Loading...
HomeMy WebLinkAboutAPA2688·oo~oo~B\ a §00&®©@ Susitna Joint Venture Dccument Number d(oB8 Please Return To DOCUMENT CONTROL Notes on I C E J A H S by R. GERARD, Ph.D., P.Eng. Department of Civil Engineering Universlty of Alberta Edmonton, Alberta, Canada T6G 2G7 . ~e9. LT~C \\SO'-\ 1. INTRODUCTION Without doubt the most dremat ie event on a northt~rn river is the format ion of a large Ice jam. Thls can cause water levels that far exceed even the largest summer, ~r open water, flood levels, with obvious consequences for riverside communities and engineering structures. Figure 1 (a) compares: breakup and summer flood levels for Fort Vermilion on the Peace River in Alberta. The location is shown in Figures 1 and 2 of Appendix II. The dominance of the breakup water levels is obvious. The view from the front door of one of the riverside homes in the town during the fourth highest flood is shown in Figure l(b). Bridge superstructures must obviously be placed well above such levels to avoid the problems shown in Figures 2 and 3, and development located to avoid the problems shown in Figures~ and·S. The sudden failure of ice jams can cause high velocity flow and the movement ·down river of large ice floes at high water levels. It is noteworthy that each pier of the bridge recently constructed at Fort Venmilion was designed to resist the full ice load of 7 HN applied at the highest breakup stage'shown in Figure l(a). Ice jams can also cause unusual scour both of the bed and banks, the latter more by the flow of water in unexpected locations rather than the physical abrasion of the ice. Ice jams are therefore an extremely important feature of river engineering in cold regions*. Yet, in comparison with summer floods, their character- istics are poorly known. lee jams can be very local and very brief, yet very damaging. In unpopulated regions they are also unrecorded. These features make it desirable that the mechanics of ice jam format.ion and behaviour be understood because statistical records of breakup water levels are few and, more importantly, unlike summer flood records, those few cannot be transposed to other locations along even the same river. ICE JAM TYPES ANr CHARACTERISTICS Ice jams can be broadly classified on the basis of the season in which they for-m -freeze-up, winter and breakup -and of their type -floating and gro1.1nded. Freeze-u.e.. Jam~ These form when the stream becomes gorged with frazil icep as shown in Figure 6, or when the down-river passage of panca~e ice becorres obstructed and a jam forms. Winter Jams These form when a mid-winter thaw causes breakup. By definition such a breakup does not extend over a long length of the strE:am. The supply of * When defining the geogravhical 1 imits of cold regions it is. well to recall events such as the ice jam in 1899 on the Mississippi River at New Orleans! (Ge rde.l , 1969). 2. ice floes Is there1:ore 1 hnited and the increase in discharge is of short duratIon. These t\1f0 features generally llmt t the magnItude of the water level Increases. ine major significance of such jams Is that they refreeze forming a formidab'le obstruction for the subsequent spring breakup. This is also a danger woth freeze-up jams (for example, see Frankenstein and As sur, 1972). · Breakup Jams Generally these are of most concern and form during the general spring ice run. After Initiation an ice jam·can develop into e floating or grounded ice jam. Floating Jam This type of jam maintains a relatively unobstructed flow of water under its full length, except perhaps for a short section near the toe (down- stream end) of the jam. It seems to be the most common type of jam aNd is sketched in Figure 7(a)~ Grounded (or Dry) Jam In this jam type the ice accumulation extends to the stream bed over a considerable portion of the length of the jam. The jam then behaves much 1 ike a rockfi 11 dam, as shown in Figure 7(b), with the character of the flow being that of flow through porous media. High water levels can therefore be expected. The discussion that follows deals with breakup jams. Such jams will obv 'i ous 1 y depend heav i1 y on the time and manner of breakup. This is briefiy reviewed first. BREAKUP AND ITS PREDICTION First, it is important to realise that there are some rivers in cold regions which rarely, if ever, experience a well-defined ice run. Such streams are generally braided and shallow with large expanses of ice frozen to the bed, such as the Delta River shown in Figure 8 (which is nowhere near a delta). Such streams are very common in N.W. North America. However for streams in which an ice run is a regular feature, the nature of breakup at a given location depends on: ( i ) ( i i) ( i i i) ( i v) (v) snow melt (magnitude and rate of rise of water level); thickness and strength of the ice cover; water level at freeze-up; quantity of ice moving down from u~stream and~ last, but definitely not least; morphology of the river. Breakup can progress upstreAm or downstream depending on the orientation of the river and Its tributaries relative to the spring Isotherms and the occurrene;e of snowmelt and/or spring r•lns.. In many Instances breakup oeeurs first along the central portions of a stream because of the breakup of a maJor tributary. However, no matter in which direction breakup progresses, it is a progression only in a very general sense; there are many local perturbations, these often taking the form of major ice jams. Breakup is instigated by changes in one or both of