| Summary: | The tidal record from the Reykjavik Harbour in Iceland is analysed for extremes. For this analysis the annual maxima of recorded high tide of the 37 recorded years are used. In addition predicted high tides from the calendar years 1-2950 are used. This number of predicted years is considered sufficient data for a fairly accurate probability distribution of the astronomical tide, but the prediction is based on astronomical constituents. The astronomical tide deviates from the recorded, and three possible deviations are discussed, two deviation series that are differences of simultaneous values and one that is differences of annual maxima. It is found that a cross-correlation has to exist between the recorded and predicted series, if a stochastic model, expressing the recorded value as the sum of the prediction and the deviation, is to be applied. One of the deviation series fulfils this requirement, its recorded and predicted series are correlated. The deviation can be separated in two parts, one correlated to the prediction and another independent of the prediction. Using this, a probability integral is derived that gives the expected value of the tidal surge level as a function of return period. This result is compared to historical flood record, and it is found that the results are reliable for return periods 30-100 years. Finally, it is recommended that a hydrodynamic computational model of the tidal motion is constructed in order to investigate the tidal surge from rare meteorological situations that may not be represented in the 37 years of record.
|
|---|