Liu Quanhua, Xue Y, Li C
Earth System Science Interdisciplinary Center, University of Maryland, College Park, Maryland 20740, USA.
Appl Opt. 2013 Jul 10;52(20):4981-90. doi: 10.1364/AO.52.004981.
The community radiative transfer model (CRTM) has been implemented for clear and cloudy satellite radiance simulations in the National Oceanic and Atmospheric Administration (NOAA) National Centers for Environmental Prediction (NCEP) Gridpoint Statistical Interpolation data assimilation system for global and regional forecasting as well as reanalysis for climate studies. Clear-sky satellite radiances are successfully assimilated, while cloudy radiances need to be assimilated for improving precipitation and severe weather forecasting. However, cloud radiance calculations are much slower than the calculations for clear-sky radiance, and exceed our computational capacity for weather forecasting. In order to make cloud radiance assimilation affordable, cloud optical parameters at the band central wavelength are used in the CRTM (OPTRAN-CRTM) where the optical transmittance (OPTRAN) band model is applied. The approximation implies that only one radiative transfer solution for each band (i.e., channel) is needed, instead of typically more than 10,000 solutions that are required for a detailed line-by-line radiative transfer model (LBLRTM). This paper investigated the accuracy of the approximation and helps us to understand the error source. Two NOAA operational sensors, High Resolution Infrared Radiation Sounder/3 (HIRS/3) and Advanced Microwave Sounding Unit (AMSU), have been chosen for this investigation with both clear and cloudy cases. By comparing the CRTM cloud radiance calculations with the LBLRTM simulations, we found that the CRTM cloud radiance model can achieve accuracy better than 0.4 K for the IR sensor and 0.1 K for the microwave sensor. The results suggest that the CRTM cloud radiance calculations may be adequate to the operational satellite radiance assimilation for numerical forecast model. The accuracy using OPTRAN is much better than using the scaling method (SCALING-CRTM). In clear-sky applications, the scaling of the optical depth derived at nadir for brightness temperature calculation at the other direction may result in an error up to about 7 K for some HIRS/3 channels. Under cloudy conditions, SCALING-CRTM may result in an error of about 3.5 K.
社区辐射传输模型（CRTM）已在国家海洋和大气管理局（NOAA）国家环境预测中心（NCEP）网格点统计插值数据同化系统中实现，用于全球和区域天气预报以及气候研究的再分析中的晴空和多云卫星辐射模拟。晴空卫星辐射已成功同化，而多云辐射需要同化以改善降水和恶劣天气预报。然而，云辐射计算比晴空辐射计算慢得多，并且超出了我们天气预报的计算能力。为了使云辐射同化可行，在CRTM（OPTRAN-CRTM）中使用波段中心波长处的云光学参数，其中应用了光学透过率（OPTRAN）波段模型。这种近似意味着每个波段（即通道）只需要一个辐射传输解，而不是详细的逐线辐射传输模型（LBLRTM）通常所需的一万多个解。本文研究了这种近似的准确性，并帮助我们了解误差来源。本研究选择了两个NOAA业务传感器，高分辨率红外辐射探测器/3（HIRS/3）和先进微波探测单元（AMSU），包括晴空和多云情况。通过将CRTM云辐射计算与LBLRTM模拟进行比较，我们发现CRTM云辐射模型对于红外传感器可以达到优于0.4 K的精度，对于微波传感器可以达到优于0.1 K的精度。结果表明，CRTM云辐射计算对于数值预报模型的业务卫星辐射同化可能是足够的。使用OPTRAN的精度比使用缩放方法（SCALING-CRTM）要好得多。在晴空应用中，为了在其他方向计算亮温而对天底处导出的光学厚度进行缩放，对于某些HIRS/3通道可能会导致高达约7 K的误差。在多云条件下，SCALING-CRTM可能会导致约3.5 K的误差。
