Application of a partial duration series model for regional rainfall frequency analysis in Michigan
In recent years, regional frequency analyses based on annual maximum series (AMS) methods have found wide application, while partial duration series (PDS) methods have not. Typically, the generalized extreme value (GEV) distribution has fit annual maxima, corresponding to generalized Pareto (GPA) distributed threshold exceedances. Previous work has shown PDS/GPA index-flood procedures to be more efficient than index-flood models based on AMS/GEV (Madsen et al., 1997). This paper discusses the application of a PDS/GPA regional index-flood procedure for updating rainfall intensity-duration-frequency (IDF) estimates for the State of Michigan. The PDS model applied herein assumes independent peaks and a Poisson-distributed number of threshold exceedances. Threshold determination, delineation of homogeneous regions, goodness-of-fit testing, and estimation of distribution parameters using L-moments are discussed. Finally, results from the PDS/GPA procedure are compared to those from a AMS/GEV index-flood method. Copyright ASCE 2004.
Bridging the Gap: Meeting the World's Water and Environmental Resources Challenges - Proceedings of the World Water and Environmental Resources Congress 2001
Application of a partial duration series model for regional rainfall frequency analysis in Michigan.
Bridging the Gap: Meeting the World's Water and Environmental Resources Challenges - Proceedings of the World Water and Environmental Resources Congress 2001,
Retrieved from: https://digitalcommons.mtu.edu/michigantech-p/8664