
The VAM requires a background (first guess) analysis of gridded U and V winds as a starting estimate of the wind field. Analysis increments are added to this background to arrive at the final analysis. Background analyses are obtained from the Computation and Information Systems Laboratory (CISL) at the National Center for Atmospheric Research (NCAR).
Two product streams are produced by this project : First-look (FLK) and Late-look (LLK). FLK products use the ECMWF Operational Analysis (DS111.1) as a background and are available within 6-months of real-time. The advantage of this product is that it is continuously updated. The disadvantage is that it is subject to changes made to the ECMWF operational system. A description of these changes can be found at: http://www.ecmwf.int/products/data/operational_system/evolution/.
LLK products use the ECMWF ERA-40 Reanalysis (DS117.0) as a background. Processing for this product is determined by the availability of ERA-40 products. At the time of writing this document, ERA-40 reanalyses were only available through June 2002. The advantage of this background, however, is that it uses a consistent analysis system for the entire period.
It is important to note that 10-meter winds from the ECMWF analyses are not neutral winds. Overall, the ECMWF wind speeds are biased low relative to the satellite observations by about 0.6 m/s. Of this, approximately 0.2 m/s can be attributed to the effects of stability (Chelton, Freilich 2005). There are no attempts to remove this bias in the FLK products. However, the resulting VAM analyses are nearly unbiased (see Figure 1) over the global oceans due to the overwhelming number of satellite ocean surface wind observations during the first-look period (1999-2007).
Satellite coverage is often sparse during the early years of the late-look period (1987-2002). There are large portions of the global oceans that are not observed within a 6-hour period. For the LLK products, the ERA-40 10m winds will be converted to equivalent neutral winds to derive the background for assimilating observations. Methods to remove the remainder of the bias are currently being examined. Check the LLK product descriptions for detailed information on the methods used for each release of the data.