THE MOVEMENT OF POLLUTED SEA WATER ,. NEAR MARMORILIK, GREENLAND; ITS CAUSES & POSSIBLE CURES ..•..... THE MOVEMENT OF POLLUTED SEA WATER NEAR MARHORILIK , GREENLAND; ITS CAUSES & POSSIBLE CURES E. L. LEWIS Frozen Sea Research Group Institute of Ocear. Sciences , Patricia Bay 9860 r1est Saanich Road Sidney, B. C., Canada vaL 482 ttAY 1978 -~ .. CONTENTS 1. Introduction and summary of recommendations 3 2. Oceanographic data from the·Marmorilik district 7 3. Circulation in arctic fjords 14 4. Hethods of retaining polluted waters within Agfardlikavsa fjord and their probable consequences 23 5. A floating boom dam 28 6. Suggested programme of environmental observations 31 ) Appendix - Hydrodynamic effects of a break in the floating dam D.M. Farmer 33 References 38 2 FIGURES 1. General view of the location 9£ the Black Angel Mine giving the positio~ pf measurement stations. 4 2. Bathymetry in the vicinity of the Black Angel Mine. 5 3. Water depth profiles along Afgardlikavsa and Qaurnarujuk fords. 6 4. Mean summer and winter profiles of tell'perature, salinity, and 9 density from Site 4 in A fjord. 5. Mean SUl'IUIer and winter profiles of tenperature, salinity, and density from Site 10 in Q fjord. 10 6. Temperature and salinity variations in moving from the head of A fjord out over the sill and seaward in Q fjord. 12 7. A schematic of wind set up. 15 8. The typical structure of sea ice. 17 9. Increase of mixed layer depth due to salt rejection by sea ice 19 during growth. ) 10. Dissolved lead concentrations in A fjord. 21 11. Resonance periods for internal \o'9-ves on the interface between seawater and a fresh water layer trapped behind a dam. 26 12. Schematic of an ice boom floating darn for fresh water retention in A fjord. 29 3 l~ Introduction and Summary of Recommendations Tailings dumped into Agfardlikavsa fjord from the Black Angel Mine at Hanrorilik, in Greenland, are to some extent soluble in sea water and high concentrations of dissolved heavy metals exist in the fjord basin. During winter months, vertical cO,nvective circulation caused by the rejection of salts by sea ice during growth mixes these high concentrations up',;ards into the surface waters which are then moved out into the neighbouring Qaumarujuk fjord by tidal exchange. Figures 1, 2 and 3 show respectively a plan of the general area defining measurement stations referred to in the text, the loca~ bathymetry ) and approximate depth sections along lines defined by 1-4-10 and 10-12-16 on Figure 1. The oceanographic and meteorological data available from the area is fragmentary and satre of it is of questionable accuracy. A selection has been mace and the fjord studied in terms of the basic dynamical. processes common to arctic fjords in as far "as the data would allO'.o1. This has proviced a base for evaluating possible solutions to the dissolved heavy metal transport problem but has sharply defined the need for furtiter i:lvestigations. "In spite of the lack of data it has proven poss~le to make a clear suggestion of the optimum method of pollution control though quanti tati ve detail awaits acquisition of better data from the fjord, particularly tiT:!e series records. As a ~ethod for permanent control of the pollution it is recOr.h~~ded that the discharge point for tailing disposal be moved to near t1te bottom of the fjord, that the density of the tailing t-laters be increased to be significantly above that existing within the fjord at that depth at present. This should be done by cooling the \.,rater prior to discharge and increasing its salt content either by removal of water by distillation to be used for domestic purposes and discharged separately or by the addition of rock salt. The optimUJ salinity for discharge cannot be computed on the basis of the present data but the matter is so important that it is suggested that an increase to 40 0 / 00 salt should be instituted as soon as possible as a stopgap measure. The full effect of this introduction of more dense tailing 'tolaterS will not