Controlling Skewness in MOVE3 Peak-Flow Record Extensions

Abstract

Streamgage record extension methods such as the maintenance of variance Type 3 (MOVE3) method improve flood frequency estimates at a target streamgage by incorporating information from a nearby, hydrologically similar index streamgage. Bulletin 17C recommends using a variation of the MOVE3 method to estimate values at the target streamgage for only a subset of the available data at the index streamgage to account for uncertainty in values estimated using MOVE3. Bulletin 17C recommends the most recent index streamgage data be used for the subset unless these data misrepresent the skewness of the extended record. However, no method is provided to select the subset if the most recent data are inappropriate. The objective of this study is to develop such a method to select the subset of peaks by extending Bulletin 17C’s MOVE3 methodology to control the skewness of the extended streamgage record. The new method allows the extended record skewness to be informed by all available peak-flow data at the index streamgage while accurately computing the variance and resulting confidence intervals for flood frequency estimates. An example is presented comparing three different variations of MOVE3 record extension, which produce extended streamgage records with the same mean and variance, but different values of skewness. In the example, the difference in skewness between the three methods causes the results of flood frequency analysis for the 1% annual exceedance probability flood to differ by about 15%, illustrating the importance of considering skewness when using MOVE3 record extension.