An Emerging Picture of Peace River Break-Up Types That Influence Ice

Home , Cadotte River, Fort Chipewyan

CGU HS Committee on River Ice Processes and the Environment th 20P P Workshop on the Hydraulics of Ice Covered Rivers Ottawa, Ontario, Canada, May 14-16, 2019.

An Emerging Picture of Peace River Break-up Types that Influence Ice Jam Flooding of the Peace-Athabasca Delta Part 2: Insights from the comparison of the 2014 and 2018 Break-ups

Martin Jasek BC Hydro, 6911 Southpoint Drive, Burnaby, BC, V3N 4X8

[email protected]

The 2014 and 2018 highly dynamic break-ups on the Peace River have been well documented but were very different from each other in that the former produced an ice jam flood in the Peace-Athabasca Delta (PAD) and the latter did not. This paper examines the differences and looks at driving environmental variables that produced very different ice jam flooding features and locations. The paper then extrapolates further to outline other possible break-up types to increase the understanding of the driving variables that affect the chances of ice jam flooding of the Peace River sector of the PAD.

1. Introduction Background information about the location, hydrology, the ice jam flooding mechanism, the high springtime flows of the unregulated Smoky River to drive the dynamic break-up on the Peace River and channel storage release that all play a role in ecologically beneficial flooding of the Peace Athabasca Delta are described in Part 1 in these proceedings (Jasek, 2019a).

This paper compares the 2014 and 2018 break-ups of the Peace River, the former resulted in an extensive flood of the PAD and the latter did not. Detailed descriptions of the 2014 and 2018 break-ups can be found in Jasek (2017a, 2017b) and Jasek (2019a, 2019b) respectively.

The term “jave” is used throughout this paper and is defined as: when an ice jam releases it forms a wave or what is called a “jave” that can break-up more ice cover in front of it and may eventually create ice jam flooding at the PAD. The term “jave” defined in Beltaos, (2008).

2. Comparison of the 2014 and 2018 break-ups

2.1 USpring snowpacks on the Alberta plains in the Peace River Basin

Spring snowpacks or Snow Water Equivalent (SWE) in the portion of the Peace River basin were well above normal in the spring of 2014 and 2018 (Figure 1a and 1b respectively). When these large snowpacks melt they drive run-off that can trigger dynamic break-ups on the Peace River. Figure 1c shows the ratio of SWE of 2018 over 2014. Eight out 13 of the snow survey sites indicated higher SWE in 2018 than in 2014 and five out of 13 showed a lower SWE. This suggests that the snowpacks were similar in the two years, slightly differently distributed and overall slightly higher in 2018 than in 2014.

2.2 USpring air temperatures

Figure 2a shows the hourly air temperatures for April – May 2014 and April – May 2018 on the Smoky River at Watino which are a good indicator of the snowmelt occurring in the low elevations of the Smoky River basin and other catchments along the Peace River between the foothills and the Town of Peace River that typically drive the dynamic break-up on the Peace River.

There are some differences and similarities in the temperature trends between the two years indicted in Figure 2a that are worth highlighting. In early April 2014 there was above freezing warm spell that lasted about a week were the same period in 2018 had mostly below freezing temperatures. Therefore, some early melt occurred in 2014 where some of the snowpack may had been lost too soon to contribute to the river flows during break-up in late April. This earlier warm spell in 2014 was enough to dynamically break-up the lower Smoky and Little Smoky Rivers from Apr 11-16, 2014 but not the upper Smoky River (Jasek 2017b). From April 12 to 22, the temperatures were fairly similar and mostly positive in the two years. The warm spells between April 18 to 23 in 2014 and 2018 was what drove the upper Smoky River to break-up on Apr 25, 2014 and what started the Lower Smoky to break-up on Apr 23, 2018.

In 2018, the warm weather continued after April 22 where in 2014 it cooled for a few days. The continuing warm weather in 2018 caused the upper Smoky River to break-up on Apr 26, 2018, only three days after the lower Smoky River broke up. By contrast, there was a 9 day delay between the lower Smoky completed break-up and the upper Smoky break-up. However, in both years, the Peace River dynamic break-up for long distances towards Fort Vermilion and the PAD did not start until the upper Smoky River had broken up.

2.3 UDischarge

Figure 2b shows the discharge at the Town of Peace River in April – May 2014 and April – May 2018. The traces start after there is no more ice affected backwater at the Town of Peace River. Even though the water volumes available from snowpack were similar in the two years, the run- off came in three pulses in 2014 and in one pulse in 2018 due to differences in air temperatures. The first pulse in 2014 in Figure 2a is not shown because the Peace River at the town of Peace River were backwater affected by an ice cover but it did break-up the lower Smoky River and 3 Little Smoky Rivers. However, both the 2014 and 2018 break-ups exceeded 4000 mP P/s at the Town of Peace River on about the same date (Apr 24-25) which started the dynamic break-up of the Peace River in both years.

In 2014, there were two discharge peaks at the Town of Peace River that drove the break-up 3 3 further downstream, about 4700 mP P/s on April 25, 2014 and another of about 4000 mP P/s on May 3-4, 2014. In 2018 there was one prolonged and much higher discharge peak between 6000 and 3 8000 mP P/s from Apr 26 to May 1, 2018. These two very different hydrographs drove the two break-up sequences in 2014 and 2018 described next.

2.4 UBreak-up sequences, ice jams and flooding in 2014 and 2018

Figure 3 shows break-up front locations and observed ice jams for the 2014 and 2018 break-ups. Horizontal lines indicate stalled break-up fronts with vertical lines terminating at the horizontal lines indicating locations and lengths of observed ice jams. Small sloping lines are indicative of slow moving break-up fronts, typically thermal break-ups. Steep lines indicate faster moving dynamic break-up fronts. Lines that are dotted are approximate break-up fronts to a few hours based on indirect data and rate of travel assumptions. Indirect data includes subsequent observations of ice runs the next day, timing of gauge increases and prior observations of intact ice cover locations a few hours to a day before. More detailed similar plots indicating intact ice covers, open water and ice runs on similar figures to Figure for the 2014 and 2018 break-ups can be found in Jasek (2017a) and Jasek (2019a) respectively.