two features: water level and Ice sheet strength. The Ice can become so weak that a low flow is sufficient to fragment and move the ice out. In this case the Ice run will be minor. At the other extreme the water level and flow can increase sufficiently to float a strong ice sheet free of the bed and banks and to fragment the ice sheet. For a competent ice cover it would seem breakup can only occur in an intermittent fashion, with ice jams forming, however briefly, to build up water levels and release surges. Such a surge will move ahead of the fragmented ice to keep the breakup front moving. As will be discussed later, the celerity of such surges can be very high. From the above discussion it would seem the three most pertinent para- meters governing the moment and manner of breakup at a given location are: (i) ( i i ) ( i i i) the difference in water level from that just after the forma- tion of a stable iee cover during freeze-up, 6H; ice thickness, ti; the numbe·~ of degree days of thaw, S, which provides a measure of the ice strength. That item (i) is relevant is supported l,:.y the graphs shown in Figure 9, taken from Shuliakovskii (1963). Item (iii) is supported by the ather graph $hown. If item (iii) is important there is little doubt that ice thickness should also be a parameter, although it may vary little from year to year at a given site. However, it should be remembered that the natural ice; thickness can be modified by the formation of a freeze-up jam, winter jam or aufeis. Presumably, for a given river morphology, the relation between breakup and these parameters \'IIi 11 have the form: _\~------------....... 3. Unfort~nately no systematic evaluation of such • function has been reported In North Amertc~. About all that can be said at present ts that breakup will not occur until about 30°C days have accumulated and the water level has increased somewhat beyond that at freeze-up. The required increase in water level c·.an be caused by snowmelt (or rain) or by •n ice jam failure upstream. Either of these can occur on the mainstream or an upstream tributary. To give some idea of the way breakup progresses Figure 10 shows a summary of the average br·eakup d.~tes for the major streams in Alberta, Canada. There are several features of interest. As mentioned above, · seve.ral streams brEakup tn their central reaches first, the breakup generaily being triggered by breakup in a major tributary. This role of tributaries in caus!ng breakup on the mainstream can be an important consideration. If the relative discharges of tributary and mainstream are changed (for example, by regulation or diversion} this will change the t.nfluence of the tributary on breakup in the rMinstream, and, consequently, may change the frequency of ice jams at and near the confluence. Also of interest in Figure 10 is the concentration of the isochrones at the \labasca -Peace conflu&nce near Ft. Vermi Hon. This is probably indicative of ice jams at the confluence and suggests that, unlike the Smoky River near Peace River town, breakup on the Wabasca is not strong enough to cause breakup on the Peace River. In addition to being important in the spring, the risk of inducing breakup imposes important constraints on the allowable range of discharges from hydro-plants (Burgi et at~, 1971; Pentland, 1973). Some field observations of br.~akup have been reported (eg. MacKay, 1965; Newbury, 1967; Johnson and Kistner, i967; Nuttall, 1970; Slaughter and Samide~·1971; Sampson, 1S73; McFadden and Collins, 1977); and Hlehe1 and Abdelnour (1975) have done some preliminary studies in the laboratory using simulated ice, but in general much of engineering significance remains to be learned about the common event ca11ed breakup. INITIATION OF ICE JAMS The initiation of ice jams during breakup will probably be a function of the same \'ariables as breakup. Hence ice jams can be expect~d with large ice thickness, heavy snow accumulation and a large and rapid increase in temperature above freezing. On the other hand, ice jams are less likely if there has been little snow or there is a gradual onset of spring. If an ice jam forms its severity will be a function of the rate of rise of water level (and the associated velocity}, the amount of ice travelling with the breakup front, and the nature of the obstacle that initiates the jam. Obviously with all these parameters fixed, the probability of a jam at a particular location will depend on the river 1110rphology. This can at least be roughly analysed. Using a simple analysis Nuttall (1973) has shown that locations of large mean depth, relative of the average for the stream, cause an Increase in concentration of floating Ice and hence constitute an hydraulic obstruction to the passage of lee and Increase the chance of jam initiation. Such locations wtll correspond to bends and narrows. At these locations, the plan form of the stream provides a further impediment to the passage of ice. The headwaters of reservoirs provide other examples and these are indeed a common location of ice jamss Sudden changes in slope f~om steep to flat also seem to be prime ice jam locations. This Is presumably related to the deepening of the flow and the decrease in velocity. Such high ice concentrations can also be caused by physical obstructions such as islands and bars. A very common physical obstruction is the ice sheet on the river, particu18rly if it is thick or more shore -or bottom -fast than normal (eg. because of freeze-up of winter