The influence of thermal stability at the air-sea interface on computed values of the transfer velocities of trace gases is examined. The novel ''whitecap'' model for air-sea gas exchange of Monahan and Spillane (1984), extended here to include thermal stability effects, is linked with an atmospheric general circulation model to compute global transfer velocity patterns of a climate reactive gas, CO2. The important terms in the model equations such as the whitecap coverge, friction velocity, neutral and local drag coefficients and the stability parameter &psgr; m(Z/L) are discussed and analyzed. The atmospheric surface level air temperature, relative humidity, wind speed and sea surface temperature, obtained from the National Center for Atmospheric Research Community Climate Model 1 (CCM1) are used to drive algorithms describing the air-sea transfer velocity of trace gases. The transfer velocity for CO2 (kCO2) is then computed for each 2.8¿¿2.8¿ latitudinal-longitudinal area every 24 hours for 5 years of the seasonal-hydro runs of the CCM1.

The new model results are compared to previously proposed formulations using the identical CCM1 forcing terms. Air-sea thermal stability effects on the transfer velocity for CO2 are most important at mid-high wind speeds. Where cold air from continental interiors is transported over relatively warm oceanic waters, the transfer velocities are enhanced over neutral stability values. The depression of computed kCO2 values when warm air resides over cold water is especially important, due to asymmetry in the stability dependence of the drag coefficient. The stability influence is 20% to 50% of kCO2 for modest air-sea temperature differences and up to 100% for extreme cases of stability or instability. The stability dependent ''whitecap'' model, using the transfer velocity coefficients for whitecap and nonwhitecap areas suggested by Monahan and Spillane (1984), produces CO2 transfer velocities that range from 13 to 50 cm h-1 for a monthly mean. High-latitude regions of both hemispheres experience winter season means of 40 to 50 cm h-1.

The global area-weighted mean CO2 transfer velocity is 19.2 cm h-1, in reasonable agreement with the 14C estimate of Broecker and Peng (1974). Although consistent with global 14C estimates, the initial version of the model predicts a factor of 2 to 3 higher CO2 transfer velocities over areas with low wind speeds relative to the parameterizations of Liss and Merlivat (1986) and Tans et al. (1990). New transfer velocity coefficients for whitecap and nonwhitecap areas are suggested that bring the low wind speed results into better agreement with observations and other models. The calculations described here suggests that oceanic gas exchange with the atmosphere is sensitive to thermal stability at the air-sea interface. This specific, turbulence-related geophysical forcing may account for a portion of the observed scatter in previously obtained experimental data that has been correlated with wind speed alone. ¿ American Geophysical Union 1993