[1] Despite decades of effort to accurately quantify whitecap fraction W using in situ photography of the ocean surface, there remains significant scatter in estimates for any given 10 m wind speed (U₁₀). It is believed that the resulting, commonly used, W(U₁₀) parameterizations do not fully account for the true variability in W, by failing to incorporate the impact of the wavefield and other environmental conditions. This paper attests to the variability in whitecap fraction attributed to these additional factors, by analyzing satellite-derived W estimates over the globe for a full year. A comparison is made between the wind speed dependence of satellite estimates and three W(U₁₀) relationships formulated from in situ photographic data. The influence of various secondary factors on W is investigated once the dominant wind speed dependence is accounted for. The W retrieval's sensitivity to secondary forcings is dependent upon microwave frequency; at 37 GHz it varies by up to 25% of the mean at a given wind speed, while at 10 GHz it is a maximum of 8%. This results from a frequency-dependent sensitivity to foam depth; at 10 GHz predominantly foam from active breaking waves is detected, while at 37 GHz thin foam in residual whitecaps is also seen. Principal component analysis is used to rank variables by their success in accounting for variability in W. After wind speed, the most important secondary factor that accounts for variability in W is the wavefield. A wind-wave Reynolds number accounts for almost as much variability in W as wind speed.

In this study, the utility of satellite-based whitecap fraction (W) data for the prediction of sea spray aerosol (SSA) emission rates is explored. More specifically, the study aims at evaluating how an account for natural variability of whitecaps in the W parameterization would affect SSA mass flux predictions when using a sea spray source function (SSSF) based on the discrete whitecap method. The starting point is a data set containing W data for 2006 together with matching wind speed U₁₀ and sea surface temperature (SST) T. Whitecap fraction W was estimated from observations of the ocean surface brightness temperature T_B by satellite-borne radiometers at two frequencies (10 and 37 GHz). A global-scale assessment of the data set yielded approximately quadratic correlation between W and U₁₀. A new global W(U₁₀) parameterization was developed and used to evaluate an intrinsic correlation between W and U₁₀ that could have been introduced while estimating W from T_B. A regional-scale analysis over different seasons indicated significant differences of the coefficients of regional W(U₁₀) relationships. The effect of SST on W is explicitly accounted for in a new W(U₁₀, T) parameterization. The analysis of W values obtained with the new W(U₁₀) and W(U₁₀, T) parameterizations indicates that the influence of secondary factors on W is for the largest part embedded in the exponent of the wind speed dependence. In addition, the W(U₁₀, T) parameterization is able to partially model the spread (or variability) of the satellite-based W data. The satellite-based parameterization W(U₁₀, T) was applied in an SSSF to estimate the global SSA emission rate. The thus obtained SSA production rate for 2006 of 4.4 × 10¹² kg year⁻¹ is within previously reported estimates, however with distinctly different spatial distribution.