jams, aufeis, or hanging dams). The mainstream ice cover is a frequent cause of ice jams at the mouth of tributaries. The Athabasca River at Fort McMurray, Alberta, (see Figure 10), is an example of a location where both hydraulic (sudden decrease in slope and ~ncrease in depth and width) and physical obstructions (islands, bars, bend, wide ice sheet) exist. Not surprisingly, therefore, it is a location where ice jams form almost annually. · FLOATING ICE JAHS Ice jam initiation has been briefly discussed. After initiation the future characteristics of a breakup jam depend on a series of hydraulic and structural constraints. A general analytical framework for determining the major characteristics· of floating iee jams was established over a decade ago by Pariset and Hausser (1961), Michel (1965) and Pariset, Hausser and Gagnon (1966). s. The analysis reflected that of an earlier investigation of the mechanics of log jams by Kennedy (1958). Some refinements to the analysis were added by Uzuner and Kennedy (1976) but, as pointed out by Beltaos (1978), the essentials remained unchanged. The latter investigator applied the analysis to two natural ice jams on the $rooky and Wapiti Rivers in Alberta with encouraging results. Another successful application is reported by Macdonald and Hopper (1972). Although further confirmation under field conditions is obviously desirable, the approach seems viable. It should therefore be possible to determine reasonable values for the maximum breakup water levels at a site caused by steady floating ice jams, using records of past breakup discharges. The latter are generally both transposable and available. Hydraulic Constraints If the surface velocity is low enough ice floes will simply accumulate against the solid ice cover or obstruction and the accumulation front will move upstream to leave behind an accumulation one layer thick. 6. However, If the velocity exceeds a erltlf:Al value the Ice floe will turn under when It eontaets the obstruction~ and will be entrained by the flow. AgaEn depending on the flow velocity, It may then be deposited under the downstream lee cover or carried on downstream. The deposition ~elocity has yet to be investigated in any detail for freez~·up or breakup conditions. If the latter occurs the ice front cannot move upstream until some event occurs downstream to lower the velocity near the front. If the former occurs, floes will accumulate under the downstream ice cover until the obstructio~ is such that the surface velocity at the front is reduced and the impinging ice floes are not entrained. The front will then begin to move upstream and the process of entra.inment, deposition and surface aceumu:lation repeated. This will result In a steady progression of the front if another condi- tion Is satisfied. It Is argued that there is a tendency fo_r the front of such an accumu- lation to be entrained by the flow as it moves· under the accumulation. This requires a local acceleration of the flow which in turn requires a lower piezometric head downstream. This causes a lowering of the water level just downstream of the f~ont, much like the lowering of the water level In subcritieal flow over a hump In the bed. A simple analysis of the hydraulics of the s~uation along these lJnes suggests that the front will be engulfed when v <. rr ( 1 -!.> {g"t (1 -s. f h I where V is the average approach velocity, h the ~pproach depth, t the accumulation thickness, and Si the relative density of the ice (Pariset, Hausser and Gagnon, 1966). A rearranged version of this relation is compared with experimental and field results in Figure 11. The agreement is noteable. Figure 11 indicates that an ice accumul~tion cannot progress upstream ifF • VII gh > 0.16 and that the dimensionless thickn~ss, t/h, of the front portion of the accumulation must be less than 0.33. Field measurements (Kivisild, 1959) suggest that the critical value of F is. actually about 0.08-o. 10. If the approach velocity is such that F > 0.1 (e.g. about 1.1 m/s for 10m depth) no accumulation Is possible (i.e. any accumulation will be continually engulfed) until a backwater due to some obstruction down- stream r~duces the velocity below the maximum accumulation velocity. If this occurs the accumulation should then progress, leaving behind a thickness that, initially, is close to 1/3 of the flow depth. Such are the hydraulic restraints placed on a floating ice jam. Structural Constraints As an accumulation progresses upstream an increased area of accumulation is exposed to the drag of the flow passing underneath. This accumulated drag must be transferred back to the original obstruction, or to the banks of the stream. To transfer this load, the accumulation must be 7. strong enough to sustain lt. As pointed out by Parlset et al. (1966) the compressive strength of tt'.e accumulation is a direct function of Its thickness. · If the thickness given by the hydro-dynamic constraints Is Insufficient to sustain the load to be transferred, the accumulation will collapse or shove until it Is thick enough. The channel Is considered narrow if the maximum thickness of the accumulation given by the hydraulic constraints is sufficient to sustain the additional load from an advance of the front by shear on the banks. In this case the accumulation thickness is governed by the hydraulic constraints. The channel is considered wide if the accumulation shoves as it lengthens. The shoving increases the maximum thickness until the drag added by ~n advance of the front Is sustained by shear on the bank as shown In Figure 12. When this thickness is reached no add;ttonal 'load Is trans- ferred to the obstruction. Thereafter the maximum accu111.1la.tion thickness remains the same despite a lengthening of the accumulation. The thickness left behind as the accumulation advances is then governed by the strength of the accumulation -that is, by the structural restraint. This maximum thickness has come to be called the equttibrium thickness. The strength of an accumulation of ice and therefore, in a wide river, its thickness depends on parameters lJ, which is related to the porosity and internal friction of the accumulation, and Cj, a cohesion parameter. The maximum accumulation thickness is glven by (Michel, 1965; Pariset et al., 1966; Uzuner and Kennedy, 1976). 