Effect of Period of Record on Flood Prediction

Abstract

One long-term gaging station record was analyzed by standard methods, breaking the record into 205 shorter periods. Results are compared. Floods predicted on basis of different 10-yr. periods vary by as much as 59%. The primary advantage of the log-Pearson type III method of frequency curve determination is its mathematical objectivity. Once the method of analysis is selected a given set of data gives only one answer. The objectivity, however, is also a disadvantage in that anomalous points can greatly distort the result without the user being aware of the situation. Furthermore, the precision of the answer from a given set of data gives an unwarranted feeling of confidence in the accuracy of the answer. Use of the graphical method of frequency curve determination is justified when there is a long enough record so that only a moderate extrapolation is needed to reach the desired probability of occurrence. The graphical method, in this case, has advantages because of its simplicity, because of the necessity to plot the data and because it is possible to use individual judgment. The graphical method is especially adaptable for design of minor structures. In this paper, the mathematical method did not appear to give more accurate results than the conventional graphical method.