Executive Summary
In the high northern latitudes, close to 60% of rivers will experience some degree of freeze-up every winter, putting northern communities located at the banks of rivers at risk and susceptible to the threats posed by ice jams and subsequent ice jam floods (Rokaya et al., 2018). The dominant river system in the NWT is the Mackenzie River and its tributaries (such as the Hay, Liard, and Peel Rivers). The Mackenzie River freezes annually, and, due to its low slope and numerous meanders, it is prone to ice jams, predominantly during spring break up (Beltaos and Prowse, 2001). Breakup ice jams are dynamic and very site specific, and are therefore difficult to predict. They can be especially unstable, and fail suddenly, causing drastic spikes in water levels (White et al., 2006). Subsequent flooding can lead to devastating destruction to northern riverine communities as ice jam floods occur quickly and without warning. These types of floods are responsible for about two-thirds of all flood damage in Canada (Beltaos and Prowse, 2001). To better aid and warn ice jam prone communities, improved forecasting and prediction tools are needed with regards to the ice jam formation and whether (and to what extent) they will result in localized flooding.
Though ice jam processes are multi-faceted, this paper only addresses the individual relationships between a few key climatic parameters with water levels and streamflows. It also compares water levels and streamflows during years of ice-jam flood occurrence (flood years) with years when ice-jam flooding does not occur (non-flood years). Climate, water level, and streamflow data for three “at risk” communities in the NWT, Fort Simpson, Hay River, and Aklavik, were downloaded and analyzed using the R statistical coding platform. Temperature, snow on ground, total precipitation, and total rain were compared against water levels and streamflows for the three communities during flood years and non-flood years to determine patterns that may aid with future forecasting efforts. The main objective of this research was to develop a predictive model that can provide an early warning system for when climatic and hydrometric indicators suggest that conditions for ice jam floods are favorable.
Though it is highly useful to understand the role of climatic variables in ice jam floods, the formation of ice jams is stochastic (Healy and Hicks, 2006). Formation and breakup and are also dependent on streamflow, channel geomorphology, ice strength and thickness characteristics, antecedent soil moisture, and location (Beltaos and Prowse, 2001) Spatial, geomorphological, and climatic discontinuities suggest that predictive tools may not be readily transferrable between sites. While some variables had strong correlations with water levels and streamflows during flood years, it seems ice jam floods cannot likely be predicted based on singular variable climate data alone. Forecasting and prediction tools need to take a systems approach due to the complex nature of ice jams (Hicks and Beltaos, 2008). Recommendations include exploring alternative climatic variables in search of stronger correlations between climatic variables and water levels or streamflows. Analysis of climate stations further upstream from stations reviewed for this report should be considered to take watershed influence into account. Lastly, all future models should take climate change into consideration as the inevitable warming of the north will have a dramatic impact on snow and ice formation.