Summary of findings
Ice jam floods occur rapidly, often without warning, and are causing unprecedented damage to communities in the Northwest Territories. This project was intended to provide an improved understanding of flood causation and aid in possible flood forecasting at NWT communities by:
- Developing a set of early-warning singular indicators of climatic and water level conditions that might indicate imminent flooding;
- Identifying data gaps and recommending potential locations for future monitoring and data collection;
- Developing code to potentially help aid GNWT hydrologists and other researchers in hydrometric and meteorological data analysis using water level and discharge data from the Water Survey of Canada along with climate data from Environment and Climate Change Canada.
Three communities in the NWT, dominated by the effects of three different rivers, were chosen for this study: Aklavik (Peel River), Fort Simpson (Liard and Mackenzie Rivers), and Hay River (Hay River). All communities are located within the Mackenzie River basin. A set of possible climatic indicators were chosen, and their respective relationship to river water levels, discharge, and ice jam floods were explored. Timing of spring melt, spring temperatures, water levels in the fall preceding spring floods, snow water equivalent (SWE), and rain-on-snow events were investigated, primarily through the use of linear regression models and comparing the respective means between flood years and non-flood years. SWE was found to have a statistically significant relationship (p-value < 0.05) with water levels in years when observed flooding was recorded in Fort Simpson (over a 12 year period). SWE was also found to almost have a statistically significant relationship (p-value of 0.06) with discharge in Hay River in years where flooding was observed. No singular climatic variable was found to be a sufficient indicator for predicting water levels or streamflow in Aklavik, Fort Simpson, and Hay River.
A few of the variables appeared to have meaningful differences between their means for flood years and non-flood years. Average spring temperatures were higher during flood years for Hay River but lower for Aklavik and Fort Simpson. Antecedent fall streamflows for Hay River and Aklavik were both greater during years when subsequent spring flooding occurred, while they appeared to be lower in the falls preceding flood years in Fort Simpson.. Notably, none of the variables shared a positive or negative relationship between the mean for flood years and the mean for non-flood years for all three communities. For each variable at least one community had a positive relationship and at least one other city had a negative relationship, however slight. Although the differences in means have not been proven to be statistically significant, these differences are notable.
Though it’s highly useful to understand the role of single climatic variables in ice jam floods, the formation of ice jams is stochastic (Healy and Hicks, 2006). This makes it difficult to ascertain the exact manner in which different variables interact with each other to either produce a flood event or not. The alignment of measured hydrometeorological variables that produced a flood event in one year may not result in a flood event in another. Prowse and Beltaos (2002) delved into the long-term effects of climate on river ice processes, and determined that certain hydro-climatic factors can overshadow others, thus making it important to look at the relationships among climatic factors and their combined influence rather than one variables solitary influence. Also of importance is the role of meteorological fluxes in tandem with those climatic relationships. Given that the climate in northern Canada is expected to change more rapidly than other regions of the world, it may prove difficult to develop and maintain statistically based models for the region (Bush and Lemmen, 2019). Regime changes in climate, and how hydro-climatic conditions interact with each other may change to a large enough degree that caution will need to be applied when making predictions based off of historical data.
In conclusion, this project indicates that ice jam floods cannot be predicted by individual climatic indicators alone. Due to warming and the non-stationarity of climate trends, especially in the north, severity and frequency of future flooding is uncertain.