Gd. Alexander et al., USE OF DIGITAL WARPING OF MICROWAVE INTEGRATED WATER-VAPOR IMAGERY TOIMPROVE FORECASTS OF MARINE EXTRATROPICAL CYCLONES, Monthly weather review, 126(6), 1998, pp. 1469-1496
A technique is described in which forecasts of the locations of featur
es associated with marine cyclones may be improved through the use of
microwave integrated water vapor (IWV) imagery and image warping of fo
recast mesoscale model fields. Here, image warping is used to optimall
y match mesoscale model output to observations of IWV measured by micr
owave sensors. In the mesoscale model simulations presented here (one
of the March 1993 ''superstorm,'' one of a rapidly deepening cyclone o
bserved in the North Atlantic in February 1992, and one of the ERICA I
OP 4 cyclone), the Pennsylvania State University-National Center for A
tmospheric Research MM5 model is initialized from the standard Nationa
l Meteorological Center (recently renamed the National Centers for Env
ironmental Prediction) operational analysis. The simulations are then
run until a time at which a Special Sensor Microwave/Imager (SSM/I) ov
erpass occurs. For each simulation, the forecast pattern of IWV is the
n compared to the field shown in the SSM/I image. In all three cases,
the MM5 moves the cyclones too slowly, and therefore places distinguis
hing features in the forecast IWV fields significantly upstream of the
ir locations as revealed in the microwave imagery. To rectify these er
rors, the grid on which the source image (forecast held) is defined is
then warped to match the target image (remotely observed IWV field) b
y choosing pairs of tie points corresponding to similar features in th
e two images. The values of all model moisture variables at all vertic
al levels are then carried to the new warped grid points and interpola
ted back to the original model grid. Model integration then proceeds w
ith the new model fields. The model results at a subsequent time after
the warping is applied are then compared with simultaneous model resu
lts in simulations in which no warping was applied as well as with mod
el simulations in which a standard nudging technique is applied. Warpi
ng results in improved forecasts of cyclone minimum sea level pressure
, tracks, and IWV fields over both the control simulations and the nud
ged simulations.