The starting date of the dynamic break-up on the Peace River in both 2014 and 2018 was Apr 26 as noted by the change in slope of the break-up front celerities in Figure 3 on that day. The break-ups in both years transitioned from thermal to dynamic not only on the same date but also a similar location, just downstream of Sunny Valley, about 700 km upstream of the PAD.

Both break-ups created ice jams around Fort Vermilion and Vermilion rapids but the 2014 break- up had stalled ice jams in place for 3 days where the 2018 break-up had stop and release of ice jams from Thompkins Landing to Vermilion Rapids on a more or less daily interval.

Figure 3 shows the break-up of the ice cover between Vermilion Rapids and the PAD was very sequential in 2014 lasting only about 1 day. By contrast the break-up of this reach in 2018 experienced multiple break-up fronts over about a 2 day period.

The most important difference between the two break-up sequences relative to PAD flooding is that in 2018 the ice in the PAD reach and well downstream on the Slave River melted out thermally prior to the arrival of the main break-up jave from upstream, where there was competent ice there prior to the 2014 break-up. In 2018 an early thermal break-up started from Rocky Point sometime on Apr 25 and 26 when the more forceful dynamic break-up was still 700 km upstream. This thermal break-up continued downstream on the Slave River and was about 76 km downstream of the PAD on the Slave River on May 2 when the jave from the main dynamic break-up from upstream arrived in the PAD reach. This was too far downstream of the PAD for the ice volume available from upstream to create an ice jam through the PAD reach; the head of the ice jam on the Slave River was apparent in a satellite image on May 4 to be 36 km downstream of the PAD. By contrast the break-up in 2014 broke-up a very competent ice cover from Vermilion Rapids all the way through the PAD reach and 10 km downstream on the Slave were it remained for more than a week causing the PAD to flood (See ice jam in PAD reach in Figure 3 lasting until May 9, 2014).

2.5 UState of the ice cover prior to the 2014 and 2018 break-ups

Figure 4 shows a comparison of the ice conditions on the Slave River just downstream of the PAD in 2014 and 2018. The May 1, 2014 images show the dynamic break-up front entering from the Peace River into the Slave River and solid ice downstream on the Slave River. The May 3, 2014 images show the ice jam for the first 10 km of the Slave River and a solid ice cover further downstream. They also show an intact ice cover on the Rivière des Roches. By contrast the April 30, 2018 images show highly deteriorated and melted reaches of the ice cover on the Slave River and also a completely open Rivière des Roches. It became obvious during the Apr 30, 2018 observation that there was insufficient ice cover and a weak ice cover that likely would not cause an ice jam if a major jave arrived from upstream. The jave did arrive May 1-2, 2018 and in fact flushed the remaining ice in the upper Slave closer to the PAD further downstream making an ice jam flood of the PAD impossible in 2018.

Figure 5a shows the ice covers downstream of major ice jams in 2018 upstream of Fort Vermilion, at Fort Vermilion and upstream of Vermilion Rapids; the ice in these reaches of the Peace River looks white and competent. Figure 5b shows the ice covers in the Peace River in the PAD reach and a location between Vermilion Falls and the PAD. The ice in these reaches of the Peace River looks dark and deteriorated. All but one photograph in figures 5a and 5b were taken on the same day (Apr 29, 2018) allowing so they have been exposed to similar degradation energy from solar radiation and warm air temperatures. Therefore, the differences in the ice covers were likely due to ice development in the fall and over the winter. The contrasting ice covers between Figure 5a and 5b explains why there was ice jam flooding in the upper reaches of the Peace River and not in the lower reaches of the Peace River in the PAD reach in 2018.

There were some short reaches of more competent looking ice between Vermilion Falls and the PAD. These included short reaches downstream of Vermilion Rapids and Boyer Rapids where over winter frazil generation caused ice covers to grow into frazil slush given it a more white and solar radiation resistant quality. The latter included the Peace Point reach. However, the majority of the ice cover between Vermilion Rapids and the PAD was dark, deteriorated and with open leads just prior to the main break-up jave arriving in this reach.

These differences in ice presence/strength between 2014 and 2018 at the PAD and from upstream to downstream at the PAD in 2018 steered this investigation to examine the antecedent conditions that may had led to these differences in ice strength or break-up resistance.

3. Antecedent conditions for 2014 and 2018 break-ups To help determine what caused different break-up sequences in 2014 and 2018 and the resulting different reaches of ice jam related flooding in those two years (the former caused the PAD to flood and the latter did not), it is helpful to determine the antecedent break-up resistance prior to these break-ups.

U3.1 The 2013 and 2017 freeze-ups

Data on freeze-up is important to determine ice conditions at break-up because one needs to know the start date of the thermal growth of the winter ice cover, the date that snow can start accumulating on the winter ice cover that retards thermal growth due to insulation or if it is sufficient enough to submerge it and form snow-ice layer on top. The type of ice that forms is also important as clear ice deteriorates much faster in the spring from solar radiation than snow ice or ice that grows through frazil slush, both of which have air bubbles that scatter the sunlight and protect the deeper ice from deteriorating.

Satellite images and gauge data were reviewed to determine the dates of freeze-up in the PAD reach, Peace Point and at Fort Vermilion. These dates are summarized in Table 1.

