Intrinsic frequency spectra of water waves in the range of 6-17 Hz were obtained as a function of both wind speed and wind stress from point measurements of wave height. In a lake with a limited fetch there are two types of surface motions causing Doppler shift in the frequencies of short waves: orbital velocity of long waves and surface wind drift. The former was estimated from long-wave amplitude by using a linear wave theory. The error in this estimate is of the order of the long-wave slope, and for this work it is typically 10%. The latter was approximated by the friction velocity. The friction velocity could be either taken as 3% of the mean wind speed measured at a height of 10 m or obtained from our direct measurements of the wind stress. The surface drift velocities obtained by these two approaches were found to be in close agreement. However, the estimate based on the mean wind speed was preferred because of its simplicity. Doppler frequency shift can be corrected in either the frequency or the equivalent spatial domain. A comparison of these techniques from both a theoretical and a practical point of view was made. The two techniques were found to produce comparable results. Experimental results showed that the spectral energy of short waves rapidly increased in response to increasing winds and jumped up by an order of magnitude when wave breaking occurred.
INTRODUCTION
The transfer of wind energy and momentum to the sea involves water surface waves of different length scales. The short gravity-capillary waves provide the roughness elements with which the wind interacts [Stewart, 1961]. The longer waves cause the wave-induced pressure fluctuations which are central to the energy flux from wind to waves (see, for example, Phillips [1977, p. 132]). The relative importance of these two stress transfer mechanisms is thought to be a function of the mean wind stress itself, but the details are poorly known [e.g., Valenzuela, 1976; Plant and Wri#ht, 1977; Snyder et al., 1981; Mitsuyasu, 1985]. The relation of the amplitude of these smallscale waves to wind speed or stress and their dependence on the presence of longer waves or mean currents have also become increasingly important from a technological point of view. This has occurred because the remote sensing techniques used to infer these other phenomena depend on the resonant interaction of the electromagnetic radiation with the smallscale surface waves (Bragg scatter). The electromagnetic theory for Bragg scattering is evidently well understood [Rice, 1951; Wright, 1968; Brown, 1978; Ulaby et al., 1981; Keller et al., 1985]. However, whether the scatterometer signal is proportional to wind speed or wind stress is a fundamental question of importance to many investigations which involve scatterometer data. (For discussions, see, for instance, O'Brien et al. [1982], Brown [1983, 1986], Pierson [1983], and Donelan and Pierson [1984].) Other features of the sea surface related to wave breaking may also contribute to the radar backscatter [Keller et al., 1986; B...