2C. 1Jp. (1 -s.)gt 2 -[(gp.S ----8 1 )8] t·-T 8 c 0 I I I where S is the channel slope, B the channel width at the bottom of the accumulation, and T is the shear of the water on the bottom of the accumulation as shown in Figure 13. Similar values of 1J have been determined from field measurements in at least two independent investigations (Pariset et al., 1966; Beltaos, 1978) and these values are not inconsistent with those found in the laboratory (Uzuner and Kennedy, 1976). From these investigations a value for lJ of 1.2 seems reasonable. Littie is known about the cohesion parameter, except that its effect seems to be small in breakup jams. Laboratory tests suggest C. ~ 100 -SOO Pa. I For uniform flow under the accumulation L E pgh.S I in which h; is the distance from the bottom of the ice accumulation to the maximum velocity point. The: ratio hi/hj, where hj is the depth of flow beneath the jam, can be found from given roughnesses by h. I ti:-= J 1 1 + k -l/~ r where k r = k. I ~ In turn, h. can be found from J v. . R .-L • 2.5 ln r• 6.2 v. ..!.. ~ where Vj ~~ . R • h.B . 2 J 1 + k 1/ .. k kb ( r )" • 2 The roughness of the ice cover seems to be related to the thickness of the ice cover (Kennedy, 1958; Nezhikhovskiy, 1964; Tatinclaux and Cheng, 10.'.,0\ --..& ........ e...._ .... -...-... •• 1~ ..._.·----• •&...--~--~& •~-:-.-. &1--.,. TL-• !-1;/{fJ/ IIOIIIU; l'lliiCII;;J\.111 ftUI.IIU ;:ju~~li:;;;J\.j ~Ill~:; ;)I'll:; Ul \.11~ l~li:O I Ju.;;;.. IIIIIOIL 1::1 k. I -. !. I where R.i is a typical floe length. A crude estimate of the form of this function is shown in Figure 14. To calculate the equilibrium accumulation thickness all these relations must be satisfied simultaneously. A suggested procedure is 1. Estimate k. I 2. .Calculate h. J 3. Determine B 4. Calculate h. I s. Calculate "[' 6. Calculate t 7. Calculate water level = h. J + 0.9t A typical calcu1ation ls detailed in Appendix I. Consideration of the above will indicate that the increased water level is caused both by the additional roughness, and the additional thickness of the accumulation! over that of the normal solid ice cover.• On large flat.rivers the former is the more important influence. In smaller steep streams the latter would probably be more important. Because it is based on unifo~m flow calculations the above calculations give 3n esti~ate for the maximum level along a floating i~e jam. However the actual water level will follow a gradually varied ~low profile as sketched in Figure 7(a). An actual example is shown in Figure 7(c). The calculation of such profiles is not considered herein, but are little more complicated than gradua'lly varied flow calculations for normal open channel situations if the downstream boundary condition can be determined. This however is difficult at present. It is important to keep this open channel behaviour of ice covered channels in mind when assessing water levels along such channels. * If the ln expression for velocity is replaced by a power function approximation, steps 4-7 collapse into the evaluation of the single expression given in Appendix 1. If, further, B varies 1 ittle with h., over the values of h. of interest, this would include steps 2 and 3Jtoo. The siple relation f~r maximum ice jam stage that then results is given in Appendix 1 and shown in Figure 15. 9. In all the above Investigations the poss1bl11ty of chann~l i~d changes have not been considered. Yet such changes could have an lmportlDnt Influence on the behaviour of the Jam and the water levels It causes, and presumably on such engineering structures as burled pfpel ines, bridge piers or spur dykes that 1te on or under the bed. A field observation of such scour is dis- cussed below. A first attempt to calculate scour under a quasi-steady floating ice jam has been reported by Mercer and Cooper (1977). Stability of Floating lee Jams If the situation that prevailed at formation changes, the accumula'tion configuration may change. For example If the discharge increases the accumulation can be expected to shove and thicken. However, if the dis- charge is reduced little should change, other than the water levels. Andres (1980) took advantage of this in their analysi5 of the 1978 ice jam at Fort McMurray. Likewise if the jam is thickened by the deposition of ice entrained upstream it should simply increase the upstream water levels. On the other hand, if the accumulation begins to melt it can become thin enough to be unstable and shove again. GROUNDED ICE JAMS These jams can he caused, for example, by the coYlaps~ of a floating ice jam, the sudden stoppage of a surge of ice and wat~r, or by blockage of the flow under a hanging dam. Given the