Effect of Record Length on Prediction of Extreme Events Using River Flow Data

This paper considers the effect of record length on the sensitivity of extreme predictions
based on the analysis of annually recorded maximum flow records for a number of locations
worldwide. Locations have been chosen based on the criterion of a minimum of 100 years of
standardised records, with extremes that closely follow standard statistical techniques.
Based on the analysis carried out, confidence of extreme predictions appear to be a function of the log
of the return period event required, the reciprocal of the square root of the record length and a
parameter unique for each river. Using the techniques outlined in this paper, methods are proposed to
give confidence on extreme predictions using limited data sets. However, more work is required to
define the unique parameter for each river, which has not been considered in this paper.

Citation:
Hames, D. (2006) ‘Effect of Record Length on Prediction of Extreme Events Using River Flow Data’ Proceedings of the AC&T, pp.107-112..