Three methods which may be used to hindcast the marine surface windfield in extratropical storms were applied in case studies of three recent storms in the eastern north Pacific Ocean. Statistical measures of difference between the alternate windfields and between each windfield and an enhanced data base of buoy and ship reports were computed. The most significant difference between the methods was found to be a negative bias in wind speed fields derived from sea-level pressure gradients. The most accurate method was the direct manual synthesis of ships' wind reports into a continuous field through the application of classical synoptic and continuity techniques and the principles of kinematic analysis. It is the method of choice in hindcast studies involving a small number of storms widely separated in time. A computer-based objective analysis method, which operates jointly on pressure fields and ships' wind observations, produced nearly unbiased windfields, given the ship report observation density available in the eastern North Pacific. For such densities, it is the method recommended in hindcast studies which require long continuous meteorological series.
As a part of the process of selection of environmental design data for offshore structures to be placed in regions exposed to severe storms, data on extreme waves, currents and surge are generated, usually by the application of hydrodynamic models in historical hindcast studies. The need to describe accurately the temporal and spatial distribution of the surface wind over the sea arises both in the calibration of the hydrodynamic models and in the application of the models to long meteorological series in order to provide a data base of environmental extremes for statistical evaluation.
Specification of the most accurate windfield description possible is especially crucial in the calibration studies. There, it is important to isolate errors in predictions of environmental factors (waves, currents, surge) which are related to model physics or numerics, from those errors introduced by deficiencies in the windfield. Of course, accurate windfield specification is desired in long-term historical simulations, though the sparsity of early historical meteorological data places an upper bound on attainable accuracy. In lieu of high accuracy it becomes important to determine the errors which characterize historical windfield descriptions, and the impact of those errors on the accuracy and reliability of environmental series generated by hindcast models. While much work has been reported recently on development, calibration and application of hindcast models, the specification and validation of windfields has received little attention, at least for extratropical storms.
This study was designed specifically to determine the errors characteristic of alternate methods which have been used to specify marine surface windfields in hindcast studies of extreme environmental factors associated with extratropical storms. The eastern North Pacific Ocean was taken as the test basin, but the results should be applicable to any extratropical region with a comparable meteorological data density.