The tide hydrographs in Appendix B show a significant variation in tlie range of the tide 

 throughout the lunar month. Compilation of the distribution function for daily ranges, and 

 perhaps for semidiurnal ranges, would have provided a significant improvement to the 

 statistical description of the tide. This compilation should be based on lunar not solar days. 

 At locations where the tide is strongly semidiurnal or strongly diurnal, a good 

 approximation to the complete tidal curve can be proved by connecting highs and lows witii 

 a sine curve. This is not true for stations with a large diurnal inequality of the tides, and 

 where the lower high water and the higher low water are sometimes approximately equal. 

 This phenomenon is illustrated in the hydrograph for St. Petersburg, Florida (see Fig. 2). 

 The compilation of a mean shape for the diurnal tide hydrograph or a distribution function 

 for hourly tides, as normalized by the diurnal range, would have been a useful addition. The 

 compilation of this normalized diurnal tide, together with distribution function for diurnal 

 range, would permit greater precision in the calculation of the probabUity of combined tide 

 and surge. 



The graphs of the seasonal cycle in tide parameters in Appendix B show that most of the 

 variation is associated with seasonal cycle in montlily MSL. An independent examination of 

 the variabUity in monthly MSL (see Fig. 6) indicates that the monthly MSL varies 

 significantiy from year to year. These factors suggest that it would have been useful to have 

 omitted the terms S^ and S$a which describe the annual cycle of the sea level from this 

 calculation, and to have used the theorem for combined probabilities (see Sec. VIII) to 

 combine the distribution function for astronomical tides calculated without regard to the 

 annual cycle with an empirically derived distribution function for monthly mean sea levels. 

 This approach appears to be especially useful in predicting the probable extreme water levels 

 expected to result from hurricanes. Hurricanes are most likely to occur in those months in 

 which the MSL is expected to be near the peak of its annual value, even in the absence of 

 hurricanes. Along the Washington-Oregon coast, the storm surges are most hkely to occur in 

 winter near the peak of the annual cycle of MSL in that region. 



86 