Table 1. Timing of Freeze-up in the PAD reach, Peace Point and Fort Vermilion. PAD reach Peace Point Fort Vermilion (km 1198 - 1243) (km 1136) (km 832) 2013 Freeze -up Date Nov 18 Nov 19 - 22 Nov 18 2017 Freeze-up Date Nov 14 Nov 15 - 22 Nov 24

Note that there are date ranges for freeze-up at Peace Point. This was because the raw water level gauge data trace continued to exhibit large fluctuations at this location after satellite images confirmed ice presence there indicating that ice movement was still occurring there intermittently. Sentinel 1 satellite images on Nov 17 and 21, 2017 indicated that freeze-up was still active between those two dates in the Peace Point reach but an ice cover was present and remained unchanged in the PAD reach. The likely reason for this is that the Peace River is steeper upstream of Peace Point to above Boyer Rapids than downstream of Peace Point to the PAD reach. Because it is steeper further upstream, it takes some ice cover consolidations at Peace Point for the ice front to develop enough stage to drown out and reduce the high water velocity at Boyer Rapids to allow freeze-up to progress further upstream.

Due to the smaller slopes and velocities downstream of Peace Point, freeze-up in the PAD reach should occur in the juxtaposed ice cover mode, allowing for rapid upstream progression of the ice cover through the PAD reach all the way to Peace Point, in about one day. This one day progression from the PAD reach to Peace Point was confirmed by Nov 17, 18 and 19, 2013 MODIS satellite images. For 2017, it was assumed that freeze-up in the PAD reach occurred on Nov 14 the day before the first gauge stage-up at Peace Point (Nov 15) a daily satellite images were not available in that year.

All signs indicate that freeze-up in Fort Vermilion occurs independently of that at Peace Point and the PAD. Based on satellite images and Fort Vermilion gauge data in Nov 2013, the ice lodged just upstream of Vermilion Rapids, initiating another ice front that progressed through Fort Vermilion on Nov 18. Based on satellite images and Fort Vermilion gauge data in Nov 2017, the ice lodged at km 1080 (between Garner Creek and Boyer Rapids), initiating another ice front that progressed upstream, eventually submerging Vermilion Rapids sometime between Nov 22 and 24 and causing freeze-up at Fort Vermilion on Nov 24, 2017. Based on satellite images, there was at least one other lodgment location between Boyer Rapids and Vermilion Rapids in 2013 and at least two in 2017 as shown in Figure 6a.

The ice fronts for the two ice seasons continued to advance upstream as indicated in Figure 6a responding to changes in air temperatures (Figure 6b) and occasionally receding downstream in mid-winter due to warm weather. Overall, the 2017-2018 winter had more rapid changes in air temperatures and more sustained warm and cold spells (Figure 6b) leading to larger oscillations in the ice front position than in the 2013-2014 winter as shown in Figure 6a. For example a major consolidation occurred in early January 2018 (Figure 6a) that was triggered by a sudden o o rise in mean daily temperatures from -30 P PC to -5 P PC (Figure 6b) and a lengthy mid-December warm spell in 2017 caused a recession in the ice cover in a month that it should normally be advancing.

3.2 UCumulative freezing air temperatures and snowfall on the ice cover

A thick ice cover is more likely to stop a dynamic break-up and cause an ice jam flood so it is important to determine the maximum winter ice cover thickness prior to ice deterioration and melt in the spring prior to break-up. Cumulative freezing air temperatures and snowfall on the ice cover are two of the most important variables in determining the end of winter thermal ice thickness. Since the largest ice jam flooding in 2014 occurred in the PAD reach and in 2018 occurred in Fort Vermilion, air temperatures and snowfall data were examined representative of both reach in both years.

Cumulative freezing air temperatures were summed starting on the freeze-up dates presented on Section 2.1. Snow cover was also computed starting from zero on the date of freeze-up. Rather than using daily cumulative winter precipitation to compute snow on the ice cover, snow on the ground measurements were used at nearby weather stations. Using snow on the ground measurements rather than daily cumulative precipitation allowed for fewer assumptions needed to compute the snow on ice cover. One did not need to consider snow settlement, melt and redistribution by wind. Snow on ground prior to freeze-up dates was subtracted from the snow on ground values in order to not count the snowfall that fell into open water.

Cumulative freezing air temperatures and snow on the ice cover for the two years and two locations is presented in Figure 7. Figure 7a shows that the 2013-2014 winter was generally colder than the 2017-2018 winter and in both years it was generally colder in the PAD reach than in the Fort Vermilion Reach.

Figure 7b shows that there was more snow on the ice cover in the PAD reach earlier in the ice season in 2017-2018 than in 2013-2014 although the maximum snow depth at the end of the winter was almost the same. Figure 7c shows the opposite in the Fort Vermilion reach; there was more snow on the ice cover earlier in the winter in 2013-2014 than in 2017-2018.

The air temperature and snow data in this section will be used to compute the end of winter thermal ice thickness for the two years and two locations to ascertain if the differences can account for the different break-up sequences in the following section.

3.3 UThermal ice thickness

A daily time step spreadsheet model was developed that used the equation developed by Ashton (1980) for simulating thermal ice growth: dh   h h 1  ρ = −  + s +  − − e i Li (Tm Ta )/  hwi (Tw Tm ) [1] dt   ki k s hsa  3 Where: e = slush porosity, ρRiR = ice density = 916 (kg/m P P), LRiR = latent heat of fusion of water = 5 3.34 x 10P P J/kg, h = thermal ice thickness, t = time, TRmR = temperature of ice at ice-water interface o o = 0 P PC, TRaR = temperature of the air, TRw R = temperature of water at ice-water interface = 0 P PC, kRiR = o thermal conductivity of ice = 2.24 W/m/P PC, hRsR = snow depth, kRsR = thermal conductivity of snow o 2 o = 0.26 W/m/P PC, hRsaR = heat transfer coefficient between air and water = 17 W/mP P/P PC, hRwiR = heat transfer coefficient between water and ice. These calibration coefficients values have produced good results by the author on the Peace River during operational forecasting for BC Hydro compared to measured thermal ice thicknesses and are within acceptable ranges for the variables.

o For the purposes of this study, the water temperature was defined at 0 P PC thus neglecting the last term. This is justifiable since above freezing water temperature occur for only short distances downstream of the ice front.

