construction wliich modifies the free travel of waves from tlie open sea. Zetler and 

 Cummings (1967) list the harmonic constants for Philadelphia, Pennsylvania, as determined 

 from observations in 1946, 1952, and 1957. The amplitude of major constituents vary by 

 about 4 to 8 percent of the minimum value, and phases vary by about 5° to 40°. Philadelphia 

 has one of the longest, most constricted channels from the open sea of any U.S. port. Tiie 

 variability of the harmonic constants at Plriladelphia is beUeved to be near an upper limit for 

 the United States. 



Derivations by the tide prediction equation, in greater detail, are given by Schureman 

 (1941), Doodson and Warburg (1941), Pillsbury (1956), Defant (1961), Dronkers (1964), 

 Ippen (1966), and Godin (1972). 



IV. TIDAL DATUMS 



Mean sea level (MSL) was widely adopted as a primary datum many years ago on the 

 assumption that it could be determined accurately and simply from the records of any 

 reasonably weU-exposed tide gage. The mean elevation of the sea surface is an equipotential 

 surface by the definition of equipotential. In principle, it might be considered that MSL 

 should be established by integrating the tide record over a long period. In practice, this is 

 not only difficult but may be undesirable because tide records frequently contain 

 irregularities of a local and nonperiodic nature due to meteorological effects and tsunamis 

 (see Figs. 3 and 4). In addition, MSL determinations are based on the average of the hourly 

 determinations of tide level. Disturbances of the type shown in Figures 3 and 4 have often 

 been smoothed before tabulating the tide record. 



The ideal length of record to be considered is generally taken as about 19 years, partly to 

 account for the cycles of 18 to 19 years in tidal ampUtude and phase, but mostly to average 

 out the more important meteorological effects. The existence of trends in the elevation of 

 the sea relative to the land for periods longer than 19 years is not explicitly recognized in 

 selecting a period of 19 years. The existence of long-period trends, however, is a major 

 factor in requiring revisions of the official datums at intervals of about 25 years. The MSL 

 must often be estimated when less than 19 years of data are available. An integral number of 

 years should be used, if possible. If less than 1 year of data must be used, the preferred 

 period is 29 days or a multiple of 29 days. Methods for minimizing the errors in short-period 

 determinations of sea level are provided by Marmer (1951) and Swanson (1974). 



Briefly, their technique compares the available record for MSL or mean tide level with 

 tlie same period of record at nearby stations with similar tide forms with a record duration 

 of 19 years or longer, to identify any long period anomaly and to assume that any anomaly 

 in the short record is the same as that at a station with long records. Marmer advocates 

 transferring the mean tide level and the mean tide range; Swanson recommends separate 

 comparisons for each datum of interest. 



If it is impractical to consider hourly values, a good approximation is provided by the 

 half-tide level, sometimes called mean tide level (MTL). 



36 



