|
Activity Number:
|
117
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #304134 |
|
Title:
|
Censored data methods for modeling annual peak flood flows
|
|
Author(s):
|
John Grego*+ and Philip Yates
|
|
Companies:
|
The University of South Carolina and Cal Poly
|
|
Address:
|
Statistics Department, Columbia, SC, 29208,
|
|
Keywords:
|
EM algorithm ; Profile likelihood ; Hydrology
|
|
Abstract:
|
Finite mixture models can be used to study the behavior of annual peak flow statistics compiled from a stream or river gage. In addition to the continuous data collected from a gage, historical flood stages may also be available. It is reasonable to assume that all intervening stages between historical events and the continuous record are right-censored. We will explore the use of the EM algorithm to model the joint distribution of the historical event, the censored flows, and the continuous record assuming a finite mixture model underlies the behavior of the annual peak flows. Standard error estimation for events of interest, typically 1 percent flood events, will be a focus of our presentation.
|
