From studies of celestial motion we know that each tidal cycle follows a simple recurring (harmonic) pattern with a fixed period. The parameters that vary from place to place are the amplitude and time lag (phase) of each constituent.
Harmonic tidal analysis involves taking a series of sea level measurements made at regular intervals, usually hourly, and resolving the amplitude and phase values for each constituent tide that comprises the observed tide at that place.
The period covered by the sea level measurements has a bearing on the information that the analysis can provide. Since the moon has the greatest effect on the tides, data from at least one month is required to account for the major lunar orbit-related effects. Longer sequences of observations enable the analysis to distinguish between constituent tides with similar periods as well as to determine very long period constituents and generally improve the quality of the results.
The analysis of sea level data for a single month will produce parameters for about 36 constituent tides - the analysis of a years' data will typically yield amplitudes and phases for 68 constituent tides. However, the amplitudes of many of these constituents are usually small (a few millimetres, or less) and as these values are affected by the amount of scatter (noise) in the sea level measurements the reliability of such results is questionable. Meaningful values can usually be obtained for the most significant 20 to 30 constituent tides for the locality.
Wherever possible predictions are based on the analysis of sea level measurements taken over a period of at least one year. This allows the average changes in mean sea level due to seasonal variations in meteorological conditions to be calculated and included in the predictions. These changes do not, however, repeat themselves exactly from year to year so it is advisable to observe and analyse sea level measurements over a number of years.
Once the amplitude and phase values for a set of constituent tides have been calculated they can then be used to create the tide curve for a specified time interval. This information can then be used to find the times and heights of maxima (high water) and minima (low water). Predictions can be made for future and past dates, however their accuracy decreases as the time difference between the measurements used in the analysis and the prediction period increases.
As predictions are given for average meteorological conditions, it follows that when conditions are not average the actual tide times and/or heights may differ from those predicted. The effects of varying meteorological conditions are discussed in the following section.