A NEW METHODOLOGY FOR FLOOD FREQUENCY ANALYSIS WITH OBJECTIVE DETECTION AND MODIFICATION OF OUTLIERS/INLIERS

ABSTRACT: Prerequisites for deriving satisfactory estimates of design floods are the objective detection and modification of outliers/inliers at desired levels of significance and a versatile technique of flood frequency analysis. Objective detection and modification of outliers/inliers has been accomplished by developing statistics for outliers/inliers at both the high and low end of the flood spectrum. An inlier at the high end is defined as a variate (gener ated or observed) which is lower than indicated by the trend of the rest of the data, and an inlier at the low end is higher than indicated by the rest of the data. The tests developed in this study can be used to check for outliers/inliers at different levels of significance, such as 0.01, 0.05, 0.1, 0.2, 0.3, and 0.4. Three transformations for converting an observed flood series to an approximately normally distributed series were tested on flood series at 28 gaging stations in Illinois. The transformations considered were the power, Wilson-Hilferty, and 3-parameter log transformation. Analyses of th