check for faulty electrical connections and filtering. Wind interfered 

 with the calibration process by generating ripples and occasionally basin 

 oscillations. The only solution was to wait for calm. Nonlinearity was 

 usually reducible to ±0.5 percent of full scale by manipulation of the 

 parallel wire probes as described by Stafford (1972) . Occasional occur- 

 rences of very heavy rains caused a short-term drift and gain problem 

 which hindered the calibration of the gages. The rain would leave a top 

 layer of water which had a different conductivity due to different chemi- 

 cal content and temperature. The different conductivities at different 

 water depths caused unequal flow of current per unit length of probe and 

 as the probe position changed relative to water depth during the calibra- 

 tion process, electronic balance and amplification changed. This was 

 another form of nonlinearity, which would be resolved only by natural 

 mixing of the water. As the heavy rains also caused generator operating 

 problems, testing was delayed until the following day [see Sec. IV, 5). 



c. Calibration Check . Another significant step in determining wave 

 record accuracy was the calibration check. Within minutes after the com- 

 pletion of each individual run, a sufficient number of calibration marks 

 were placed on the record to determine the amount of drift and change of 

 gain that had occurred since the initial calibration. Drift is not an 

 impairment to data reduction, as long as the trace remains within the chart- 

 paper grid. In determining reflection coefficients, the change of gain 

 can be disregarded, since 



AgH^ - AgH^ Ag (H^ - H^) H^ - H^ 



R AgH^ + AgH^ Ag (H^ + H^3 H^ + H^ 



where 



K^ - reflection coefficient 



H^ = height of wave envelope antinode 



H^ = height of wave envelope mode 



Ag = change of gain 



Stafford (1972) verified that the change of gain occurring during a run 

 is linear to the degree that the initial calibration is linear. Therefore, 

 for reflection coefficients the linearity of the initial calibration is 

 the prime consideration. 



The change of gain is important in analyzing absolute wave heights. 

 Immediately after each run, the calibration was checked and the amount of 

 gain change since the initial calibration for each channel was recorded on 

 the wave record and in the log. The log data showed gains of from 0.5 to 

 4 percent above the initial calibration, with an estimated 80 percent of 

 the measurements within this range; e.g., in 1970, only four cases of 10- 

 to 11-percent gain were observed. Losses on the calibration checks were 



30 



