辐射传输模式对地基微波辐射计观测亮温的模拟能力分析

Assessment of Radiative Transfer Models Based on Observed Brightness Temperature from Ground-Based Microwave Radiometer

提供高时间分辨率大气温度湿度廓线的地基微波辐射计近年来广泛使用,多通道观测亮温的数据质量是大气廓线产品合理性的基本保障。一般定期液氮绝对定标可以更好维护亮温数据质量,但实际操作颇为不易。辐射传输模式作为一种辅助工具,可以检验和认识地基微波辐射计观测亮温的数据质量。本文针对三个辐射传输模式:MonoRTM、ARTS和MWRT,结合北京探空观测资料、北京观象台和河北香河站同类型的德国RPG地基微波辐射计观测资料,分析比较了三个模式的模拟与观测亮温差异,评估不同辐射传输模式对地基微波辐射计观测的模拟能力。地基微波辐射计14个通道观测亮温与模式模拟的差异统计比较发现:三个模式的模拟结果与地基微波辐射计大部分通道的观测亮温都很接近,与观测结果具有很好一致性(如相关系数高达0.99),而对温度通道ch8(51.26 GHz)和ch9(52.28 GHz),三个模式模拟与观测相关系数明显较低(<0.80),并且存在显著的绝对偏差(4~5 K),表明模式在这两个通道的模拟能力有待提高。三个模式中,MonoRTM模式在温度通道ch8、 ch9和ch10(53.86GHz)存在明显的系统性偏差,尤其是ch8高达5K;ARTS模式对水汽通道ch1(22.24 GHz)的模拟能力相对较弱;MWRT模拟与观测亮温在多个通道上相对更为接近和稳定,尤其系统性偏差最小。此外,探空廓线与地基观测站的空间位置不一致,对地基微波辐射计水汽通道的模拟结果影响较为显著,而对水汽不敏感的温度通道影响甚微。两地观测亮温与模式模拟的比对,初步表明北京观象台地基微波辐射水汽通道的观测质量有待改进。

Ground-based microwave radiometers(MWRs) have been widely used in recent years due to the fact that they provide atmospheric temperature and humidity profiles with high temporal resolutions. The quality of the multi-channel brightness temperature(TB) is the measure used to determine the quality of retrieved atmospheric profile products. In general, periodic absolute calibration using liquid nitrogen best maintains the quality of TB observations, but this operation is complex and difficult. As an auxiliary tool, radiative transfer model can be used to determine the TB quality of MWRs. Using observations from the Beijing radiosonde and two RPG MWRs located at the Beijing Observatory(GXT) in Beijing and the Xianghe site(XH) in Hebei, we evaluated three radiative transfer models—MonoRTM, ARTS,and MWRT—by comparing their simulations with those of the corresponding TBs observed at both sites. The results show that for most of the 14 MWR channels the simulations of the three models are highly consistent with the observed TBs(i.e., correlation coefficients up to 0.99), but for the temperature channels ch8(51.26 GHz) and ch9(52.28 GHz),there was a significant absolute deviation(approximately 4-5 K) between the simulated and observed TBs, and the correlation coefficient decreased significantly(<0.80), which indicates that the model simulation at these two channels must be improved. Of the three models, MonoRTM showed obvious systematic deviation at temperature channels ch8,ch9, and ch10(53.86 GHz), particularly at ch8, which had a bias of up to 5 K. The ARTS model generated the worst simulation at the water vapor channel ch1(22.24 GHz). The MWRT simulations were relatively more stable and closer to the corresponding TB observations at 14 channels, with the least systematic deviation. In addition, the locations of the radiosonde measurements differed from that of the MWR site, which had a significant impact on the simulations for the MWR’ s water vapor channels. Comparisons of the observed and simulation TBs at both sites indicates that the observation quality of the water vapor channels for the MWR at GXT must be improved.