中文摘要：

已有的SeaWinds散射计反演结果误差评价包括两种方法：一是与其他数据来源进行对比,如与数值气象预报产品或浮标风场数据对比;二是从风场反演结果的空间一致性进行判断。文中采用二维几何中心矩定量描述象元解空间中模糊解区域（与真解最接近的模糊解区域）的图谱特征,如图像椭圆的主轴倾角、长短轴比以及总灰度值;研究了总灰度值与风矢量反演误差的关系。结果表明,当总灰度值大于某一阈值时,风矢量反演误差将被限制在一个较小的范围之内。根据图像椭圆总灰度的这一特性,设计了一种新的基于总灰度约束特性的滤波算法。通过与浮标风矢量观测值的比较发现,新滤波方法比传统的圆中数滤波方法在风向反演误差、风向误差的标准差以及模糊解识别正确率等3个指标方面有所提高,特别是模糊解识别错误的数目有较明显的下降（由289下降至240,下降约17%）,这预示着新方法有较好的应用前景。

英文摘要：

There are two bias assessment methods for scatterometer wind retrieval：1）Compare the data with those from other sources,for example,with the numerical weather product（NWP）or the buoy wind data;2）Check the coherence of wind fields.In this article,two-dimensional center geometry matrix is adopted to describe the atlas properties of the ambiguity area in the solution space of wind cell,for example,the direction of long axis,the ratio of long/short axis,and the total gray value of the picture ellipse,then the relation between the total gray value and the retrieval biases of wind vector is discussed.The results show that the retrieval biases of wind vector will be bounded in a limited range when the total gray value is larger than a threshold.According to that,a new filter algorithm,which is called gray value constraint filter,is developed based on the property of the constraint effect of the total gray value of picture ellipse of the wind vector solution space on the retrieval biases of wind vector.The comparison with buoy wind data shows that the new filter algorithm is better than the algorithms of circle medium filter CMF）in the accuracy of retrieval wind direction,the standard deviation of wind direction bias and the accuracy of the identification of the ambiguity are improved,especially in the performance of identifying the ambiguity,the wrong identification number decreases from 289 for CMF to 240 for new filter,decreasing by approximately 17%.It is expected that the new filter algorithm would have a good application prospect.