面向资料同化的FY-4A卫星GIIRS探测仪偏差特征分析和偏差订正

Bias Characteristics and Bias Correction of GIIRS Sounder onboard FY-4A Satellite for Data Assimilation

干涉式大气垂直探测仪(Geostationary Interferometric Infrared Sounder,简称GIIRS)是国际上第一部对地静止卫星平台上的高光谱红外大气垂直探测仪,能为对流尺度区域模式预报提供所需的高时空和高光谱分辨率的大气状态信息。本文利用高分辨率区域模式WRF及其同化系统WRFDA对GIIRS观测的偏差(观测亮温减去模拟亮温,记为O-B)分布特征进行了全景分析,结果表明:长波通道O-B偏差和标准差普遍小于中波通道,且都存在一段受污染的通道。O-B偏差的日变化和偏差与卫星天顶角的关系相对较弱,而所有筛选通道的偏差都与亮温值及卫星的扫描阵列位置有关,偏差的水平分布主要表现出“阵列偏差”的特征。2020年重新定标后,GIIRS观测数据质量比2019年有明显提高。在此基础上进一步进行了偏差订正试验,试验发现选取扫描阵列作为偏差订正的主要因子,都能有效地改进2019年和2020年筛选出的GIIRS通道的偏差,订正后O-B和O-A的系统性误差(偏差)都变小。该研究结果可为全球/区域模式中同化GIIRS长波及中波通道的辐射资料提供参考。

The geostationary interferometric infrared sounder(GIIRS) is the first hyperspectral infrared(IR) sounder onboard the geostationary meteorological satellite— FengYun-4A satellite. The GIIRS can provide atmospheric state information with a high temporal, spatial, and spectral resolution that could enhance the prediction of convective-scale regional models. This paper presents a panoramic analysis of the bias characteristics of the GIIRS with the high-resolution regional model WRF and its assimilation system WRFDA. Results reveal that the bias(observations minus simulations, or O-B) and standard deviation of long-wave infrared(LWIR) channels are generally smaller than those of middle-wave infrared(MWIR) channels, and both have some contaminated channels. The diurnal variation of biases is relatively unremarkable, and the O-B bias has a weak relationship with the satellite zenith angle. However, the biases in all selected channels are related to the value of brightness temperature observations and the scanning position of the satellite. In addition, the spatial distribution characteristics of the bias are relevant to the bias characteristics impacted by the scanning position. After re-calibration of the GIIRS in 2020, the quality of the GIIRS observation data is significantly improved compared with that in 2019. Based on this, further bias correction tests are carried out using correction factors relevant to the scanning position. Results reveal that both the systematic errors(bias) of O-B and O-A have been significantly reduced, implying the effectiveness of the bias correction scheme. This study provides implications for GIIRS radiance assimilation in global/regional models.