Freeze-up Jam Observations on the Dauphin River

Transcription

1 CGU HS Committee on River Ice Processes and the Environment 19 th Workshop on the Hydraulics of Ice Covered Rivers Whitehorse, Yukon, Canada, July 9-12, Freeze-up Jam Observations on the Dauphin River Lucas Wazney 1a, Shawn P. Clark 1b, and Alexander Wall 1c 1 Department of Civil Engineering, University of Manitoba, 15 Gilson Street, Winnipeg, Manitoba R3T 5V6 1a umwaznel@myumanitoba.ca 1b Shawn.Clark@umanitoba.ca 1c Alexander.Wall@umanitoba.ca The 52 km long Dauphin River that drains Lake St. Martin into Lake Winnipeg in Central Manitoba has been selected as an excellent location to conduct detailed field investigations of freeze-up jam formation. This undertaking is part of the NSERC / Manitoba Hydro Industrial Research Chair in River Ice Engineering. The winter field monitoring season presented another extensive freeze-up jam, caused by the relatively steep slope that occurs on the lower 11.2 km of the river. While previous years field work relied on data from instrumentation deployed throughout the entire winter, the data during this season was supplemented by two field trips that were perfectly timed to coincide with the active freeze-up jam formation. Visual observations from the shoreline during ice shoving events, as well as unmanned aerial vehicle observations of these events, provided the authors with a much more detailed perspective on the processes involved in the ice jam development that often accumulates to thicknesses in excess of 3 m. These observations will be presented in detail, combined with the corresponding water level data and trail camera imagery from selected monitoring locations.

2 1. Introduction The study of river ice dynamics has been given greater attention at breakup than at freeze-up. This is perhaps because breakup ice jams can form rapidly and cause backwater effects that can flood land very quickly. The thickness of a breakup jam has historically been modeled using theory derived from soil mechanics; the ice is thought to behave as a granular material with strength (resistance) resulting largely from frictional interlocking of the ice blocks. As confining pressures on the ice accumulation increase, so too does the cover s ability to resist the external forces of water shear and gravity attempting to drive it downstream. Freeze-up ice jams differ from breakup jams in that the air temperature is below 0 C, causing continual heat loss from the river (Michel 1991). However, they are often modeled in the same way as breakup jams. Michel (1991) noted that the poor documentation of freeze-up ice phenomena in the field results in the inability to reject numerical model outputs which are based on these assumed behaviours. The formation of additional ice, in the form of sintering between ice floes or the freezing of interstitial water between ice floes at the surface, can add substantial strength and resistance to mechanical thickening (Michel 1971). The effect of freeze-bonding has been shown to add strength to a floating ice accumulation in laboratory experiments by Schaefer and Ettema (1985) and Urroz and Ettema (1987). It has been observed in the field that ambient air temperatures during freeze-up are linked to ice dynamics, and can control whether an ice cover is more prone to frontal progression or shoving events (Andres 1999, Michel 1984). Michel (1991) suggested that shoving is much less likely during freeze-up due to freezing effects, unless the ice front progresses very quickly. This paper will summarize field observations made on the Dauphin River during freeze-up Site visits were timed to coincide with ice cover progression through the relatively steep lower reach. Persistent cold weather resulted in high volumes of incoming ice during cover advancement, resulting in a leading edge that was poorly defined (i.e. there was no distinct separation between a stationary cover and incoming frazil pans). Instead, the authors observed the upstream portion of the ice cover to be an actively consolidating transition zone, several hundreds of metres long, which included regions of swiftly moving surface ice pans, followed by a gradual deceleration zone, and eventually a stationary zone. The ice could appear stationary at a given location for a short time (on the order of minutes or longer), then suddenly the cover would mobilize and the ice would shove towards a stationary point located some distance downstream. Qualitative observations made from shore and with an unmanned aerial vehicle (UAV) were supplemented by water level measurements and meteorological data to understand the process of ice cover progression on the lower Dauphin River. 2. Overview of Dauphin River Freeze-up Monitoring Program The Dauphin River freeze-up ice regime has been monitored since as part of the NSERC / Manitoba Hydro Industrial Research Chair in River Ice, with efforts intensifying in each subsequent year. In , five monitoring locations were added, bringing the total to 16 (13 of which are in the lower 12 km reach). Figure 1 shows a map of the study area, while Table 1 shows the equipment deployed at each monitoring location for The equipment was installed over two site visits conducted on October and October 18-20, 2016.

