Testing Less Restrictive Quality Control Measures for Satellite Winds in NAVGEM

Satellite-derived winds are one of the most beneficial data categories in the U.S. Navy’s global numerical weather prediction model, NAVGEM (Navy Global Environmental Model). The operational suite (NAVGEM v1.4, T425L60) includes quality control measures that screen out satellite wind observations according to producer-assigned quality indicators, but also apply a number of other screening measures that are re-examined here.  These measures include disallowing atmospheric motion vectors between 425 hPa and 625hPa, or below 975 hPa; masking over some land regions; and rejecting observations with wind speeds which differ from the background forecast field by more than 8-12 m/s as a function of pressure level. It should be noted that despite this rather heavy-handed filtering, satellite winds rank at the top in the amount of forecast error reduction they provide to the system.  However, the recent availability of a large number of high quality and high resolution winds from Himawari-8 and Meteosat-10 has called into question the practice of excluding many of the available observations.  Reducing the number of observations that are screened out in quality control checks that precede the variational analysis system will bring in observations at higher spatial resolution, and should allow us to take more full advantage of both the 4D-VAR and ET (ensemble transform) techniques, since NAVGEM’s data assimilation system, NAVDAS-AR (Naval Research Laboratory Atmospheric Variational Data Assimilation System – Accelerated Representer), is a hybrid of the two methods.  Admitting observations from the blocked-out pressure levels will increase spatial resolution in the vertical, but also in the horizontal, as these levels of the atmosphere are currently a relative data void.  NAVDAS-AR applies a 3-sigma check on all observations, setting them aside while a precursory analysis is made, then rechecking them against the adjusted background.  This 3-sigma check should be restrictive enough to prevent harming the forecast due to large observation errors.  (Some have even argued that the 3-sigma check itself can be too restrictive. Tavolato and Isaksen, 2015.) This presentation will examine the characteristics of satellite wind data with less restrictive screening measures, describe the impact on NAVGEM of assimilating more large innovations, and recommend changes and monitoring strategies.  Increasing the amount of information that is used from Himawari-8 and Meteosat-10 will improve forecasts now and also set the stage for making optimum use of information from other next-generation meteorological satellites, such as GOES-R.