The slush porosity e was set to = 1 (no slush) as slush is not dominate in the Fort Vermilion reach and only for relatively short distances downstream of Vermilion Rapids and Boyer Rapids (including Peace Point). This has become readily apparent in recent satellite images not shown herein due to space limitations. Also Beltaos (2007) found that the thermal ice thickness in three out of four years was higher at Peace Point than in the PAD reach, this possibly due to the local effect of frazil slush at Peace Point from Boyer Rapids.

The thermal ice growth Equation 1 allows for a changing snow depth with time. This is important as snowfall earlier in the winter has a larger effect in reducing the thermal ice thickness due to insolation than a snowfall later in the winter.

Equation 1 does not include the formation of snow ice (when the snowfall is heavy enough to submerge an ice cover, flood and refreeze on top of the thermal ice). However, the thermal ice thickness at each time step was tested to determine if the snow was heavy enough to submerge the ice cover and no occurrences in the 2014 and 2018 simulations were found.

Figure 8 shows the results of the thermal ice computations along with the snow on the ice cover. Figure 8a shows that the end of winter thermal ice thickness in 2014 (0.98 m) in the PAD reach was significantly thicker than in 2018 (0.69 m) while the thermal ice thicknesses in the Fort Vermilion reach were similar in the two years; as a ratio, the computed end of winter ice thickness in the PAD reach in 2018 was 70% of that in 2014. This explains why the ice was so thermally deteriorated in in the PAD reach in 2018 compared to 2014. Therefore, it would take less energy input in the spring to melt the ice cover in this reach in 2018 than it did in 2014. The reason for the higher thermal ice thickness in the PAD reach is the colder winter in 2014 than in 2018 but also reduced snow cover on the ice in 2014 than in 2018.

To demonstrate how important the timing of the snowfall is on the end of winter ice thickness, a thermal ice thickness calculation was done assuming that the snow on ice depth increased linearly from freeze-up to about the maximum value of 0.35 m in mid-March which was about the maximum value in 2014 and 2018. The results showed that this calculated end of the winter ice thickness in the PAD reach in 2018 was 92% of 2014 compared to 70% if the actual timing of snowfall was used in the simulations. This shows that the timing and amount of snowfall on the ice cover is critical in determining if there will be a competent enough ice cover in the PAD reach and upper Slave River to arrest a dynamic break-up, cause an ice jam, and flood the PAD.

Figure 8b indicates that the end of winter ice cover in the fort Vermilion reach were similar in 2014 and 2018. It is not surprising then that ice jams occurred in both years in this reach albeit in slightly different locations. However, the discharge was much higher in 2018 than in 2014 (Figure 2b) so one would presume that the break-up should have gone quicker through this reach in 2018 than in 2014 but the opposite is true (Figure 3). There may have been a difference in the ice quality in the two year to explain this discrepancy.

The ice cover in 2018 at Fort Vermilion did have an apparent snow layer on top despite the submergence criteria not being met (Figure 9). Inspection of the timing of snowfall in the region indicated heavy snowfalls in the 3 days before freeze-up in Fort Vermilion (Figure 10b). Perhaps the ice pans and large skim ice sheets had a lot of snow on them prior to their arrival at the ice front creating this snow ice layer. In 2014 the snowfall just prior to freeze-up at Fort Vermilion was not as high as it was in 2018 (Figure 10b). Snowfall preceding freeze-up in the PAD reach was smaller and not dissimilar in 2014 and 2018 (Figure 10a).

3.4 UThermal ice decay

o Cumulative degree-days above -5 P PC (∑TR-5R) can be used as an indicator of thermal ice

deterioration, Bilello (1980). Figure 11 shows a plot of ∑TR-5R for Fort Chipewyan and Fort Vermilion in 2014 and 2018.

Figure 11 shows that the PAD reach in 2014 had the lowest ∑TR-5R than further upstream at Fort Vermilion in that same year and also less than both locations in 2018. This likely enhanced the probability of jamming in the PAD reach 2014 in addition to the already thicker end of winter thermal ice cover there in 2014 (Section 3.3).

In Fort Vermilion the ∑TR-5R was lower in 2018 than it was in 2014 and this was perhaps the reason why the dynamic break-up took longer to work its way through this reach (Section 2.4) in 2018 despite the higher discharges in 2018. In addition to the possible presence of snow-ice in 2018 in the Fort Vermilion reach (Section 3.3).

3.5 UDischarge at freeze-up

Some previous studies (Prowse and Conly 1998, Beltaos et al. 2006) have theorized that because Peace River winter discharges are higher post-regulation which causes higher freeze-up ice elevations, that this higher ice level requires greater discharge in the spring to create a dynamic break-up (Beltaos et al. 2006), thereby causing the frequency of ice jams to decrease post- regulation (Beltaos 2014, 2018). Although this is recently under debate (Timoney et al. (2018), Hall et al. (2018), Beltaos (2019)), the discharge at freeze-up in the Fort Vermilion and PAD reaches of the Peace River are examined herein.

It should be noted that freeze-up stage is not the only factor affecting break-up resistance. Ice strength and thickness play an important role which are determined by many factors such as how cold the winter is, how much snow falls on the ice cover, presence and porosity of slush under the ice cover, and in the spring, how much melt and strength degradation occurs prior to the arrival of the spring freshet discharge that may or may not be sufficient to create an ice jam flood of the PAD.