limited and irregular depths of most natural channels, the formation of such jams is an obvious possibility. The destructive jams described by Barnes (1928) and Frankenstein and Assur (1972), on the Allegheny and Israel Rivers respectively, were known to be grounded. The description of the Moira River ice jams at Belleville (Lathem, 1974) suggests that they only became threatening when the passage under the ice was blocked -that "is, when the jam primed. Mathieu and Michel (1967) found that if the ratio of the flow depth beneath a floating jam was less than the largest dimension of the entrained floes, the jam would •prime' and bf!come a grounded jam. As stated by Michel (1971) 11 in such jams the headlosses are considerable compared to those of a simple [floating] jam. It has been impossible to determine these losses in a general manner because of the seemingly fQrtuitous length of grounding in ~ach ease and the variable solidarity of the accumulation of the floes 11 • This states the problem succinct1y. However, given the possibility that such jams may be responsible for the highest breakup water levels, much further work is required on this type of jam, if only to establish a reasonable upper limit on the high water levels possible. 10. ICE JAM FORMATION AND FAILURE: THE UNSTEADY CASE Almost all past work has been concerned with almost-steady flow past an· tee jam. However an examination of reports recordeo in archives and told by eye-witnesses reveals Important features of observed ic2 jams that are difficult to explain from steady flow considerations. A minor but typieal example is provided by Johnson and Kist ner (1967). During breakup of the Me:!d~ Rt~r tm th~ north slooe of Alaska "a flow of brownish river water about 40 em in height was-progressing over th-e topof tile river. ice (June 7) ••• at the pace of a fast walk, perhaps~ km/h. A floe [sic] of jumbled ice blocks choked the channel behind the slush wave. This ice flow [sic] at times overflowed the unbroken ice or simply created ice blocks as it advanced. The advancing ice flow [sic] with its slurry of water and ice blocks jammed quite suddenly when it reaehed a narrowing of the channel 0.5 km below camp. The river, now completely choked with jumbled ice blocks, rose rapidly, about 2m in 1.5 hours •••• On the afternoon of the 10th a very high water level allowed the ice jam to slip downstream ••• evidently a similar ice jam had broken upstream ••• this time considerable ice arrived from upstream and the river was choked with i~e blocks for several kilometres upstream [see Figure 16a]. On the night of the 11th the entire ice floe [sic] broke ••.• After the dam [sic] released, the river level dropped briefly on the 11th and again on the 13th leaving both banks lined with vertical cliffs of ice blocks 3-4m high (Figure 16b). A characteristic of the more dramatic reports is the extremely rapid rise in water levels. For example, in the Athabasca River at Fort McMurray in 1875 "in less than an hour the water rose 57 feet, flooding the whole flat and mowing down trees, some 3ft. diameter, like grass •.• " (Moberly and Cameron, 1929); on the Peace River near the Mikkwa River confluence in 1886 "the ice in the Peace River struck during the night and about 2 a.m. the water rose rapidly tn the Red [Mikkwa] River. Two feet more gf rise would have put it over the banks .•• 11 (Hudson's Bay Co. Journal, Red River, 1886); on the Athabasca River 35 km upstream of the House River confluence in 1936 "During the night they [three men] awakened to find ~hree feet of water in the room. Scrambling into some clothes they waded out and untied their horses and tried to find higher ground. The water rose so rapidly that all they could do was to climb a tree. Lee and Cinnamon got a safe one and climbed higher a~s the water rose. They could see Donaldson in difficulties and shouted to him, but he appeared unab 1 e to climb or the sap l i ng wou 1 d not :$u~1port him and he gradua 11 y sank out of sight .•• '' (Athabasca Echo, 27 April 1936; Athabasca, Alberta); on the Red Deer River near Red Deer the water rose 11 min about 3 hours and removed the superstructure of a CNR bridge (Morris, 1976). Such rapid increases can only be explained by the action of surges created by the failure, and perhaps the reformation, of ice jams. That such surges occur is supported by the several reports in the literature of very high velocities. For example Killaly (1887) IJbserved '1the ice [on the Missouri River] in the neighborhood of St. Joseph .•• came down from above with a rush, causing a sudden rise in the river •••• The river 11. foamed and hissed. The whole wate~ay was filled with broken i~e grinding along the bottom, and pitching and tossing on the surface. The water itself was not to be seen, as the mass of broken lc:e, and drift rolled by- forest trees and masses of brush, wreckage of all sorts, whirling around, and forced into the •tr by the upward action of heaving ice. A gorge [jam] .had broken above •••• " Doyle 0977) reports on breakup in 1977 on the Athabasca River at Fort McMurray: uFlood wave estimated to be 5 m high rushes downstream p~st bridge tossing ice blocks Into air as it passes at an estimated velocity of S - 6 m/s". With such behaviour possibly preceding the formation of an lee jam it is difficult to imagine they would take up •h--P~-P1v ~hsPs~+aPie+t~e &RuiesftaA ~ft ARs1ue:Aft e•-AAv ~1ft~~;ftft :~m~-t..lit'liijjo ._.,..,....,._,, 17 ....... .-... ~.11iiC 1ill'-11ilit>..,' .._, ... ,,.