基于雷达回波概率特征的雷达部分遮挡区域识别算法

Radar Partial Shielding Region Identification Algorithm Based on the Probability Characteristics of Radar Echoes

部分遮挡是影响雷达观测资料质量的重要误差源之一。基于雷达回波的概率特征,本文提出一种不依赖高精度地形信息的雷达部分遮挡区域识别算法。该算法无须以人工方式建立样本数据库,只需直接检测实际业务应用中雷达连续观测的反射率因子,统计雷达探测范围内的每个格点出现大于某一反射率因子阈值(定义为Z阈值)的概率,并通过设置适当的概率阈值(定义为C阈值)来确定不同遮挡性质的雷达探测区域。应用该算法,使用2012年5-10月金华、衢州及上饶雷达连续探测资料,识别得到金华、衢州和上饶三部雷达距离地面3 km高度层上CAPPI的完全遮挡区域、部分遮挡区域和无遮挡区域;同时,通过对比2013年4月29日金华、衢州、杭州、上饶、黄山雷达3 km高度重叠探测区域的雷达反射率因子,检验了雷达部分遮挡区域识别结果的有效性,并分析了算法阈值的敏感性;最后,结合剔除部分区域弱回波的方案,对雷达组网拼图算法进行优化改进。结果表明:通过选择适当的Z阈值和C阈值算法,可以有效识别雷达部分遮挡区域;算法阈值的设置是识别部分遮挡区域的关键,较为适中的Z阈值有利于C阈值的选择;优化后的组网拼图算法可以有效解决部分遮挡问题,有效地提高了组网拼图数据的质量。

Partial blockage is one of the most important error sources of radar observation. A new shielding region identification algorithm without any dependence on the terrain information with high resolution is proposed.The algorithm first computes the probability of every grid's probability of emerging specified reflectivity exceeding a threshold which is defined as Z threshold only through checking the continuous operational radar observation data without using any sample database established manually,and then selects adequate probability threshold which is defined as C threshold to delineate the region with different shielding characters. Applying the algorithm with the continuous operational observation dataset during the period betw een May and October in 2012 at Jinahua Quhou and Shangrao,the completely shielding region,partially shielding region and no shielding region on 3 km CAPPI are derived separately. With the comparison of the observation data at April 29 in 2013 in the overlapping region on the 3 km CAPPI betw een Jinhua,Quzhou,Shangrao,Hangzhou and Huangshan,the validation of the identification results of the three radar sites are tested and the parameter sensibility of the Z and C threshold is analyzed,too. The multi-radar mosaic algorithm is optimized applying the partial shielding region results of the radar sites above to enhance the quality of multi-mosaic. The results show that the shielding region with different characters can be identified effectively by selecting adequate Z and C threshold; the selection of the Z and C threshold is the key and moderate Z threshold is beneficial to the further selection of the C threshold; the partial shielding effects can be easily eliminated based on the identification results and the data quality of multi-radar mosaic can be enhanced effectively based on the optimized multi-radar mosaic algorithm.