Predicted winds for Chesapeake Bay from LFM and observational data

"Observed overwater winds from two stations on Chesapeake Bay have been filtered to remove high freguency variability and compared to LFM output at nearby grid points. Stepwise linear regression was then used to develop predictive eguations for the overwater winds using modeled winds and pressures for input. With three LFM variables, regression equations explained from 64 to 71 percent of the variance in the u and v wind components and from 50 to 52 percent of the wind speed variance for Holland Island Bar Light and York Spit stations."