3 DRLL01 DRLL02 DRLL03 DRLL04 DRLL04a DRLL05 DRLL05a DRLL06 DRLL06a DRLL06b DRLL07 DRMET DRLL08 DRLL08a DRLL09 DRLL10 ± Kilometres River DRLL03 05LM006 Lake Winnipeg Dauphi n DRLL02 DRLL01 Lake St. Martin Kilometres Source: Esri, DigitalGlobe, GeoEye, Earthstar Geographics, CNES/Airbus DS, USDA, USGS, AeroGRID, IGN, and the GIS User Community Figure 1. Map of Dauphin River monitoring locations. Chainage along the lower reach of the river is shown in red. Environment Canada station 05LM006 is also shown for reference. Table 1. Instrumentation deployed for freeze-up monitoring season. Water Level Logger Water Temperature Logger Trail Camera Barometric Pressure Logger Meteorological Station Also new to monitoring efforts in was the attempt to time a site visit to coincide with dynamic ice cover advancement from DRLL08 to DRLL04 (past observations have shown the cover likely becomes grounded near DRLL08a). To help achieve this goal, a satellite-linked camera was installed at DRMET, which took pictures hourly and sent one photograph daily to a website to be viewed in the office at the University of Manitoba. Site visits were conducted on December 9 and December 12-14, 2016 to obtain visual observations of the forming ice cover. A survey was conducted using a Leica real time kinematic (RTK) unit on February 21-22, 2017 to obtain the top of ice profile. Site visits were also conducted on March and May 15-16, 2017 to survey near-shore ice transects and bank elevations using the RTK unit and a terrestrial

4 Elevation [m] DDOF [ C-d] laser scanner to obtain an estimate of the ultimate ice thickness at select locations. A final site visit on May 24-25, 2017 was conducted to collect the deployed equipment. 3. Timing of Site Visit during Freeze-up Air temperature and water level data collected in and were analyzed to obtain an estimate of the number of degree days of freezing (DDOF) that are required to (i) develop a competent ice cover at the outlet to Lake Winnipeg, and (ii) coincide with ice cover progression through the steeper, lower reach. It was found that approximately 50 DDOF were required to initiate cover progression near DRLL10, while at 150 DDOF the cover was progressing up the river in the region near DRLL06 and DRLL05. Figure 2 shows water surface elevations at DRLL06 and DRLL05 in and , respectively, and their corresponding DDOF DRLL06 ( ) DRLL05 ( ) DDOF ( ) DDOF ( ) Nov 12 Nov 17 Nov 22 Nov 27 Dec 02 Dec 07 Dec 12 Dec 17 Dec 22 Dec 27 Figure 2. Water levels in steeper reach and corresponding degree days of freezing (DDOF) in and At approximately 150 DDOF each year, the ice cover appears to be advancing up the steeper reach. In order to time the site visit during freeze-up successfully, weather data from Fisher Branch was monitored daily. Using the recorded air temperatures, a cumulative DDOF value was calculated. Forecasted air temperatures were used to project the cumulative DDOF value a week in advance. On December 8, the cumulative DDOF was approximately 40 C-d, and the forecast for the following week called for daily average temperatures consistently below -20 C. Unfortunately, around this time the satellite-linked camera had battery problems and was not sending daily photographs to the website. To err on the side of caution, a one day site visit was conducted on December 9 to assess the ice cover on the river and fix the satellite-linked camera. Based on the observed state of the ice cover, another site visit was planned for December Observations during Freeze-up The first site visit was made on December 9, 2016, corresponding to approximately 50 DDOF. Efforts were focused on the portion of the river downstream of DRLL06a, as ice was observed to be freely moving upstream of this location. Observations began at DRLL10 at approximately 12:00 pm. At each site, photographs and video were taken from shore. In the first pass from

