shown in Figure 1. Table 1 gives some details of the gage locations, 

 gage types, and water depth at the gages. The gage designs and charac- 

 teristics are described in detail by Thompson (1977) , and for the Great 

 Lakes accelerometer buoy installations in Thompson (1978). 



The continuous-wire staff gage (manufactured by the Baylor Company, 

 Hoiiston, Texas) is the more accurate sensor because the gage directly 

 senses surface waves and it interferes very little with waves being 

 measured. The continuous -wire gage and the pressure gage (fabricated 

 at CERC) give conparable surface wave height estimates when the pressure- 

 gage record is compensated for the depth-dependent attenuation of the 

 dynamic component of pressure due to surface waves. Suitably compen- 

 sated pressure-gage spectra are also comparable to continuous -wire gage 

 spectra except at high frequencies (Esteva and Harris, 1970). 



The accelerometer-buoy gage or Waverider (manufactured by Datawell, 

 Haarlem, the Netherlands) gives wave heights and spectra which are 

 reasonably comparable to those given by the continuous-wire gage for 

 frequencies between 0.065 and 0.5 hertz, including virtually all impor- 

 tant frequencies in the Great Lakes (Pitt, Driver, and Ewing, 1978) . 

 The step-resistance staff gage (fabricated at CERC) has been shown to 

 have a consistent bias toward high wave heights (Esteva and Harris, 

 1970) . Significant height estimates from the step-resistance gage may 

 be 30 centimeters too high during low wave conditions and 20 percent too 

 high during high wave conditions. Aside from the tendency for over- 

 estimating wave height and energy, Esteva and Harris show evidence that 

 spectra for step-resistance gage records are reasonably consistent with 

 continuous-wire gage spectra. 



Time series of sea-surface elevation from staff gages, subsurface 

 pressure from pressure gages, and double-integrated sea-surface vertical 

 acceleration from buoy gages were transmitted by telephone line to the 

 CERC laboratory and used to generate pen-and-ink, strip-chart records 

 and digital records on computer-compatible, seven-track magnetic tape. 

 Digital records are 20 minutes long. A digital record from a single 

 station contains data points at 0.25-second intervals giving a total of 

 4,800 data points per location per 20-minute record. Further details of 

 CERC field wave data collection procedures are given in Thompson (1974, 

 1977) . 



2. Analysis . 



a. Spectral Analysis . The digital field wave gage records were 

 analyzed by the CERC routine analysis system (Thompson, 1977). One 

 record per 6 hours is routinely analyzed; however, one record per 2 

 hours was analyzed for some of the Great Lakes data. The routine anal- 

 ysis times are chosen to approximately coincide with synoptic times 

 (0100, 0700, 1300, and 1900 e.s.t.). The procedure includes editing 

 each 20-minute time series to eliminate data points which are obviously 

 bad, coroputing and testing the distribution function of the edited data 

 points, and con^juting the variance or energy spectrum for the time series 



19 



