Regression and fuzzy logic based ice jam flood forecasting

Abstract

In Canada, ice jam events have frequently produced the most extreme and dangerous flood events on record, resulting in millions of dollars in associated damages. However, our ability to forecast such events remains quite limited. An example of this is the Athabasca River at Fort McMurray, Alberta, where severe ice jam events have been documented for over 100 years, and where breakup has been monitored intensively for the past 25 years. Despite these efforts, no reliable flood forecast model is yet available. Here, the use of Fuzzy Expert Systems is explored to examine their potential for developing long lead time ice jam risk forecasts for this site. The developed System identified seven out of twenty two years that had the potential for high water levels, including all four years where high water levels actually occurred. These preliminary results suggest that Fuzzy Expert Systems are promising tools for long range ice jam flood forecasting. © 2008 Springer-Verlag Berlin Heidelberg.