5 DRLL10 to DRLL07 (observations ranged from 12:00 pm to 2:00 pm), the cover appeared to be rough and stationary, but thinner than the final cover configuration observed in Blocks of ice, estimated to be on the order of 5-10 cm or less in thickness, protruded from the top of the ice cover. Shear planes were evident near the banks; often more than one shear plane was present, possibly indicating that more ice remained attached to the shore with each subsequent shove of the ice cover at that location. Border ice was composed of frazil pans that were solidified together by the freezing of interstitial water. The ice cover near DRLL08a appeared thicker and rougher than the cover downstream, indicating that the toe of the jam may be near this location (similar to previous years). Figure 3 shows the ice cover at DRLL07 at 1:56 pm. Figure 3. Stationary ice cover at DRLL07 at 1:56 pm on December 9. Flow is from right to left. Buffalo Creek can be seen in the background. The ice cover at DRLL06a (about 1.4 km upstream of DRLL07) was observed at 2:30 pm to be moving at an estimated velocity of about 0.5 m/s. The cover at this location had nearly 100% surface ice concentration, and was composed of relatively thin pans that appeared to be partially frozen to each other as they moved downstream. There were not distinct ice floes with open water separating them. A photograph taken from shore is shown in Figure 4.

6 Figure 4. Ice cover at DRLL06a at 2:30 pm on December 9. While conducting a UAV flight from 2:50-3:00 pm at DRLL06b, the cover at this location and downstream near DRLL07 began moving. The appearance of the ice cover at this time is shown in Figure 5. Figure 5. UAV photo at DRLL06b looking downstream (taken at 2:50 pm on December 9). Note the shear planes along the river banks. Cover was stationary at the time of the photo, but portion between the shear lines was found to be moving at 2:57 pm during shoving event. Observations made from 3:00-3:15 pm at locations DRLL07 to DRMET during the shove event were as follows. The entire cover mobilized, and was moving with average velocity of about m/s. Portions of the cover, especially near the banks, were moving at different rates, causing ice pieces to move on top of each other. The cover was making a consistent sound of ice

7 sliding and crushing against itself. The cover appeared to be getting rougher on the surface, possibly indicating the same was true for the underside of the cover. An immediate increase in water level was evident; the road between DRLL07 and DRMET was suddenly flooding over at its lowest point. In total, the shove event lasted approximately 20 minutes. The appearance of the ice cover from DRLL07 to DRMET during the shove is shown in Figures 6-8. Figure 6. Ice cover at DRLL07 at 3:05 pm on December 9 during shove event. The entire cover was moving, right up to the bank. Flow is from right to left. Figure 7. Ice cover between DRLL07 and DRMET at 3:08 pm on December 9 during shove event. Ice piled up on the bank in the foreground was stationary, but the majority of the cover was moving with an estimated velocity of about m/s.

8 Figure 8. Ice cover just downstream of DRMET at 3:17 pm on December 9. The photograph was taken just as the cover became stationary again. The water levels near DRMET show the rapidity and severity of the backwater effect caused by the shove. The data shows that the cover mobilized at DRMET at about 2:18 pm, when the water elevation was m. The shove first caused the water level at DRMET to drop to m, as the water temporarily came out of storage. Approximately 30 minutes later, the water levels at DRMET began to increase abruptly, indicating that the jam attained a new, stable configuration downstream and began causing a backwater effect again. Approximately 30 minutes later, at 3:18 pm, the water level reached a peak value of m, then decreased to m, possibly due to the surge of water temporarily increasing discharge. A similar water level pattern is shown at DRLL07. Upstream, at DRLL06b, the shove is signified by a sharp decrease in water level at 2:50 pm, followed by a steady increase when the cover stabilized downstream. The water level at DRLL06a drops slightly later, at 3:00 pm, and shows a similar steady increase thereafter. Downstream, at DRLL08 and DRLL08a, the water levels show a response to the surge of water passing each location, but levels return to their original values following the surge. This indicates that the cover did not thicken appreciably at, or just downstream of, these locations. The timing and sequence of water level changes suggest that the cover failed between DRMET and DRLL08 first, causing the cover to mobilize from downstream to upstream (e.g. DRLL06b responds before DRLL06a). These observations are summarized in Figure 9. UAV images taken after the shove indicated that the downstream extent reached just upstream of DRLL08, meaning that about 2 km of the stationary ice cover mobilized during the event (from DRLL06b to DRLL08). The portion of the cover that collapsed became visibly rougher on the surface than the portion downstream that remained stable (see Figure 10). It is likely that the underside of the cover also became rougher in the shoved region, and, coupled with the increased thickness, caused the increased backwater effect on water levels upstream. Quantifying the change in surface roughness before and after the shove event would be beneficial for numerical model applications.