-,_.,._.:f"~._ ...,. .. ,._,, ._-IIVJJ.iill\'l!:f ii6~"tii>W ... fJ • ,..,..,. •• ,,:J J .. ,.,,.-. In particular, the lncreas.ed possibility of priming a grounded ice jam when such 'ice surges' are halted to reform a jam is obviou!i. Consideration of the result of a sudden halting of such a surge suggests the answer to another anomaly. The quotation given abi)Ve reports a 17 m increase in water level just after the passage of a surge on the Athabasca River at Fort McMurray in 1875. If this is simply caused by the passage of a surge released by an ice jam failure upstream, this ice jam would have had to be at least double this height -say about 35 m high. Although such an ice jam may be possible in the deep valley of the Athabasca River upstream of Fort McMurray, it is unlikely. However, i1F the consequence of a surge reflection caused by the sudden reformation of the jam downstream of Fort McMurray is considered, a much lower ·initial S\JJrge, and hence a lower upstream ice jam, Is required to explain the increase in water level noted. This line of reasoning, and the analysis of surges created by ice jam formation and failure has been pursued by Henderson and Gerard {1981). This analysis considered the consequences of sudden complete failure and subsequent reformation of ice jams. It confirmed the change in water level downstream of an ice jam immediately after failure cannot be more than half the initial water level difference across the jam. It also showed that extremely high velocities can be expected downstream of such a failure. A field example of high velocities after a partial jam failure has been reported by Gerard (1975). The 2-3m standing waves created by this sudden discharge is shown in Figure 17. Figure 18 shows another example on the Yellowstone River in Montana. Doyle (1977) reports velocities as high as 6 m/s caused within an ice jam as it readjusted within an ice jam as it readjusted. Both Henderson and Gerard (1981) and Beltaos and Krishnappan (1981), the latter using numerical techniques, have investigi'tted the behaviour of the jam documented by Doyle (1977) and report gt~d agreement between prediction and observation. Measurements of the propagation~ surges! bot~ ir1 the upstream and downstream direction~. have been reported by Calkins (1981). Although often of short duration (from minutes to hours) the possibility of unusual scour by such events is obvious; to quote Killaly (1887) again "On the 24th [February] a gorge occurred •••• ihe river hurled itself, with great force, against dyke No. 6, and washed along its face ••• in a few hours the who1e face of the dyke had been undermined; the .channel having scoured out a depth of thirty-four feet [from the account this se~ms to represent about 4 m of scour]. The dyke 'turned over'!" BREAKUP WATER LEVELS As mentioned before, a major incentivt.! for developing an understanding of ice jam behaviour is the need to predict breakup water levels for river engineering design purposes. These are often more Important than water levels c•used by summer, or open water, floods. They should therefore be subject to at least as much scrutiny In a river engineering Investigation. Analytical Estimates $9~ indic•tign Qf wh~t th~~e levels might b~ can be determin~d by analysis. Lower Bound If no Ice jams are expected to form at the location of interest the breakup water level will be closely related to the freeze-up water level. As discussed previously indications are that, for a reasonably competent floating ice cover, breakup will occur when the water level rises about a metre or so abo~e the maximum winter stage. This relation . can be refined for a particular site if some observations on the tlme of breakup are available. After the relationship has been established breakup water levels for · various past years can be estimated from winter discharge records and estimates of the thickness and roughness of the ice cover at the time·of maximum winter stage. A probability analysis can be carried out on these estimates to fix a lower bound on the breakup stage distribution. It should be noted that in many locations these no-ice-jam levels.will be above the 2-5 year summer flood levels. Upper Bound On the assumption that only floating jams can form and that they form downstream c,f the site each year, an upper boun~ for the probab i 1 i ty distribution of breakup water levels can be estimated using discharge records and the analysis of floating i~e jams described above. If no grounded jams form the actual probability distribution should be somewhere between these bounds, depending on the probability of an ice jam forming in the reach each year. Unfortunately, this probability is difficult to determine. The other ~imitation on the above analysis is that jams other than simple floatir,g jams may form 'in or near the rea·ch of interest. As pointed out above, the present understanding of breakup events other than quazi-steady floating jams is very poor. Hence because of these limitations on the current ice jam state-of-the- art, the above deterministic estimates must be supplemented by as much information on actual past breakup water levels as possible. Empirical Estimat!! 