AMSU Balance Winds
This method begins with a statistical retrieval technique which is applied to the AMSU-A radiances to provide temperature profiles of the atmosphere. Using NCEP 100 hPa height and temperature analyses as an upper boundary condition, the hydrostatic equation is integrated downwards using the AMSU temperature profiles to give the geopotential height as a function of pressure. Assuming a geostrophic, linear, or nonlinear balance exists between the height field and the streamfunction, the nondivergent wind field may be retrieved from the height field.
Temperature retrievals derived from AMSU-A radiances over the Arctic north of 60°N were collected for both NOAA-15 and NOAA-16 polar orbiting satellites for two periods. A winter case consists of data from 2-17 December 2004 and a summer case from 13-31 July 2007. Radiosonde data from Arctic stations were also collected in order to verify the results of the satellite-based technique.
Plotted above are the wind speed bias (solid) and vector root mean square error (dashed) of the geostrophic, linear-balance, and nonlinear-balance winds derived using the AMSU technique when compared to the winds measured by radiosondes. The results are for matches between the satellite technique and the radiosonde within 1.5 hours and 200 km. Also shown is the mean radiosonde wind speed (solid blue) as a function of pressure. Plotted on the right-hand side of the figures is the number of matchups at each level. The behavior of the bias and RMSE with height is similar in the winter and the summer: at 1000 hPa the bias is 3-5 m s-1, then becomes negative away from the surface with a minimum of -3 to -5 m s-1 near the jet level. Above 250 hPa the bias moves back towards zero. The RMSE has minima in the free troposphere and above 200 hPa, and maxima near the surface and near the jet level.