9 Elevation [m] DRLL06a DRLL06b DRLL07 DRMET DRLL08 DRLL08a Surge of water from shove continues past DRMET Water reaches new level based on steady discharge and ice configuration downstream Propagation of surge downstream Cover mobilizes, water comes out of storage Jam attains stable configuration downstream Dec 9 6:00 AM Dec 9 12:00 PM Dec 9 6:00 PM Dec 10 12:00 AM Time Figure 9. Water levels near DRMET during shove event witnessed on December 9. Measurements were taken at 10 minute intervals. Figure 10. UAV photograph taken on December 9 at 3:30 pm after the shove event. View is looking downstream from DRMET. A distinct change in thickness is evident upstream of DRLL08, indicating the extent of the shove event.

10 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM 12 AM 2 AM 4 AM 6 AM 8 AM 10 AM 12 PM 2 PM 4 PM 6 PM 8 PM 10 PM 12 AM Elevation [m] The water levels between 8:00-9:00 pm on December 9 indicate another minor shove event; this time the record at DRLL07 shows the passing of the surge, but no substantial thickening. The cover likely failed between DRLL06b and DRLL07 to cause this event. The attenuation and dampening of the surge can be seen by observing the water level response at downstream locations. Another site visit was conducted from December 12-14, Monitoring efforts were focused mainly on site DRLL05. Observations made from shore were supplemented by UAV photographs and video. A plot of the water surface elevations during these observations is shown in Figure DRLL04a DRLL05 DRLL05a DRLL06 DRLL06a Dec 11 Dec 12 Dec 13 Dec 14 Figure 11. Water levels near DRLL05 from December Measurements were taken at 10 minute intervals. At 1:05 pm on December 12, the cover at DRLL05 was thin, rough, and still moving at approximately 0.5 m/s. There were frozen portions along the banks that were stationary. At 2:30 pm, the cover at DRLL05a was observed to be completely stationary and rough. A trail camera was set up to take pictures at 5 minute intervals at this location to capture any further shoving events. At 3:00 pm, the cover at DRLL07 was also stationary and rough, with some open water leads. Frazil slush could be seen appearing from underneath the cover in these open leads, suggesting that under ice transport of slush could be helping to erode and smoothen the under surface of the cover. The cover at DRMET at 3:10 pm was observed to be stationary and rough, looking very similar to its appearance on December 9. Another shove event took place around 4:00 pm on December 12, sending a surge of water downstream and thickening the cover at DRLL05a, DRLL06, and DRLL06a. Trail camera images at DRLL05a during this event are shown in Figure 12. DRLL05 was revisited at 4:15 pm. At this time the cover was moving slowly. A few hundred metres upstream of DRLL05, UAV video showed a gradual transition zone from free-moving, discrete ice floes to a slower moving cover made up of attached floes and slush that covered nearly 100% of the river (see Figure 13). Figure 14 shows a photograph taken from shore at DRLL05 at 4:22 pm on December 12.

11 Figure 12. Trail camera images at DRLL05a during shove event on December 12. Figure 13. Ice conditions near DRLL05 at 4:15 pm on December 12. Flow is from bottom to top. In the foreground, discrete ice floes with areas of open water and frazil slush move downstream freely. A ridge can be seen where the backwater effects of the ice cover cause the floes to slow down.

12 Figure 14. Moving ice cover at DRLL05 at 4:22 pm on December 12 (left) and at 9:13 am on December 13 (right). In both pictures, the ice at the banks is stationary while the portion in the middle is moving downstream. As shown in Figure 11, another significant shove event occurred overnight at about 3:00 am. Unlike the shove on December 9, the sequence of water level responses indicate that the collapse of the ice cover during this event cascaded from upstream to downstream (i.e. DRLL04a responds first, followed by DRLL05, etc.). At 9:13 am on December 13, the cover at DRLL05 was observed to be much thicker than the previous evening (see Figure 14), but the ice was still moving with a relatively high velocity in the centre of the channel (estimated at about m/s). As shown in Figure 14, ice blocks about cm thick were protruding from the ice surface at the banks. By about 9:20 am, the cover had temporarily stopped moving at DRLL05. Figure 15 shows a UAV image at this time, indicating an open water area existed just before a meander downstream of DRLL05. Cracks running in the streamwise direction were observed in the halted cover. Moving ice continued to pack against the stationary cover a few hundred metres upstream of DRLL05. Figure 15. UAV photograph taken at DRLL05 at 9:20 am on December 13. View is looking downstream.

