The Effect of Outliers on Flood Frequency Estimates

摘要

Flood frequency analysis plays an important role in the design of hydraulic structures and in the delineation of floodplains. There are several factors in the flood frequency analysis procedure that depend on the user's judgment or input. One such consideration, the detection and treatment of outliers, is investigated in this study. The Bulletin 17B (IACWD, 1982) defines outliers as data points that depart significantly from the trend of the remaining data. Hypothesis-testing-based approaches are widely used to detect outliers. These outliers, however, are not automatically eliminated from flow records. Verifiable ground-based information is needed for both the retention and elimination of outliers. In the absence of such information, they are subject to modification as a compromise solution. As the results of this study show, each of these choices (outlier retention, modification and elimination) has impacts on the ensuing flood frequency analysis. The study-results also show that statistical outliers are often present in flow records and that the number of outliers identified can vary depending on the test and significance level chosen for outlier detection.