It is often easier to use data that has been processed into a
regular latitude-longitude grid. Data from polar orbiting satellites
is archived in swaths, relative to the satellite track, rather than a
global grid. Satellite data must be averaged in places where tracks
cross, and gaps in coverage must be filled. The problem with gaps
could be solved by averaging over a sufficiently long time, but many
researchers are interested in averages over shorter times. The
rapidly growing use of satellite observations has lead to many
techniques for averaging and/or filling gaps. We have generated
gridded wind products using several of these techniques.
A pyramidal averaging in time has been used to generate 1x1°
(global fields of daily winds. For a detailed description of the
technique see
how we created wind fields.
Optimally Determined Pseudostresses
|
Pseudostress is the product of the wind speed and the wind vector.
A variational technique (direct minimization) was used to optimize
fields of NSCAT pseudostress. The weights used in the optimization
were optimized through cross validation. This technique smoothly
blends observations with a background field. In this case, the
temporally average wind products were used as backgrounds. The
available fields are 1x1° (resolution daily fields for the Indian Ocean.
Wind stress can be determined directly from scatterometer observations.
Strictly speaking friction velocity (the square-root of the
kinematic stress) is determined, and then it is converted to a stress.
Half degree resolution field were determined by averaging the stresses
for each month.
