QSCAT/NCEP blended winds (version 4) were extracted from the global data files for the Mediterranean Sea, including the Black Sea and about 10° into the Atlantic: 19° W to 42° E, and 30° to 47°N.
For more information, see CoRA's Blended Winds, or NCAR's archived DS744.4 - QSCAT/NCEP Blended Ocean Winds.
Sample plots of zonal and meridional winds are shown below, including sample 0.5° x 0.5° ECMWF analysis
forcing data, provided by the Mediterranean
Forecasting System, MFS, at the Istituto Nazionale di Geofisica e Vulcanologia.
Click on the images for enlarged versions of plots.
The QSCAT/NCEP blended winds are based on QSCAT scatterometer data. These satellite data were
used to compute spectra based on 10° across-swath tracks. Two sample QSCAT revolutions
(no. 8715 and 8716) are shown below in the Mediterranean Sea. The data resolution is 25 x 25km.
Whenever possible, cross-swath tracks were selected for the Eastern and Western Mediterranean.
Two of those are marked with red arrows.
Whenever there are more than two consecutive observations missing along the 10° track (e.g. due to rain contamination at wind speeds < 15 m/s), the track was not sampled.
Those tracks were collected for all years (2000-2003). On average, there are about 5300 tracks
per year. Sample locations of those tracks are shown below for a 7 day period in March 2003,
when 109 tracks were collected.
In addition to QSCAT 10° tracks, also NCEP derived tracks were extracted. These data
are based on the NCEP operational analyses (so-called "final analysis or NCEP-FNL), that
have been interpolated in space and time to the wind vector locations for each QSCAT
swath (provided M.Freilich and Barry Vanhoff, Oregon State University).
For comparison, also the ECMWF data were sampled along 10° tracks. The picture below shows ECMWF's 0.5° grid locations, and the 11 tracks sampled from each 6-hourly field.
Spectra for all 3 data sets (QSCAT revs, NCEP collocated, and ECMF) are shown below.
On the plot of ECMWF spectra, bootstrap confidence intervals are indicated for the year
2000 power spectral density (PSD) averages. The data were partitioned into ten subsets and
average spectra were computed for each subset. The vertical bars indicate the range of the
ten subset averages.