Global Evaluation of the Fidelity of Clouds in the ECMWF Integrated Forecast System

Weather forecasting centers mainly assimilate infrared sounder data in clear-conditions or in channels with their main sensitivity to the atmosphere above the cloud tops. Sometimes channels with stronger cloud sensitivity are used in overcast conditions, but currently no cloud information is used from infrared sounders, and all-sky assimilation approaches are still under development. However, cloudy radiances could already be used for validating the quality of clouds in forecasts in a passive assimilation. We illustrate this by comparing the radiances observed (obs) with the AIRS (Atmospheric Infrared Sounder) to calculated radiances (cal) based on the clouds specified in the ECMWF (European Centre for Medium Range Weather Forecasting) Integrated Forecast System (IFS). Our analysis is based on a 12 hour ingest of AIRS data into the ECMWF assimilation system. We show that the standard deviation of (obs-cal) using the 1231 cm-1 atmospheric window channel is a metric of the fidelity of the clouds in the IFS. The global standard deviation of 5 K appears to be dominated by clouds in the IFS which are not seen in the AIRS data, and vice versa. Our metric capitalizes on the unique sensitivity of infrared sounders to clouds for the routine monitoring of the fidelity of clouds in weather forecasts.气象预报中心主要在晴空条件下，或选用对云顶以上大气具有主要响应的探测通道，开展红外探测仪（infrared sounder）的数据同化工作。部分多云天气场景下，会采用对云体响应更强的探测通道，但当前尚未直接使用红外探测仪获取的云信息，全天空同化（all-sky assimilation）方法仍处于研发阶段。不过，借助被动同化（passive assimilation）技术，已可利用有云条件下的辐射亮温数据，验证气象预报中云系的模拟质量。本研究通过对比大气红外探测仪（Atmospheric Infrared Sounder，简称AIRS）获取的观测辐射亮温（observed, obs）与基于欧洲中期天气预报中心（European Centre for Medium Range Weather Forecasting，简称ECMWF）集成预报系统（Integrated Forecast System，简称IFS）指定云系参数计算得到的模拟辐射亮温（calculated, cal），对上述思路进行了验证演示。本次分析基于向ECMWF同化系统导入12小时时长的AIRS探测数据完成。研究表明，利用1231 cm⁻¹大气窗口通道计算得到的（obs-cal）标准差，可作为衡量IFS云系模拟保真度的量化指标。全球范围内5 K的标准差异常，主要由ECMWF IFS模拟出但AIRS未探测到的云系所主导，反之亦然。本研究所提出的指标，依托红外探测仪对云体的独特响应特性，可用于日常监测气象预报中云系模拟的保真程度。