On estimating the surface wind speed over drifting pack ice from surface weather charts

Abstract

A recent publication in this Journal (Feldman et al., 1979) on estimating the surface wind speed over drifting pack ice from surface weather charts has prompted us to offer the following comments :(1) Feldman et al. have used linear regression equations based on the studies of Hasse and Wagner (197 1) and Hasse (1974a, b) to estimate surface wind speed U(at 10 m level) in terms of the geostrophic wind speed G; these authors, however, do not mention the surface wind direction. Without appropriate determination of the deviation of the surface wind direction from the geostrophic wind (cross-isobar angle), a knowledge of surface wind speed alone is not sufficient in studying pack ice movement or the movement of isolated floes. The cross-isobar angle, which is a function of surface Rossby number and atmospheric stability, can vary from 15 to 25 ' for typical stability values observed over high latitude ocean areas (see Haltiner and Martin, 1957). The classical study of Campbell (1965) which Feldman et al. refer to, obtains mean wind stress field by constructing a field of surface Rossby number over the polar ocean and applying Blackadar's (1962) results ; thus Campbell appropriately takes into account surface wind speed as well as direction.(2) Feldman et al. mention using data of Smith (1971, 1973) and Smith et al. (1970) for calculating the regression of U on G and G on U as shown in their Figure 2; they have omitted the set of data at site 1 because it gave a negative correlation coefficient of -0.58 between U and G. This negative value of the correlation coefficient appeared doubtful to us and accordingly we carefully examined the data of Smith et al. (1970) collected over the Gulf of St. Lawrence. Using all 9 observations in Table I of Smith et al. (1970), we obtained a linear correlation coefficient (between U and G) of + 0.95 which appears reasonable because the geostrophic wind speed and the surface wind speed at 10 m level should be positively correlated. Feldman et al. appear to have used only 5 of the 9 observations to obtain a correlation coefficient of -0.58. It is our contention that this negative correlation coefficient may not be due to widely spaced isobars as speculated in the paper.(3) For estimating surface wind speed U over the Arctic Ocean, Feldman et al. propose the use of regression Equations (2), (3) and (4) of their paper; these regression