13 At 9:58 am on December 13, the cover at DRLL05 began moving slowly again. The ice cover a few hundred metres downstream remained stationary, resulting in the open water section to remain. The cover at DRLL05 stopped again at about 10:10 am. This event was not detectable in the water level records. From 10:24-10:37 am, another shove occurred at DRLL05. The ice cover moved notably faster during this shove than the previous one, reaching an estimated velocity of about m/s. This time, the cover downstream of DRLL05 mobilized as well, causing the open water area to close up. The stationary front worked its way past DRLL05 at approximately 10:37 am. After this shove event, the water level rose enough to flood the interstitial ice that was present between the protruding ice blocks at the bank. Although smaller than some other events, this second observed shove event is detectable in the water level records in Figure 11. In general, over the day the cover would go through cycles of being stationary, then shoving downstream. As the cover became thicker at a given site, shove events became less frequent. At 2:49 pm on December 13, a thin, moving layer of frazil slush (near 100% surface concentration) was observed from shore at DRLL04a. The cover was observed to be stationary at DRLL05 just prior to this. At 3:00 pm, a few hundred metres downstream of DRLL04a, the moving frazil slush was observed to be slowing down, creating a parabolic front as fast moving ice packed against the slower moving ice downstream (similar to Figure 13). The parabolic front progressed upstream, although the cover itself was still moving downstream. At 4:27 pm the cover at DRLL05 was stationary, but the ice near the banks was more flooded than what was observed around 2:45 pm. The cover appeared to have moved again since this time as well. A large crack could be seen in the cover, extending from the bank and angled upstream towards the centre of the channel. Recorded water levels indicate another shove event occurred at approximately 6:30 pm on December 13. Again, this event appears to have started from an upstream shove that cascaded downstream. On December 14 at 9:00 am, water levels at DRLL04 were observed to be much higher than the previous evening, flooding a low lying section between the river and PTH 513. At 9:10 am at DRLL05, conditions were noticeably different from the previous evening as well (see Figure 16). The cover near the centre of the channel appeared rougher, and the crack had closed up. Along the banks, large ice fragments had been pushed up and were riding up against trees. The shear lines that once distinguished shore-fast ice from the main channel ice cover were now gone. In Figure 17, trail camera images at DRLL05a show that the cover became much thicker and rougher overnight at this location as well. At 9:21 am, DRLL06 looked very similar to the previous day (observed at 12:39 pm). As shown in Figure 18, the ice fragments along the banks had not moved at all, and the cover near the centre of the channel was not noticeably thicker or rougher. Water levels in Figure 11 corroborate the visual observations at these three sites.

14 Figure 16. Photographs at DRLL05 taken at 4:27 pm on December 13 (left) and 9:10 am on December 14 (right). Figure 17. Trail camera images at DRLL05a taken at 4:30 pm on December 13 (left) and 8:30 am on December 14 (right). A significant shove event overnight caused the ice cover to become much thicker and rougher. Figure 18. Photographs at DRLL06 taken at 12:39 pm on December 13 (left) and 9:21 am on December 14 (right). No significant changes occurred to the cover overnight. 5. Observed Process of Cover Advancement Prior to the site visits in December 2016, it was envisioned that the ice cover would have a distinct ice front as it advanced upstream from Lake Winnipeg. Due to the relatively steep slope of the lower reach (0.16%), ice floe stability was predicted to be an important process that may determine the rate at which the ice front advanced. Instead, field observations indicated that a well-defined ice front did not exist. At a given time, it was possible to find a location where incoming ice transitioned from fast, free-moving ice to a thin, stationary cover. The incoming ice had a surface concentration of nearly 100%, and took the form of small floes that were partially frozen together and spread out over the entire river width (see Figure 4). Ice contacting

