Evaluation of a New Method to Quality Control Satellite-derived Polar Winds in the NCEP GDAS/GFS

Atmospheric Motion Vectors (AMV) are routinely generated from geostationary and polar orbiting satellites and they are incorporated into most global numerical weather prediction models. The AMV thinning strategies vary in the modeling community, but they are usually based on thresholds of the quality flags from the winds producers and the deviation of the AMV from the model background.

In the NCEP Global Data Assimilation System/Global Forecast System (GDAS/GFS), Moderate Resolution Imaging Spectroradiometer (MODIS) polar winds are discarded if the zonal or meridional component of the wind deviates more than the threshold of 7 m/s from the background state. Since wind speeds vary across three orders of magnitude, using this fixed threshold has the disadvantage of retaining too many slow winds (regardless of direction) and discarding high-speed winds that do not deviate very much (as a percentage) from the background. To alleviate these issues, a new method is being evaluated which discards observations based on the AMV vector departure from the background, normalized by the logarithm of the AMV speed: the Log Normalized Vector Departure (LNVD). For AMVs to be retained, the LNVD method requires slower speed winds to be in better agreement with the background, while allowing higher speed winds to deviate more, as compared to the current fixed threshold.

Results of current GDAS/GFS experiments using the LNVD to screen MODIS and Advanced Very High Resolution Radiometer (AVHRR) polar AMVs will be discussed.