12. As noted pre~viously breakup waf:er levels are very site-specific. Therefore tc1 be useful the wat1er 'level records must come from very near the site of interest. Sometimes information is available from residentst whether permanent or itinerant (eg. farmers, trappers}. Other times 13. information can be gleened from archives of a nearby community (newspapers, biographies, ~lntenance records, journals, family photographs, etc.). In some eases a standard hydronetrlc gauge Is Installed In or near the reach, although failure of these lnsta11atlons during breakup is eonnon. If sueh a gauge exists the original chart recordings or field notes must be examined. If an ·tee jam did form the water level changes may be rapid tnd will meke interpretation of the chart difficult. An example is shown in Figure 19. However, more often than not, there are neither Inhabitants nor gauges near the reach of interest. The only available information ts then that which can be deduced from environmental evidence such as trim lines, windrows, and damaged vegetation. Of the latter the most important items are the ice scars left on trees by high ice, an example of which is shown in Figure 20.. The elevationsof these scars provide a lower bound on the higher breakup water levels that have occurred during the life of the trees. If the scars are sampled as shown in Figure 21, and their age determined by tree-ring dating (Sigafoos, 1964; Parker and Lozsa, 1973), an approximate history of past high breakup water levels can be reconstructed. A typical record completed in this way is shown in Figure 22. On the ba~is of this observational data, both historical and environmental, another estimate of the breakup water level probability distribution can be made. A method for carrying out a probability ~nalysis of such unorthodox data is described by Gerard and Karpuk (1979), exerpts of which are included herein as Appendix II. An engineering assessment of the results of the analytical and empirical investigations will allow a compromise probability distribution for breakup water levels to be chosen. This should then be combined with the estimated probability distribution for summer floods to get the required probability distribution for design. Joint Probability Analysis The two types of floods are more or less independent and are not mutually exclusive (ie. both can occur in a given year). Hence the probability of one or both exceeding a given stage in a year, P, is given by where Pb • probabi~ity of a breakup flood exceeding the chosen stage in a year; likewise for ~ummer flood$f This joint probability will obv!ously be higher than either of the other twa. A typical situation is shown in Figure 1 (a). Max!mum 'Probable' Breakup Water Levels As for summer floods it is very useful to have some estimate of the maximum breakup water level that could occur. Like a11 things associated with ice jams, this is difficult to assess. The potential is exemplified 14. by the following description of. an lee jam on the Yukon ~iver (Henry, 1965).: f•Tf1e highest jam causing the greatest depth of flooding, according to re,lable reports, occurred at Auby, Alask•. Ruby Is built on a hillside, one of the few villages situated well above the river. In the spring of 1930 a big Ice jam formed and the water backed up to the porch level of the present Northern Commercial Company store. Bo~ts, tied to the porch, were at.least 35 feet above normal river levels. The river valley Is 12 to 15 miles wide at Ruby and remains abc>ut the same for miles downstream. So the jam extended at least 15 mules across [sic) and rose to a height of 65 feet. No one knew the loeatfon of the blocking Jam down rIver. 11 • With a long well-gounded jam in an entrenched valley the water level is ,presumably limited only by the discharge and the supply of ice from upstream -t~e latter being a constraint that should not be overlooked. However, in a reach with a well-developed flood plain, water will be able to move around the toe if the water level rises above the flood plain. The maximum water level should then be a metre or so above the lowest passage on'the flood plain. This mechanism limited the water 1evel of the 1963 ice jam on the Hcleod River in Alberta shown in Figure 23. (Note that levee construction to provide protection against sunwner .~loods ·co~ld remove this safety valve·.) Although a particular reach may be free of grounded jams it may still be within the backwater from a grounded jam In an entrenched reach d~­ st.ream, or in the path of a surge released by the sudden failure of one upstream., Hence at present ~~ttle more than a qualitative assessment of maximum breakup water levels is possible, but nevertheless such an assessment should be made. 15. REFERENCES Andres, D.O. (1980) The breakup process and the docwnentat I on of the 1 978 Ice jams on the Athabasca River at Ft. McMurray, Proc. Workshop on Hydraulic Resistance of River Ice, Canada Centre for Inland Waters, Burlington, Ontario, Sept., pp.1~3-161. Barnes, H.T. (1928) Ice Engineering, Renouf Publishing Co .. , Montrea~, 36lt p. Be 1 taos, S. (1978) Field Investigations of l~e Jams, IAHR, Lulea, Sweden, pp. 357-371. Beltaos, S. and Krishnappan, B.G. (1981) Surges from t~e jam releases: a case study, Proc. 5th Canadian Hyd~o­ technical ConYerence, Fredericton, May, pp. 663-681. Burgi, P.H., Borland, W.M., Greene, K.J., Hayes, R.B. and Peter, B.J .. (1971) Ice problems in winter operation: recommendations for research, Internal Report, Engineering and Re~earch Centre, Bureau of Reclama- tjcn, Denver, Colorado. Calkins, D. (1981) D.i scuss ion of Henderson and Gerard (1981). Doyle, P.F. (1977) 1977 breakup and subsequent ice jam at Fort McMurray, A~berta Research Council, Transportation and Surface Water Engineering Division, Report