Neuro-Fuzzy Logic Model for Breakup Forecasting at Fort McMurray, AB
At many sites in Canada river breakup presents a source of flooding concern, as the transition from an ice covered river to open water can sometimes be dramatic. Water levels can rise sufficiently quickly to fracture and move the ice cover while the ice is still reasonably strong and an ice jam can...
| Main Authors: | , , |
|---|---|
| Other Authors: | |
| Format: | Text |
| Language: | English |
| Subjects: | |
| Online Access: | http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.603.4240 http://cripe.civil.ualberta.ca/Downloads/13th_Workshop/Mahabir-Hicks-Fayek-2005.pdf |
| Summary: | At many sites in Canada river breakup presents a source of flooding concern, as the transition from an ice covered river to open water can sometimes be dramatic. Water levels can rise sufficiently quickly to fracture and move the ice cover while the ice is still reasonably strong and an ice jam can form if this moving ice comes to a halt. The safety of residents and property along a river may be compromised by the rising water levels behind a jam or by the water associated with the release of a large ice jam. Fort McMurray AB, is one such site and consequently an extensive database of relevant hydrological, hydraulic and meteorological variables has been created to support ice related research in the Athabasca River basin. Over one hundred variables have been investigated with multiple linear regression methods. Results indicate that several combinations of variables can be used to develop a model for maximum water level to be expected at breakup, but data from three seasons (fall, winter and spring) are needed to produce reasonably reliable forecasts. Therefore, only short term forecasts (only a few days warning) are possible with this model. This paper reports on a more sophisticated non-linear analysis undertaken using fuzzy logic theory, a form of artificial intelligence modeling that can incorporate both expert knowledge and historical occurrences. Here, using a logic rule base derived using artificial neural networks, a breakup forecasting model is developed which provides comparable results using fall and winter data only, thus providing a long lead time forecast (several weeks warning) of potential maximum water levels at breakup. |
|---|