基于梯度信息的微波辐射亮温资料质量控制方法

A quality control method based on gradient information of microwave radiation data AMSU-A

卫星微波垂直探测器的辐射观测资料在数值预报中的同化应用使得数值预报水平有了巨大的飞跃。微波资料的质量控制是保证观测资料成功同化的关键所在。文章提出一种基于AMSU-A(Advanced Microwave Sounding Unit-A)辐射亮温资料梯度信息的新质量控制方法,将亮温梯度距平值明显较大的资料作为被降水污染或因为其他原因出现的"坏"的资料。利用中尺度非静力WRF(Weather Research and Forecasting)模式和区域三维变分同化,针对"海鸥"(2008)和"圆规"(2010)2个个例,对比旧质量控制中的降水检测和阈值检测方法,评估该方法用于AMSU-A资料同化时对台风数值模拟的情况。研究表明,旧质量控制方法将会使一些"坏"的微波观测资料同化进入模式,降低模式分析场的质量,进而导致同化结果有较大误差。相对于旧方法获得的分析场,利用基于亮温梯度信息的质量控制方法可使更多"坏"的观测剔除,同化后模式初始时刻的位势高度场和风场更接近于真实情况。与传统AMSU-A辐射资料的同化相比,新质量控制方案使2个台风路径数值模拟的偏差有明显的减小:"海鸥"个例中,模拟台风路径误差的最大改善比为12,路径误差改善约540km;"圆规"个例的最大改善比为13,模拟路径误差减小118km。

Quality control (QC) is a key for microwave sounding radiance data assimilation in numerical weather prediction (NWP) models. In this study, a new approach of QC for Advanced Microwave Sounding Unit-A (AMSU-A) radiance was developed by introducing brightness temperature gradient of view point. In this approach, the data were considered to be contaminated by precipitation if their gradients were anomalously large, and would be discarded. Two typhoon cases,Kalmaegi in 2008 and Kompasu in 2010, were simulated by WRF (Weather Research and Forecasting) model to see the impacts of the new QC method on AMSU-A radiance. In the new QC cases, as the contaminated data removed from the three-dimensional variational assimilation (3DVar), the tracks of both typhoons were much closer to the observations due to the increments of initial geopotential height and wind fields. In addition, the typhoons’ intensities were much closer to the observations with more reasonable temperature field for the warm structure of typhoon core development. In the original cases,with much more contaminated AMSU-A radiance in WRF, both typhoons’ tracks and intensities were much worse simulated.An improvement ratio to simulated track error was 12, that is an improvement to 540 km compared to the track without AMSU-A radiance assimilated, in the Kalmaegi case. And the ratio was 13 with 118 km improvement to simulated track in the Kompasu case.