Comparisons between geopotential analyses derived from rawinsondes (RAOB) and the VISSR Atmospheric Sounder (VAS) generally exhibit differences that are ultimately related to the horizontal density and placement of the respective observations and the vertical resolution inherent in the instruments. In order to overcome some of the inconsistencies that appear, two strategies have been developed which allow the analyses to communicate through the derived variable, geostrophic potential vorticity. The first incorporates the statistics of RAOB derived potential vorticity into the VAS vorticity analysis. This is accomplished by making a least-squares adjustment to VAS while constraining it to have first and second moments identical to the RAOB analysis. The other approach makes mutual least-squares adjustments to RAOB and VAS vorticity analyses subject to the dynamic constraint that forecast and hindcast of potential vorticity to the time midway between analyses are equal. The forecast and hindcast are made from a two-parameter baroclinic model. In both procedures, the heights are recovered from adjusted vorticities by inverting the elliptic operators that relate height to vorticity.
Data from the GOES-East satellite at 1430 GMT 6 March 1982 are used along with rawinsonde data at 1200 GMT to test the schemes. The statistical adjustment approach makes synoptically meaningful adjustments to the VAS analysis over the Gulf of Mexico and Gulf coast region, but fails to correct the obvious discrepancies over the continental United States. The dynamic scheme succeeds in making meaningful adjustments over both the Gulf of Mexico and the continent which result in improved vertical motion fields.