There are a limited number of water level stations in the Fort Vermilion and the PAD reaches, and they are often inaccurate by several tens of metres and sometimes by more than a metre at freeze-up since the ice motion during the freeze-up process moves them around. Thus, it is better to examine the discharge at freeze-up rather than stage at these gauges. Furthermore, due to the unsteady nature of freeze-up, the freeze-up stage can be locally much different just a few km away so a single gauge is not representative of the freeze-up stage for a given reach that spans 10s of km (operational experience by the author when monitoring freeze-up on the Peace River between the Town of Peace River and Dunvegan from 2001 to 2019).

To examine the discharges in the 2013-2014 and 2017-2018 ice seasons the Comprehensive River Ice Simulation System Program (CRISSP) was used to route regulated Peace Canyon flow releases and tributary inflows to the downstream limit of the model at Fort Vermilion. The CRISSP model is described in Shen (2005) and its application to the Peace River is described in Jasek and Pryse-Phillips (2015). The other advantage of using routed flows using the CRISSP model is that it accounts for the discharge being abstracted due to upstream ice cover formation and discharge increases due to ice cover retreat downstream.

Figure 12a shows the discharge at Fort Vermilion for the two winters. On the day of freeze-up at 3 Fort Vermilion, the discharge was dropping and at about 1100 mP P/s on Nov 18, 2013 while it 3 was fairly steady at about 1700 mP P/s on Nov 24, 2017. Therefore, it is likely then that the freeze- up stage was higher in the Fort Vermilion reach in 2017 than in 2013. However, later in the 3 winter, the discharge in both years peaked at about 1800 mP P/s which would have evened out the maximum ice level to some degree leading to a similar break-up resistance if more dominant than the ice strength already discussed.

Figure 12b shows the discharge at Fort Vermilion in 2013-2014 and 2017-2018 but shifted by two days to account for the flow travel time to the PAD reach. This translation is inherent to some error because freeze-up processes between Fort Vermilion and the PAD reach is abstracting water into storage as multiple ice fronts were documented in this reach (Section 3.1). Note that winter time tributary inflows between Fort Vermilion and the PAD reach would not be significant and not that different from year to year relative to the total flow in the river. On the day of freeze-up in the PAD reach in 2013 (Nov 18) and 2017 (Nov 14), the discharges were 3 3 about 1450 mP P/s and 1600 mP P/s respectively. Therefore, it is likely then that the freeze-up stage was only slightly higher in the PAD reach in 2017 than in 2013. Again, later in the winter, the 3 discharge in both years peaked at about 1800 mP P/s which would have evened out the maximum ice level to some degree leading to a similar break-up resistance if more dominant than the ice strength already discussed. In this case, the slightly higher freeze-up stage in 2017 in the PAD reach would have been welcomed to promote ice jam formation there since in 2018 the ice deteriorated before the main jave arrived from upstream.

4. An emerging picture of Peace River break-up types that influence ice jam flooding of the Peace-Athabasca Delta

Based on what has been observed in the 2014 and 2018 break-ups, a conceptual model of break- up drivers and resistance has been developed. These are listed below. a) discharge potential for break-up: snowpack and Peace Canyon releases b) weather driver for break-up: rapid or gradual warm-up in spring c) break-up resistance upstream of the PAD reach d) break-up resistance in the PAD reach and upper Slave River

Although there is always a continuum of conditions in all of these categories and there could be a sudden shift from one state to another, (such as a rapid initial warm-up followed by a sudden and lengthy cool down during mid break-up) it helps to get an idea about potential break-up outcomes by considering these categories as binary.

Table 2 classifies the break-up drivers and resisters for the 2014 and 2018 break-ups and resulting outcomes.

Table 2. Break-up drivers and resisters for the 2014 and 2018 break-ups and resulting outcomes. Break-up drivers and 2014 2018 resisters Discharge Potential High High Weather Driver Rapid warm-up Rapid warm-up Break-up resistance upstream Low (between Vermilion Low (between Vermilion of PAD Rapids and the PAD reach) Rapids and the PAD reach) Break-up resistance in PAD High Low reach and upper Slave River Result PAD flood Dynamic break-up flush- through the PAD reach

If we take the four break-up drivers and resister categories and treat them as binary we come up with 16 outcomes, some of which may be the same or similar or even a range of results. This is th presented in a poster displayed at this 20P P Workshop on the Hydraulics of Ice Covered Rivers (Jasek, 2019c). Only 1 combination of the 16 results in a “guaranteed” ice jam flood of the PAD and another combination has the possibly of a range of outcomes that includes the potential for an ice jam flood of the PAD but it is not guaranteed. The other 14 categories result either in a thermal break-up, a dynamic break-up flush-through or a small or moderate ice jam flood in the PAD reach but unlikely extensive flooding of the PAD.

5. Summary and conclusions The 2014 and 2018 break-ups on the Peace River were examined to provide insights into the conditions needed to cause flooding of the PAD. In both years the snowpack available for run-off was similar to one another, above normal and the weather warmed quickly enough in both years to initiate dynamic break-ups on the Peace River.

The break-up of the Smoky River played a key role in initiating a dynamic break-up of the Peace River in both 2014 and 2018. Although the initial break-up of the lower Smoky and Little Smoky Rivers formed ice jams within a 100 km or so reach downstream of the Peace-Smoky River Confluence in both 2014 and 2018, it was the javes from the upper Smoky River break-up that were powerful enough to cause a sustained dynamic break-up of the Peace River in both years.

The break-up sequence in 2014 and 2018 was examined and there were some similarities and some differences. The most important difference between the two break-up sequences relative to PAD flooding is that in 2018 the ice in the PAD reach and well downstream on the Slave River melted out thermally prior to the arrival of the main break-up jave from upstream. In contrast in 2014, there was competent ice the PAD reach when the highly dynamic break-up front arrived.