hecorr.e apparent for a number of years but should still minimise -the spread of heavy "rretals even after the mine has closed. ~"1hile the mine is still operating i t "is recommended that a fresh water surface layer be retained in Afgardlikavsa fjord by reeans of a floating darn across the sill. This recommendation is subject to the COl:!!Ir.ent of a competent engineer with experience in ice boom construction on the feasihili ty of such a structure. ) It is furthe~ recommended that a program of environmental measurements be instituted imrrediately enabling proper quantitative design criteria to be applied to the above recommendations for fresh water retention and tailing water density increase. Approximately $35,000 \'lorth of instrurr.ents would be used for at least one year. ) ,' -,, o 1 L ! 2 km FIGURE 1 General view of the location of the Black Angel mine giving the positions of measurement stations. I o o ,oo~ Q -/ ( FIGURE 2 DEPTH IN METERS o I 1 • km 2 I Bathymetry in the vicinity of the Black Angel inine. ~STATION (y 'NUMBER 4 ~_ _ _ _+,1_ _ _ _-t~2_ _ _ _-t3L+_':""'-i4~_ _ _-i5 km 25 50 75 , 0" FIGURE 3 Water depth profiles along Agfardlikavsa and Qaurnarujuk fjords. 100 The location of the station numbers is shown in Figure 1. 125 150 175 10 2 3 12 STATION 4 NUMBER 5 G @ 7 8 9 25 50 .. ::; '. 75 , 100 ~ .< 125 '- ," 150 175 I I 200 225 I I ) 2. Oceanographic Data from the" l>Jarmorilik District The data available derives from two sources; that tabulated in the reports issued by the Geological survey of Greenlnnd under the general heading Recipient unders¢gelse Agfardlikavsa and Qaumarujuk Fjords (hereinafter A ~ Q Fiords) from 1972 through until the latest report covering 1976-77, and from casual observations made by staff collecting data for these reports. TWo thousand tons per day of tailings in 6,000 m3 of salt tailing water of varying salin"ity and of temperature about 16°C are pumped through an outlet into A fjord at a depth of 25 rn where the water is 50 m deep. Sea water for use with the tailings is extracted in the region of the sill at a depth of a few meters. The tailings are fairly evenly distributed and except when air bubbles are inadvertently discharged with the tailing waters no turbidity has been noticed on the surface of the fjord at any time of year. It is remarkable that no suspended tailing materials have been noticed outside A fjord at any time, and increases in the lead and zinc concentrations in the bottom sediments of Q fjord have" been attributed to chemical processes causing deposition of dissolved material. Measurements were made in one year when there was a labour strike in the mine but no "changes in pollution levels were noticeable. Run off from the nearby ice cap has been estimated as 2 m3 per second during September, a figure which would appear to be conservative. Flows of this type normally show a strong diurnal variation, a function of the incident solar radiation, and a discharge peak in Mayor June of the year at the time of surface snow melt from the land. Probably 90t of the run off occurs in the period June - october inclusive. It has been said that half the total volume of A fjord is avail~le as discharge fresh water each year. A thumb nail calculation shows that if this flowed evenly into the fjord over 5 months, about 4 m3 per second of fresh water would be required. Tides are mixed diurnal and semi-diurnal with a range of about 180 cm (spring) and 100 cm (neap). No measurements have been made of tidal currents over the sill of A fjord but if one assumes that tidal currents are uniform in the water mass a simple calculation of the volume exchange required in the tidal period would indicate that velocities were below 3 crn/sec. It must be emphasized that this does not mean that velocities over the sill are beneath that value but that those that can be ascribed to tidal forces acting alone on a homogeneous water mass are. The fjord freezes up around Christmas time and ~ m of sea ice grows by the following April. One