the traditional method (such as wind field must be randomly distributed, non-block fuzzy), extracting the true wind vector solutions to overcome the fuzzy block. In order to apply this method widely, a crucial issue is to determine and eliminate the coverage typhoon cloud by operational satellite equipment properly. The method provides a new idea for exacting data of non-cyclone wind field on the ocean surface.

     

/

返回文章
返回