The higher ice cover break-up resistance in the PAD reach in 2014 compared to 2018 resulted in an ice jam flood in the former and not the latter; in 2014 the dynamic break-up stopped and jammed about 10 km from the PAD downstream on the Slave River causing a large ice jam to flood the PAD but in 2018 the dynamic break-up flushed through the PAD reach and well downstream on the Slave River so no ice jam formed and there was no flooding of the PAD.

Antecedent conditions for the 2014 and 2018 break-ups were examined to determine what caused the different break-up results in the PAD reach. The analysis indicated that in 2014 the PAD reach had a thicker end of winter ice cover than in 2018 and also experienced less ice decay in 2014 than in 2018 prior to the arrival of the main break-up jave. This explains the difference in the ice jam flood at the PAD in 2014 and the dynamic break-up flush through the PAD in 2018. End of winter thermal ice thickness calculations indicated that the timing of snowfalls over the winter were critical in determining an accurate end of winter thermal ice cover that can be used as an index for ice jam formation to assess the possibility of ice jam floods in the PAD.

A conceptual model for determining PAD floods was developed using 4 binary categories: discharge potential, rapid weather warm up or not, break-up resistance upstream of PAD, break- up resistance in PAD reach and upper Slave River, resulting in 16 outcomes. Only 1 combination of the 16 results in a “guaranteed” ice jam flood of the PAD and another combination has the possibly of a range of outcomes that includes the potential for an ice jam flood but it is not guaranteed. The other 14 categories result in either a thermal break-up, a dynamic break-up flush-through or a small or moderate ice jam flood in the PAD reach but unlikely extensive flooding of the PAD.

6. Acknowledgments The author would like to thank Kerry Paslawski Timberoot Environmental Inc. and David Verbisky, Trek Aerial Surveys for field support during the 2018 break-up monitoring., Stéphane Hardy, Dromadaire for providing Satellite imagery, Stefan Emmer, Alberta Environment and Parks for providing break-up data and snow survey data. Hydrometric and weather data was obtained from Environment and Climate Change Canada. Also, the author would like to thank Dr. Richard Tortorella and the members of the CRIPE Scientific Committee for their thorough reviews.

8. References Ashton, G.D. 1980. Freshwater ice growth, motion, and decay. In: Colbeck, S.C., ed. Dynamics of snow and ice masses, Academic Press, 261-304.

Beltaos, S., Prowse, T.D., Carter, T., 2006. Ice regime of the lower Peace River and ice-jam flooding of the Peace-Athabasca Delta. Hydrological Processes 20(19): 4009–4029.

Beltaos, S. 2007. Hydro-climatic impacts on the ice cover of the lower Peace River. Hydrological Processes 22, 3252-3263.

Beltaos, S. (editor). 2008. River Ice Breakup. Water Resource Publications, LLC., Highlands Ranch, Colorado, USA, 462 pp.

Beltaos, S., 2014. Comparing the impacts of regulation and climate on ice-jam flooding of the Peace-Athabasca Delta. Cold Regions Science and Technology 108: 49–58.

Beltaos, S. 2018. Frequency of ice jam flooding in the Peace-Athabasca Delta. Canadian Journal of Civil Engineering. 45: 71-75.

Beltaos, S. 2019. Reply to discussions by Timoney et al. (2018) and Hall et al. (2018) on “Frequency of ice-jam flooding of Peace-Athabasca Delta”. Canadian Journal of Civil Engineering. 00: 1–6 (0000) dx.doi.org/10.1139/cjce-2018-0724.

Bilello M.A. 1980. Maximum thickness and subsequent decay of lake, river and fast sea ice in Canada and Alaska, report 8—6, U.S. Army Cold Regions Research and Engineering Laboratory: Hanover, NH, 160 p.

Hall, R.I., Wolfe, B.B., Wiklund, J.A. (2018). Discussion of “Frequency of ice jam flooding in the Peace-Athabasca Delta. Canadian Journal of Civil Engineering. 00: 1–3 (0000) dx.doi.org/10.1139/cjce-2018-0407.

Jasek, M., 2017a. The importance of break-up front celerity in the genesis of a spring flood: The 2014 Peace River Ice Jam Flood of the Peace-Athabasca Delta. Proceedings of the 19th Workshop on the Hydraulics of Ice Covered Rivers, Whitehorse, Yukon.

Jasek, M. 2017b. Peace River 2014 break-up and ice jam at the Peace-Athabasca Delta – field investigations and analysis. BC Hydro Engineering Report No. E1312.

Jasek, M. 2019a. An emerging picture of Peace River break-up types that influence ice jam flooding of the Peace-Athabasca Delta part 1: the 2018 Peace River break-up. Proceedings of the 20th Workshop on the Hydraulics of Ice Covered Rivers, Ottawa, Ontario.

Jasek, M. 2019b. Break-up of the Peace and Smoky Rivers, 2018. BC Hydro Report currently in draft.

Jasek, M. 2019c. Peace River break-up and PAD ice jam flooding flowchart. Poster presentation at the 20th Workshop on the Hydraulics of Ice Covered Rivers, Ottawa, Ontario.

Jasek, M., Pryse-Phillips, A. 2015. Influence of the proposed Site C hydroelectric project on the ice regime of the Peace River. Canadian Journal of Civil Engineering – Special Issue on River Ice Engineering. Vol. 42, No. 1, pp. 645-655.

Prowse, T.D., Conly, F.M. 1998. Effects of climate variability and flow regulation on ice-jam flooding of a northern delta. Hydrological Processes. 12, 1589-1610.

Timoney, K., Smith, J.D., Lamontagne, J.R., Jasek, M. (2018). Discussion of “Frequency of ice jam flooding in the Peace-Athabasca Delta. Canadian Journal of Civil Engineering. 00: 1–4 (0000) dx.doi.org/10.1139/cjce-2018-0409.

