||SSM/I Data Validation
The validation results presented below pertain to SSM/I version 4 results described in two publications:
The above two articles are in PDF format. You will need a PDF viewer such as Adobe Acrobat Reader in order to view the files online. These articles have been published in the following journals:
Mears,C., D.Smith, and F.J.Wentz. 2001, "Comparison of SSM/I and buoy-measured wind speeds from 1987 - 1997", Journal of Geophysical Research, Vol. 106(C6), 11719-11729.
Meissner,T., D.Smith, and F.J.Wentz. 2001, "A 10-year intercomparison between collocated SSM/I oceanic surface wind speed retrievals and global analyses", Journal of Geophysical Research, Vol. 106(C6), 11731-11742.
As the version 5 algorithm has been derived from version 4, we anticipate similar validation results and expect to see an improvement in the boundary current regions located in the GCM comparisons.
RSS performs extensive algorithm validation to determine the quality of our data. We maintain a dataset of over 10 million buoy observations to use in validating the SSM/I products we produce.
Validation of SSM/I Wind Speed
In order to validate the SSM/I derived wind speeds, we compare our wind speed products with in situ measurements of wind speed made by a large set of buoys moored in a number of regions in the worlds oceans, and with surface winds estimated by two GCMs. This provides an important check of our retrieval algorithms, and also helps to identify any problems that might occur with the satellite instrumentation.
Comparison with in situ wind speeds
We obtained buoy-measured winds speeds from a set of over 100 open-ocean buoys that regularly report hourly winds. Buoys are located in the Tropical Pacific Ocean (most of these are members of the Tropical Atmosphere Ocean Project (TAO) array) and along the coast of North America (administered by the National Data Buoy Center (NDBC)). Over the study period (1987-1997), there were over 285,000 overflights of buoys by SSM/I instruments that have valid wind data for both the buoy and an SSM/I instrument. Buoy wind speeds were corrected to a standard height of 10 m assuming a simple logarithmic vertical wind speed profile. The SSM/I wind speed data were collocated with the buoy measurement and quality-controlled for contamination by rain or ice. We found that the mean difference between SSM/I and buoy winds is typically less than 0.4 m/s when averaged over all buoys, and the standard deviation is less than 1.4 m/s. Global means for each satellite are presented in the table below:
(WSSM/I – WBuoy)
(WSSM/I – WBuoy)
Mean errors for a given buoy-satellite pair typically range from –1.0 m/s to +1.0 m/s, with standard deviations less than 1.4 m/s. Examination of a scatter plot for an individual buoy (Figure 1.) revealed a high degree of correlation between SSM/I and buoy wind speeds.
We show the geographic distribution of mean differences in Figure 2. Here, color coded squares are plotted at each buoy location indicating the mean SSM/I-Buoy wind speed difference for that buoy. We observed small regional biases, particularly in the trade wind regions, where SSM/I winds are 0.5 to 1.0 m/s higher than those measured by buoys.
Comparison with General Circulation Model Winds
We collocated modeled winds from the NCEP reanalysis and the ECMWF TOGA analysis to SSM/I wind using trilinear interpolation of the model outputs in both time and space. Global averages of the wind speed for SSM/I were typically 0.3 to 0.7 m/s higher than those produced by the models with the spatial standard deviation calculated over maps of monthly mean differences being about 1.2 to 1.4 m/s. Regional biases tended to match those found by the buoy-SSM/I comparisons, except for significant areas of low-biased SSM/I winds over the boundary-current regions at the eastern side of the ocean basins (where almost no buoys are located far enough from land to be used in this study), and a region of high biased SSM/I winds in the equatorial Pacific. Subsequent comparison of buoy wind speeds to those produced by the models indicated that the modeled wind speeds are generally too low in this equatorial region. SSM/I winds are also higher than modeled winds in the southern trades.