15 the stationary front (which was often parabolic in shape) would typically ride up on the cover and remain at the surface. The classical underturning/sinking of ice floes that depends on the Froude number and water velocity at the leading edge was not evident in the observations. A simplified, conceptual schematic of the observed ice cover progression is shown in Figure 19. At a given time during freeze-up, the ice covered portion of the river can be broken down into separate regions, depending on the approximate average cover thickness (related to the degree of consolidation). stationary front Lake Winnipeg Region 3 Region 2 Region 1 moving ice (~100% surface concentration) Region 5 Region 4 Figure 19. Conceptual schematic of ice progression on the lower Dauphin River. Suppose first that all the regions are stable in the configuration shown. As the incoming ice continues to pile onto the stationary cover of Region 1, the stationary front moves upstream. As the cover in Region 1 lengthens, eventually it will suddenly begin to move, and the stationary front jumps to a new location downstream, where the ice is thicker. The stationary front then progresses upstream from this new location, as the mobilized ice upstream packs against the stable cover. Based on the observations, it is hypothesized that in order to trigger a shove event in Region i, the cover length in Region (i 1) must attain a critical value. Referring to Figure 19, the cover in Region 1 would progress upstream through a series of small shoves at a relatively high frequency. At each shove, the stationary front jumps to the interface between Region 1 and Region 2. Each subsequent shove lengthens the cover in Region 2. Eventually, a shove will occur that increases the cover length in Region 2 beyond the critical length, triggering a higher order shove. In this case, the stationary front will jump to the interface between Region 2 and Region 3, and the shove will cause the cover in Region 3 to increase in length. This

16 process will repeat itself until a shove causes the cover in Region 3 to exceed its critical length, triggering a third order shove that effectively increases the cover length of Region 4. The simplified schematic only shows five regions, each with a relatively uniform ice thickness. In reality, the thickness is likely not uniform within each region, and more regions exist. It does, however, provide an idealized conceptualization of the relationship between frequency of a shove event and its severity as evidenced in the water level data. The frequency of a shove event may also impact how cold air temperatures add strength to a forming ice cover, and vice versa. If a region towards the toe of the ice cover does not experience frequent, small shoves (but rather large, occasional shoves), this gives more time for the freezing effects to add strength to the cover and potentially prevent these shoves from occurring altogether, resulting in a stronger, thinner cover. On the other hand, it is possible that the strength addition from freezing effects causes shoves that are less frequent, but more catastrophic than what would be experienced under mild weather conditions. In this case, the ultimate ice thickness may attain greater values in cold weather than mild weather. The occurrence of subsequent shoves is also impacted by the discharge that caused the mechanical thickening of the cover during the prior shove. As seen in the water level records, each shove involves the release of water from storage which sends a surge downstream. During the passing of this surge, the discharge is temporarily increased (i.e. Q surge > Q steady ). If the cover at a given location collapses under the external forces applied by Q surge, then it will take more force to cause a subsequent collapse of the ice cover at that location. Andres et al. (2003) used the term primary consolidation to describe the small, frequent shoves that occur near the upstream end of the cover very early in the formation period, and before interstitial freezing develops. They denoted the collapse of an existing consolidated cover as secondary consolidation. They further note that three conditions can trigger secondary consolidation events: (i) rapid advancement of the leading edge that increases the external forces on the ice cover faster than the strength addition by thermal crust growth, (ii) surges from upstream shoves that cause instability and exceed previously established equilibrium, and (iii) significant changes in air temperature that melt the crust at the surface of an ice cover that provides the majority of the resistance to mechanical thickening. During the freeze-up period of the Dauphin River, air temperatures were persistently cold, disqualifying condition (iii) above as a probable cause of secondary consolidation events. Based on the visual observations during site visits and the water level data collected, conditions (i) and (ii) were likely the governing factors in the major shoving events. The ice supply from upstream was consistent and abundant, likely causing a steady increase to the forces attempting to drive the cover downstream. When the resisting forces were exceeded, a secondary shove event took place. Even small shove events sent surges of water downstream that could be detected in the water level records and in trail camera images. These surges could also contribute to destabilizing the ice cover and lead to secondary shoving events. 6. Estimation of Ultimate Ice Thickness The equilibrium thickness of an ice jam can be calculated using Equation 1 (Beltaos 1983):

