such as "C" and "D" where no apparent relation exists. And again there 

 are sections such as "E" where there is a wave-like motion but double 

 the frequency of the waves . 



3. Moderate winds (November 12, 1968) 



The results of this run confirm the impression of largest tem- 

 perature fluctuations occurring near the wave surface. 



For this run the wind was 6.6 m/sec; the water temperature at 

 a depth of 6 cm. was 15.6 C and at 15 cm. was 15.7 C. The mean air 

 temperature indicated by the float mounted thermistor was 12 C. It 

 rode 3.5 cm. above the water surface. 



The fixed thermistor was repositioned for the last half of 



this run. Initially it was 7 cm. above mean water level and indicated 



o 

 a mean temperature of 10.7 C. It was relocated 15.5 cm. above mean 



water level and registered a mean temperature of 10.3 C. These tem- 

 peratures were checked in each case with a mercury thermometer. 



Figure 4 shows a section for the latter part of the run with 

 the fixed thermistor at 15.5 cm. Again the larger fluctuations are 

 measured with the float -mounted rather than with the fixed thermistor 

 but the general shape and phase agreement is not as strong as in run 

 2 though still present. The agreement was better, as might be expected, 

 in the first half of this run. The comments made about wave-temperature 

 signal agreement still stand. 



These data suggest that there was a strong temperature gradient in 

 the first few centimeters above the lake surface and possibly just be- 

 low the lake surface. Reconstructing the data from the moderate wind 

 run suggests the model shown in Figure 5. The closer the thermistor 

 is to the surface, the larger the temperature gradient it is in and 

 the larger the temperature fluctuations it will measure. The float- 

 mounted thermistor measured the largest temperature fluctuations be- 

 cause it was being moved up and down in the temperature gradient by 

 the waves. This being the case, one should expect a strong correla- 

 tion between the temperature fluctuation and wave records which, as 

 observations show, was not always present. The clue for the rather 

 loose to zero correlation is given by the calm run. There are appre- 

 ciable inhomogeneities in the air temperature which areadvected past 

 the sensors by the wind. Thus the temperature signal registered by 

 the float-mounted thermistor is made up of the combination of advected 

 temperature variations and temperature variations due to movement 

 through the mean vertical temperature gradient. 



From analyses made so far, wave and temperature fluctuation spectra 

 provide inconclusive evidence regarding the validity of this model. 

 There is a small peak in the October 30 (light winds) fixed-thermistor 

 temperature spectrum exactly at the peak of the wave spectrum (Fig. 6). 



380 