Shen, H.T. 2005. CRISSP1D programmer’s manual. In CEATI Report No. T012700-0401. Department of Civil Engineering, Clarkson University, Potsdam, N.Y.

a) b) c)

Figure 1. SWE as a % of normal snow survey results for the Smoky and Peace River Basins in Alberta a) April 1, 2014. b) April 1, 2018. c) April 1, 2018 as a % of April 1, 2014.

30 a) 2014 - Smoky River at Watino 20 2018 - Smoky River at Watino C) o 10

-10 Air Temperature ( Air Temperature

-20

-30 1-Apr 6-Apr 11-Apr 16-Apr 21-Apr 26-Apr 1-May 6-May 11-May b) 9000 2014 Discharge at the Town of Peace River 8000 2018 Discharge at the Town of Peace River 7000

6000 /s) 3 5000

4000 Discharge (m Discharge 3000

2000

1000

0 1-Apr 6-Apr 11-Apr 16-Apr 21-Apr 26-Apr 1-May 6-May 11-May Figure 2. a) April – May 2014 and 2018 hourly air temperatures at the Smoky River at Watino and b) 5 minute discharge at the Town of Peace River. Slave at Fitzgerald gauge 1350

1300

1250 Start of Slave River Rocky Point 1200 Sweetgrass Landing PAD

1150 Peace Point Boyar Rapids 1100

1050

1000 Garden Creek

950 Vermilion Rapids

900 Wabasca River

850 Fort Vermilion

800

750

Distance from Bennett Dam Dam from Bennett (km) Distance 700 Tompkins Landing Locations of interest 2014 Break-up Fronts Carcajou 650 2014 Ice Jams 2018 approximate break-up fronts based on indirect data and rate of travel assumptions 600 2018 Break-up Fronts Notikewin River 550 2018 Ice Jams 2018 approximate break-up fronts based on indirect data and rate of travel assumptions 500 Sunny Valley Cadotte River 450 Berreth Flats Gauge 400 Town of Peace River Smoky River 350 15-Apr 16-Apr 17-Apr 18-Apr 19-Apr 20-Apr 21-Apr 22-Apr 23-Apr 24-Apr 25-Apr 26-Apr 27-Apr 28-Apr 29-Apr 30-Apr 1-May 2-May 3-May 4-May 5-May 6-May 7-May 8-May 9-May 10-May Date in 2014 or 2018

Figure 3. Break-up fronts and ice jams on the Peace River in April and May 2014 and 2018.

Figure 4. Comparison of the ice conditions on the Slave River just downstream of the PAD in 2014 and 2018. The 2014 ice cover is present, white and competent. The 2018 ice cover is missing in large sections, dark and appears weak. a) b)

c) d)

Figure 5a. Ice covers prior to break-up in 2018 were white and competent a) downstream of a major ice jam at km 724 on Apr 27, b,c) downstream of an ice jam that caused flooding in Fort Vermilion on Apr 29. d) looking upstream from km 886 on Apr 29. a) b)

c) d)

Figure 5b. Ice covers prior to break-up in 2018 were dark and deteriorated. a) km 947, b) PAD reach Moose Island crossing at km 1218, c-d) PAD reach from Moose Island crossing to Rocky Point (km 1218 – 1228). Photographs taken Apr 29, 2018. Slave at Fitzgerald gauge 1350 a) 1300 Start of Slave River 1250 Rocky Point 1200 Sweetgrass Landing PAD Peace Point 1150 Boyar Rapids 1100 1050 Locations of interest 1000 Garden Creek 2013 - 2014 Freeze-up Ice Front or Break-up Front or toe of Break-up ice jam 950 Vermilion Rapids 2013 - 2014 approximate fronts based on indirect data and rate of travel assumptions 900 Wabasca River

850 Fort Vermilion 2017 - 2018 Freeze-up Ice Front or Break-up Front or toe of Break-up ice jam 800 2017 - 2018 approximate fronts based on indirect data and rate of travel assumptions 750 700 Tompkins Landing Carcajou 650 600 Notikewin River 550 500 Sunny Valley Cadotte River

Distance from Bennett Dam (km) Bennett from Distance 450 Berreth Flats Gauge Town of Peace River 400 Smoky River 350 300 Dunvegan 250 200 BC-AB Border 150 100

50 Peace Canyon Dam 0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 17-Apr 2-May Date in 2014 or 2018

b) 30 2013 - 2014 - Town of Peace River 20 2017 - 2018 - Town of Peace River

10 C) o

-10

Air Temperature ( Air Temperature -20

-30

-40 1-Nov-17 16-Nov-17 1-Dec-17 16-Dec-17 31-Dec-17 16-Jan-18 31-Jan-18 15-Feb-18 2-Mar-18 17-Mar-18 2-Apr-18 17-Apr-18 2-May-18

Figure 6. a) Freeze-up ice fronts, break-up fronts and toe of ice jam locations for the 2013-2014 and 2017-2018 ice seasons. b) Mean daily air temperatures at the Town of Peace River for the 2013-2014 and 2017-2018 ice seasons. a) 0 2013 - 2014 - Fort Chipewyan -500 Days) 2013 - 2014 - Fort Vermilion - C

o 2017 - 2018 - Fort Chipewyan -1000 2017 - 2018 - Fort Vermilion

-1500

-2000

-2500

-3000 Cummulative Freezing Temperatures ( Cummulative Freezing Temperatures -3500 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 17-Apr 2-May 17-May 0.6 b) 2014 - Snow on Ice based on Fort Chipewyan snow on ground 2014 Freeze-up in PAD Reach 0.5 2018 - Snow on Ice based on Fort Chipewyan snow on ground 2018 Freeze-up in PAD Reach 0.4

0.3

0.2 Snow Depth on Ice (m) Ice on Depth Snow

0.1

0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 17-Apr 2-May 17-May 0.6 c) 2014 - Snow on ice based on High Level snow on ground 2014 Freeze-up in Fort Vermilion 0.5 2018 - Snow on ice based on Fort Vermilion snow on ground 2018 Freeze-up in Fort Vermilion 2018 - Snow on ice based on High Level snow on ground 0.4

