基于精密单点定位的大气可降水量反演及精度分析

摘要:

采用精密单点定位算法对河池、宝山、海口三站地基全球卫星导航系统(Global Navigation Satellite System，GNSS)资料开展6组卫星数据质控方案对大气可降水量(Precipitable Water Vapor，PWV)反演试验，基于探空资料和中国气象局PWV业务产品对反演结果进行精度检验并对比分析各种PWV资料在降水过程中的变化特征。结果表明：卫星数据质控对反演效果影响明显，对残差sigma＞2的卫星数据进行剔除能有效改进反演效果，与探空、PWV业务产品的相关系数可提升0.03～0.25，均方根误差可减少0.54～10.7 mm；对进入滤波器前残差＞3倍误差的卫星数据以及进入滤波器后sigma＞2的卫星数据进行剔除后，反演效果最佳，与探空、PWV业务产品的平均相关系数超过0.93，平均偏差为-1.26～-1.91 mm，具有较高的反演精度，可弥补常规PWV资料时间分辨率偏低的不足；反演的PWV对短时强降水发生指示作用较其它两种明显，降水发生前后10－30 min内PWV迅速增加和减少约10～15 mm，这一现象可以作为局地短时强降水预报的辅助判别条件。

Abstract:

This study investigates the retrieval of atmospheric precipitable water vapor (PWV) through the application of the Precise Point Positioning (PPP) algorithm utilizing the Global Navigation Satellite System (GNSS) data from Hechi, Baoshan, and Haikou stations. We conducted six experimental tests to analyze the effects of satellite data quality control on PWV retrieval accuracy, comparing our findings with sounding data and operational PWV products from the China Meteorological Administration. The results indicate that implementing effective quality control substantially improves inversion accuracy, eliminating satellite data with residual error more than 3 times before entering the filter and residual sigma values greater than 2 after entering the filter, which enhances correlation coefficients with sounding or PWV products by 0.03 to 0.25 while decreasing root mean square error by 0.54 to 10.7 mm. Optimal inversion accuracy is obtained by filtering out satellites with residual errors exceeding three times the threshold prior to analysis, resulting in an average correlation coefficient above 0.93 and a mean error ranging from -1.26 to -1.91 mm. This approach successfully mitigates the challenges associated with the low temporal resolution of conventional PWV data. The PWV retrieved via the PPP method demonstrates a enhanced sensitivity to short-term heavy precipitation events compared to other products, exhibiting rapid changes of 10 to 15 mm within 10 to 30 minutes before and after the rainfall,which can be used as an auxiliary discriminating condition for local short-term heavy precipitation forecast.