Models of LP3 Regional Skew, Data Selection, and Bayesian GLS Regression

As discussed in Gruber et al. Bulletin 17B employed by US Federal Agencies uses a skew map developed in 1974 by C. Hardison. With current computing power and advanced spatial statistical methods, a much improved skew analysis using currently available and much longer data sets is possible. This study employed the Bayesian Generalized Least Squares (B-GLS) framework and diagnostic statistics introduced by Reis et al. with data from seven states in the southeast United States. This paper describes a the methods for identifying and screening for redundant sites corresponding to nested watersheds with similar drainage areas. Special attention is devoted to developing an improved cross correlation model of annual peak flows. The B-GLS analysis demonstrates that regional skewness models can be highly accurate with a small model error variance and an effective record length in excess of 45 years.