SWE/77/01, 25 p. Doyle, P.F. and Andres, D.O. (1979) 1979 Spring breakup and ice jamming on the Athabasca River near Fort McHuray, Alberta Research Council, Transportation and Surface Water Engineering Division, Report SWE/79/05, 32 P'• Frankenstein, G. and Assur, A. (1972) Israel River ice jam, IAHR, Leningrad, U.S.S.R., pp. 153-157. Gerard, R. 0975) . Preliminary observations of spring ice jams in Alberta, IAHR, Hanover, New Hampshire, pp. 261-277. Gerard, R. and Karpuk, E.W. (1979) Probability analysis of historical flood data, ASCE, Journal of Hydraulics Division, V. 105, No. HY9, Sept., pp. 1153-1165 (see Appendix II). Ge rde 1 , R. W. (1969) - Characteristics of the Cold Regions, CRREL, Monograph 1-A, pp. 30-33,. Henderson, F.H~ ~nd Gerard, R. (1981) Flood waves caused by ice jam formation ~nd failure, Proceedings of the IAHR Ice Symposium, Quebec City, ~w •• 1, p. 209. Henry, W.K. (1965) r The Ice Jam floods of the Yukon River, Weatherwlse. v. 18, April, pp. 80-85. Johnson, P.L. and Kistner, F.B. (1967) 16. Breakup of lee, Meade River, Alaskao CRREL, Speci~l Report 118, 12 p. Kennedy, R.J. (1958) Forces Involved in pulpwood holding grounds, Engineering Journal~~ Vol. 41, January, pp. 58-68. Ki 11a1y, H. H. (1887) The works on the River Missouri at St. Joseph, Transactions of the Canadian Society of Civil Engineers, Vol. 1, pp. 48-67. Lathem, K.W. (1974) Ice regime investigations on the Hoira River at Belleville, Ontario, Proceedi~gs of Seminar on the Thermal Regime of River Ice, Published as National Research Council Technical Memorandum No. 114, January 1975, pp. 109-120. MacDonald, E.G. and Hooper, H.R. (1972) Hydraulic model simulation of ice jamming during diversion of the Nelson River, engineering Journal, Vol. 55, No. 10, pp. 42-49. MacKay, D.Ko (1965) Breakup on the MacKenzie River and its Delta, 1964, Geographical Bulletin, Vol. 7, No. 2, pp. 117-128. Mathieu, B. and Michel, B. (1967) Formation of dry ice jams, Proceedings of the 12th Congress of IAHR, Vol. 4, pp. 283-287. McFadden, 1'.T. and Collins, C.H. (1977) Ice breakup on the Chena River, CRREL Report 77-14, 44 p. Mercer, A.G. and Cooper, R.H. (1977) River bed scour related to the growth of a major ice jam, Proceedings 3rd National Hydrotechnical Confer~nce, Canadian Society for Ciyil Engineering, laval University, Quebec City. Hi che 1 , B. ( 1965) Static equilibrium of an ice jam at breakup, Proceedings Eleventh Congress of IAHR, Vol. 5, pp. 37-48. Michel, B. (1971) Winter regime of rivers and lakes, CRIEL ~nograph 111-Bla, 130 p. Michel, B. and Abdelnour, R. (1975) Break-up of a solid river ice cover, IAHR, Hanover, N.H., pp. 253-259. Moberly. H.J. and Cameron, W.B. (1929) When fur was King, J.M. Dent and Sons, ltd., Toronto,. p. 151. 17. Morris, R. 0976) Minutes of ETAC Project Committee on Ice effects on bridges, Winnipeg. Newbury, R. (1967) The Nelson River: A study of subarctic river processes, Ph.D. thesis submitted to Johns Hopkins University, Maryland, U.S.A. Nez h ~ Kuovsk i y, R. A •. (1964) Coefficient of roughness of bottom surface of slush ice cover, Soviet Hydrology, No. 2, pp. 127-150. Nuttall, J.B. 0970) Observations on break-up of river ice in north central Alberta, Canadian Geotechnical Journal, Vol. 7, No. ~. pp. ~57-q63. Nuttall, J .. B. (1973) River modifications and channel improvements, Proceedings of Seminar on Ice Jams in Canada, published as National Research Council Tech- nical Memorandum No. 107, pp. 83-91. Parker, H.l. and lozsa, L.A. (1973) Dendrochronological investigations along the MacKenzie, liard and South Nahanni Rivers, N.W.T., Technical Report No. 10, Glaciology Division, Water Resources Branch, Environment Canada, Ottawa. Pariset, E. and Hausser, R. (1961) Formation and evolution of ice covers on rivers, Transactions of the Engineering Institute of Canada, Vol. 5, No. 1, pp. 40-49. Pariset, E., Hausser, R. and Gagnon, A. (1966) Formation of ice covers and ice jams in rivers, Journal of the Hydraulics Division, ASCE, Vol. 92, No. HV6, pp. 1-24. Pentland, R.S. (1973) Ice formation and jamming on the South Saskatchewan River below lake Diefenbaker, Proceedings of Seminar on Ice Jams in Canada, National Research Council Technical Memorandum No. 107, Ottawa, Ontario, pp. 122-151. Sampson, F. (1973) The ice regime of the Peace River in the vicinity of Portage Mountain development, prior to and during diversion, Proceedings of Seminar on Ice Jams, University of Alberta, Edmonton, published as National Research Council Technical Hemor·andum No. 107, pp. 158-178. Shulyakovski i, l.G. (1963) Manual of forecasting ice formation for rivers and inland lakes, Manual of Hydrological Forecasting No. It, Israel Program for Scien- tifi~ Translations, Jerusaleum, 1966, 245 p. Sigafoos, R. S. (1964) Botanical evidence of floods and floodplain deposition, U.S. Geo- logical Survey, Prof. Paper 485-A, 35 p. Slaughter, C.W. and S~lde, H.R-(1971) Sprlng breakup of the Delta River, Alaska, CRREL Spectal Report 155, 32 p. Tatlnelaux, J.C. and Cheng, S.T. (1978) Characteristics of river lee jams, IAHR, Lulea, Sweden, pp. 461-~75. Troebst, C,.C. (1963) The art of survival, Translated by Oliver Coburn, Doubleday & Co., New Yot·k, 312 p. Uzuner, H.S. and Kennedy, J.F. (1976) Theoretical model of river ice jams, Journal of Hydraulics Division, ASCE, Vol. 102, No. HY9, pp. 1365-1383. Water Survey of Canada (1974) lee thickness and break-up data for selected r1vers in Alberta, CalQarv Branch. Water Survey of Canada, Inland Waters Directorate,