17 Equilibrium Thickness [m] e = S o B 2μ(1 s i ) (2f o ) ( f i ) μ ( 1 s ( q i gs ) ) o f o s i S o B [1] { } where e is the equilibrium ice thickness, S o is the bed slope, B is the top width, μ is the strength coefficient, s i is the specific weight of ice, f o is the composite friction factor, f i is the ice friction factor, q is the unit discharge, and g is gravitational acceleration. To obtain a ballpark estimate for equilibrium thickness in the lower Dauphin River, Equation 1 was used with typical values suggested by Beltaos (1995), which may be more suitable for breakup jams; μ = 1.2, s i = 0.92, f o = 0.2, and ( f i ) = 1.2. Cohesion representing freeze-bonding in the ice cover is not included f o in this formulation. Figure 20 shows the calculated values of equilibrium ice thickness for a range of discharges and top widths corresponding to locations DRLL05, DRLL05a, and DRLL06. The discharge recorded at Environment Canada station 05LM006 ranged from approximately m 3 /s during the freeze-up period in December B = 170 m (DRLL05) B = 150 m (DRLL06) B = 130 m (DRLL05a) Discharge [m 3 /s] Figure 20. Calculated ranges for equilibrium thickness at specific locations along the lower Dauphin River. As mentioned in Section 2, surveys were conducted during site visits on March and May During the first survey, ice was still attached to the upper banks of the river, and towards the centre of the channel the cover had dropped to rest on the lower bank as the water levels decreased at the start of breakup. RTK surveys were conducted at three locations: DRLL05, DRLL05a, and DRLL06. A photograph of the ice conditions at DRLL05 on May 15 is shown in Figure 21. The surveyed transects are shown in Figure 22.

18 Elevation [m] Elevation [m] Elevation [m] Figure 21. Ice conditions at DRLL05 on May 15. A section of thermal ice runs between the shore-fast ice in the foreground and the mound of rubble ice towards the open water portion of the channel DRLL Distance along Transect [m] DRLL05a Figure 22. RTK survey transects at sites DRLL05 (top), DRLL05a (middle), and DRLL06 (bottom). Water levels at the time of the ice survey are shown with dotted lines. The shaded point at DRLL05a indicates the elevation at the bottom of a shear crack. As shown, the ice profile typically lies approximately parallel to the channel bank towards the centre of the channel. The points where more than 8-10% of the ice is above the water surface indicates that the bottom of the ice is resting on the ground. It was assumed that the near-shore thickness was roughly representative of the ultimate thickness attained in the middle of the channel. Table 2 summarizes the measured and calculated ice thickness at DRLL05, DRLL05a, and DRLL06. Ice Ground Water Level Distance along Transect [m] DRLL Distance along Transect [m]

19 y y [m] Table 2. Measured and calculated ice thickness at select locations on the lower Dauphin River Ice Thickness [m] Location RTK survey Equation 1 DRLL DRLL05a DRLL The measured thickness was lower than the equilibrium thickness calculated with Equation 1 at DRLL05, while at DRLL05a and DRLL06, the measured thickness was greater. Figure 11 indicates that DRLL05a and DRLL06 both underwent significant shoving events that caused rapid changes to their respective ice thickness, following the progression of the cover past each location. It is possible that the increased discharge and ice momentum during these events caused the actual cover thickness to exceed the theoretical values. Conversely, the water level record at DRLL05 shows that once the cover was able to progress past this location, there was no significant shove event that caused the cover to thicken substantially at this site. In this case, it is possible that the effects of freeze-bonding and surface crust growth caused the actual ice cover to be thinner than the value from Equation 1. A terrestrial laser scanner was also used in conjunction with the RTK unit to obtain three dimensional surfaces of the near-shore ice cover and the channel bank at DRLL05. The scanner works by creating a point cloud of millions of points which can then be resampled and interpolated to a gridded mesh if desired. The surfaces were geo-corrected and subtracted to obtain a spatial estimate of near-shore ice thickness at DRLL05, as shown in Figure 23. March 16 Ice Surface May 16 Ground Surface Elevation [m] z Elevation [m] z Ice Thickness [m] z x x x [m] Figure 23. Terrestrial laser scans at DRLL05. Three dimensional surfaces of the ice and ground elevations are shown on the left. The computed ice thickness is shown on the right. The ground elevation points surveyed with the RTK unit are shown with black squares.

