Hurricane Boundary Layer Wind Profiles Estimated From Satellite SAR Images

Satellite synthetic aperture radar (SAR) imagery of normalized radar
cross-section (NRCS) provides an image of the cm-scale sea-surface
roughness as illuminated by a side-looking radar beam over a range of
incidence angles at very high spatial resolution (25 to 100 m
pixels). The NRCS depends on the incidence angle, the relative azimuth
angle between the radar beam and the horizontal wind direction and on
the surface wind speed. This relationship is described by geophysical
model functions (GMFs). Given wind directions, the surface wind speed
can be determined using the GMF. Nearly always, boundary layer roll
vortices over a wide range of scales impress a signature associated
with their modulation of the surface stress on the sea surface that
can be extracted using image processing or local Fourier analysis. The
resulting surface wind vector retrievals are of good quality and can
be produced routinely at 1 km spacing. As previously demonstrated, it
is possible to make good estimates of the sea-level pressure gradients
from these surface wind vectors using a boundary layer
model. Least-squares optimization is used to estimate a sea-level
pressure pattern from these surface pressure gradient vectors. Thus,
from a SAR image we can extract good estimates of both the upper and
lower boundary conditions on the tropical cyclone boundary layer
(TCBL).

We apply these boundary conditions to a diagnostic, nonlinear
similarity model of the TCBL in order to produce vertical profiles of
the mean wind. The similarity TCBL model can incorporate a wide range
of local turbulence closure models that are consistent with standard
numerical model parameterizations. Different closures result in
significantly different mean wind profiles. If it is possible to
validate the wind profile extraction methodology, we can assess the
usefulness of the different turbulence closures. Drop sondes provide
single realizations of the wind profiles acquired over the drop
trajectory during the drop time and cannot be considered as the
vertical mean wind profile. This makes it difficult to validate
co-located SAR-extracted wind profile individually. We will present
the SAR wind profile extraction technique and preliminary attempts at
validation.