0.3

0.2 Snow Depth on Ice (m) Ice on Depth Snow

0.1

0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 17-Apr 2-May 17-May o Figure 7. a) Cumulative Freezing P PC-Days at Fort Chipewyan and Fort Vermilion after freeze-up in the PAD and Fort Vermilion reach respectively b) Snow on ice in PAD reach based on snow on ground in Fort Chipewyan, c) Snow on ice in Fort Vermilion reach based on snow on ground in Fort Vermilion and High Level, for the 2013-2014 and 2017-2018 ice seasons. a) 0.6 0.4

0.2

0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 17-Apr 2-May 17-May -0.2 ve & Snow on Ice (m) +ve (m) Ice on Snow & ve - -0.4

-0.6

-0.8 2013 - 2014 - Thermal Ice Thickness based in Fort Chipewyan data 2017 - 2018 - Thermal Ice Thickness based in Fort Chipewyan data

Thermal Ice Thickness (m) (m) Ice Thickness Thermal -1 2013 - 2014 - Snow on Ice based on Fort Chipewyan snow on ground 2017 - 2018 - Snow on Ice based on Fort Chipewyan snow on ground -1.2 b)0.6 0.4

0.2

0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 17-Apr 2-May 17-May -0.2 ve & Snow on Ice (m) +ve +ve (m) Ice on Snow & ve - -0.4

-0.6

-0.8 2013 - 2014 - Thermal Ice Thickness based on High Level Data 2017 - 2018 - Thermal Ice Thickness based on Fort Vermilion data 2017 - 2018 - Thermal Ice Thickness based on High Level Data 2013 - 2014 - Snow on ice based on High Level snow on ground ThermalIce Thickness(m) -1 2017 - 2018 - Snow on ice based on High Level snow on ground 2017 - 2018 - Snow on ice based on Fort Vermilion snow on ground -1.2

Figure 8. Computed thermal ice thickness for 2013-2014 and 2017-2018 ice seasons in a) the PAD reach and b) the Fort Vermilion reach of the Peace River. The 2014 snow on ground data from Fort Vermilion was not available so computations based on nearby High Level data is shown for both years.

Figure 9. Snow ice layer apparent in the ice cover Apr 29, 2018 after break-up at Fort Vermilion. 1.5 15 a)

1 10

0.5 5 ve & Snow on SnowIce & (m)ve +vee - 0 0

2013 - 2014 - Thermal Ice Thickness based in Fort Chipewyan data

-0.5 -5 SWE (mm) Daily Precipitation 2017 - 2018 - Thermal Ice Thickness based in Fort Chipewyan data

2013 - 2014 - Snow on Ice based on Fort Chipewyan snow on ground Thermal Ice Thickness(m) -1 2017 - 2018 - Snow on Ice based on Fort Chipewyan snow on -10 ground 2013 - Fort Chipewyan Daily Precipitation

-1.5 -15 1-Nov 6-Nov 11-Nov 16-Nov 21-Nov 26-Nov 1-Dec 6-Dec 11-Dec 16-Dec 21-Dec 26-Dec 31-Dec

1.5 15 b)

1 10

0.5 5 ve & Snow on SnowIce & (m)ve +ve - 0 0

2013 - 2014 - Thermal Ice Thickness based on High Level Data

-0.5 -5 SWE (mm) Daily Precipitation 2017 - 2018 - Thermal Ice Thickness based on Fort Vermilion data 2017 - 2018 - Thermal Ice Thickness based on High Level Data

Thermal Ice Thickness(m) -1 -10 2013 - 2014 - Snow on ice based on High Level snow on ground

2017 - 2018 - Snow on ice based on High Level snow on ground

-1.5 -15 1-Nov 6-Nov 11-Nov 16-Nov 21-Nov 26-Nov 1-Dec 6-Dec 11-Dec 16-Dec 21-Dec 26-Dec 31-Dec

Figure 10. Computed thermal ice thickness, snow on the ice cover and daily precipitation during the initial freeze-up periods in 2013 and 2017 at a) PAD reach and b) Fort Vermilion reach. 400

2014 - Fort Chipewyan cumulative temperatures above -5 deg-C 350 2018 - Fort Chipewyan cumulative temperatures above -5 deg-C 2014 - Fort Vermilion cumulative temperatures above -5 deg-C 300 C

o 2018 - Fort Vermilion cumulative temperatures above -5 deg-C 5 - Total prior to break-up or date of main jave arrival 250 Days above

- 200 C o

150 Cummulative 100

0 1-Apr 6-Apr 11-Apr 16-Apr 21-Apr 26-Apr 1-May 6-May 11-May o o Figure 11. Cumulative P PC-Days above -5 P PC for Fort Chipewyan and Fort Vermilion in 2014 and 2018. 2500

2000

1500 /s) 3

1000 2014 Discharge at Fort Vermilion

Discharge (m Discharge 2014 Freeze-up in Fort Vermilion 500 2018 Discharge at Fort Vermilion 2018 Freeze-up in Fort Vermilion

0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr 2500

2000

1500 /s) 3

1000 2014 Discharge at Fort Vermilion shifted ahead 2 days

Discharge (m Discharge 2014 Freeze-up in PAD Reach 500 2018 Discharge at Fort Vermilion shifted ahead 2 days 2018 Freeze-up in PAD Reach

0 1-Nov 16-Nov 1-Dec 16-Dec 31-Dec 16-Jan 31-Jan 15-Feb 2-Mar 17-Mar 2-Apr Figure 12. Computed discharge for 2013-2014 and 2017-2018 from CRISSP model a) Fort Vermilion and b) Fort Vermilion discharge shifted 2 days to approximate discharge in the PAD reach.