20 The estimates of ice thickness obtained from the terrestrial laser scanner are in good agreement with those obtained from the RTK surveys. The scanner data shows that the approximate ice thickness varies slightly in the streamwise direction, but is pretty consistently in the range of m. These ice thickness values can be used in the future to calibrate numerical ice models of the Dauphin River. The agreement between methods suggests that using the RTK unit is a more economical and efficient solution to estimate ice thickness (the scans of DRLL05 took about 2-3 hours to complete). However, the scanner s ability to capture a high resolution of data points on the ice surface may be used in the future to estimate the relative roughness of the surface, which might be indicative of the under-ice roughness as well. 7. Conclusions The Dauphin River was monitored again during freeze-up Equipment deployed to collect weather data, water temperatures, water levels, and photographs were supplemented by site visits that coincided with the dynamic progression of the ice cover up the steeper, lower reach. Together, the quantitative records and qualitative observations make up a fairly complete dataset that aids in the understanding of ice cover advancement at freeze-up. This dataset can be used to assess the ability of numerical models in simulating freeze-up processes. The major conclusions of this field season are as follows: Using previous years data to approximate the number of degree days of freezing before ice began progressing up the steeper, lower reach was successful. Site visits were timed almost perfectly to coincide with dynamic shoving events, many of which were observed in person. The supply of ice from upstream transitioned from discrete ice floes which were clearly separated by open water areas to smaller, thinner pans that were partially frozen together and covered nearly 100% of the river surface. Ice floe submergence/underturning was not observed in the field. Incoming ice seemed to always ride up the stationary front and remain at the surface. The leading edge was not well defined. The upstream portion of the advancing cover was an actively consolidating zone, where incoming ice transitioned from fast moving pans to a slow moving cover that decelerated towards the stationary front. Parabolic ridges could be seen in this transition zone where ice was riding up on top of slower moving ice downstream. The ice cover progressed upstream through a series of shoving events. As the stationary front passed a given location, the cover would remain stationary for a period of time. Suddenly, the cover would mobilize and begin moving downstream. The stationary front would jump to the most downstream extent of the shove and once again begin to progress upstream. Each shove event sent a surge of water downstream that could be detected in the water level loggers, which took measurements at 10 minute intervals. If the cover at a given location downstream was strong enough, it would simply rise and fall to pass the surge without thickening. Resolution of measurements will be increased in future water level logger deployment (e.g. 1 minute intervals) to better capture the transient propagation of the surge waves. Shove events could be initiated from downstream and cause cover mobilization to cascade upstream, or vice versa. The latter was more commonly observed, especially in the smaller shoving events.

21 Acknowledgements The authors gratefully acknowledge the financial support provided by the Natural Sciences and Engineering Research Council of Canada and Manitoba Hydro. The support of all members of the Scientific and Technical Committee of the NSERC / Manitoba Hydro Industrial Research Chair in River Ice Engineering is appreciated. Field assistance by Nolan Bray was also instrumental. References Andres, D The Effects of Freezing on the Stability of a Juxtaposed Ice Cover. Proceedings of the 10th Workshop on River Ice, Winnipeg, Manitoba, Canada: Committee on River Ice Processes and the Environment. Andres, D., G. Van Der Vinne, B. Johnson, and G. Fonstad Ice Consolidation on the Peace River: Release Patterns and Downstream Surge Characteristics. Proceedings of the 12th Workshop on the Hydraulics of Ice Covered Rivers Edmonton, Alberta, Canada: Committee on River Ice Processes and the Environment. Beltaos, S River Ice Jams: Theory, Case Studies, and Applications. Journal of Hydraulic Engineering. Beltaos, S Theory. In River Ice Jams, edited by Spyros Beltaos, Highlands Ranch, Colorado, USA: Water Resources Publications, LLC. Michel, B CRREL Monograph III-B1a: Winter Regime of Rivers and Lakes. Hanover, New Hampshire, USA. Michel, B Comparison of Field Data with Theories on Ice Cover Progression in Large Rivers. Canadian Journal of Civil Engineering 11: Michel, B Critical Physical Processes in the Numerical Modeling of River Ice. Proceedings of the 6th Workshop on the Hydraulics of Ice Covered Rivers, Ottawa, Ontario, Canada: Committee on River Ice Processes and the Environment. Schaefer, J.A., and R. Ettema IIHR Report No. 296: Experiments on Freeze-Bonding between Ice Blocks in Floating Ice Rubble. Iowa City, USA. Urroz, G. E., and R. Ettema Simple-Shear Box Experiments with Floating Ice Rubble. Cold Regions Science and Technology 14 (2):