The Assimilation of Cloud-Affected Infrared Radiances in Fully-Cycled Data Assimilation Experiments

Traditionally in data assimilation, infrared radiances are only assimilated in cloud-free fields of view. This can exclude as many as 85% of measurements in these observations. Recent efforts have extended the Gridpoint Statistical Interpolation (GSI) data assimilation system to include measurements affected by clouds. While this is in theory desirable, there are many difficulties, including, but not limited to, the nonlinear nature of clouds on the observations, the difficulty of detecting multilayer clouds within a single field-of-view, spectral variations in cloud emissivity, and the separation of atmospheric and cloud signatures. Previous efforts have shown this effort has the potential of increasing the number of assimilated observations. The methods, however, have a major sensitivity to cloud heterogeneity. This study will focus on the implementation of previous approaches in a modern Goddard Earth Observing System Model, Version 5 (GEOS-5) atmospheric data assimilation system, of which the GSI is the main component. This presentation will examine the behavior of these measurements on the analysis solution, background feedbacks, and numerical weather forecasts.