of the most remarkable features of the data available from the reports is the difference in the fall and winter profiles of zinc and lead concentrations shown in Table 1, taken from the 76/77 report. It is seen that during the summer period heavy metal concentrations are at a low level outside A fjord and that within the fjord high concentrations are only seen below the SeptelI'be!:' 1976 Station Narch 1977 ph No. Depth, m 1 0 10 20 30 SO 53 3.5 4.2 8.0 149 682 708 118 521 643 4 0 10 20 30 50 68 5.5 4.8 9.9 125 884 942 10 0 10 20 30 50 75 100 112 0 10 20 30 SO 75 100 150 178 17.2 31 18.5 8.6 0 10 20 30 50 75 100 150 200 5.7 2.8 0.9 1.2 1.7 6.7 11.0 5.1 5.4 12 ) 16 TABLE 1: Zn 2.1 3.5 Zn 183 151 153 142 179 177 141 3.6 3.9 5.2 93 631 635 197 165 137 123 108 -119 169 138 138 110 112 112 7.3 4.2 2.9 9.7 8.0 20.9 28.0 26.2 1.3 2.3 1-6 4.4 5.6 12.3 16.1 150 132 16.3 3.6 1.5 5.7 7.4 ~.8 195 145 194 13 .0 2.5 2.6 2.6 2.2 2.1 127 89 56 1.7 1-5 2.0 0.9 0.5 8.0 2.6 2.9 4.6 5.7 12.3 22.6 12.3 1-7 150 125 42 2.5 3.9 1.3 2.6 1.4 3.1 124 90 30 0.7 1.2 0.5 1.3 0.6 1-0 5.0 3.3 1.6 2.0 2.6 6.5 9.0 5.5 3.1 79 44 15 5.7 1.5 1.0 0.9 0.3 41 23 9 1.8 1.6 0.6 0.2:' - 0.3 1.6 "73 at.. -._-the- -various -Dissolved - - - - - . - _zfnc . _ - - and - - - -lead - - - -co~centrations - - - - - - - - -_._- _._-- .. _-.-. stations in Ph ~b~ 0.3 9 SITE 4 TEMP. ,"C ...•........ 1 -+I ______+-____ I -2~____ -_'~____ 0 I I SALINITY 30 0/00 2 -+~ I 32 25 24 33 I srGMA-T 23 ~4 I ----- 31 I I ____3+-____ I -~-.-.- 26 27 28 29 Ot-----~--~~--~----~----~--~ -' _. _. _. -.-: -:::."7:.-:-,,' ....... .;.-:"....... - - -:.:...::': .:..::.::...: ....: ~-..:. .. ' . . . . \ -- ..... ' , I ,c 40 . >- nrm !1. 0 \ I \ I I i :z: w '\ SUMMER rrrm rJm 80 ) o T I I I I I 40 E I ...:z:a. I ll.I o I rrhrr 80 WINTER 120 FIGURE 4 l1ean summer and winter profiles of temperature,. salinity and density from Site 4 in A fjord. as (p Ip -1) 10 3 where P Sigma-T is defined is the density of the sea''''ater T T' max at temperature T and p the maximum density of fresh water (near 4 D C) . max . a 1 2 TEMP. "C. . .......... 1 4 3 2 I SALINITY 32 31 D/ OO - - - --- 34 33 I I SIGMA -T .-._.-._. 23 24 a 25 26 - - --=-.=.!..-:...:..-::: '--..:, -=';:;: E ..... 40 w 0 ...., ---- .. ., I \ \ , '. ...... ', " , " ~ \ . 80 29 ' ~ :. :J: 2& :....-..-. -:-:,.... • .:...:.:... . . . . . . . . . . . . . . . . . . . . . . . . . . ............. .... 27 \ . .,! SUMMER ,, , 1 1 rr/m min m'm ,, \ . 120 ,, T 1 1 I 40 E .. " \ .. \ . .. , :J: .... w 0 \ \ 80 WINTER FIGURE 5 ) \ \ \ .. . . 120 \ ,, , , \ i 1 \ I firm rrtm rdm Mean summer and winter profiles of temperature, salinity and density from site 10 in Q fjord. Sigma-T is defined as (p Ip -1) 10 3 where P is the density of the seawater T . T' max at temperature T and p the maximum density of fresh max water (near 4 D C) . ) 11 depth of tailing injection. In the winter of 1977 the concentrations are more or less even from top to bottom and it is clear that material is being moved out over the 5111 (depth 21 m) to contaminate the upper waters of Q fjord. The higher levels shown in the deeper waters of station 10, in September 1976, are present at Station 12 but not at Station 16 (Figure 1). They appear to be remnants of the values at that station found in March, 1976, circulation during the summer of 1976 having reduced concentrations above SO m to those shown in the table. Clearly, there is a difference in water movement between summer and winter and to interpret this the 'density profiles at the various stations are required, calculated from measurements of terrperature and salinity. Figures 4 & 5 show these parameters as a composite of information acquired from all the reports_ A selection has had to be ~ade on a subjective basis as some of the measurements would appear to be erroneous. For example, on occasion, more dense water apparently overlies less dense water, a situation that can only occur for a short period under ext~emely dynamic conditions_ Data collected through the sea ice in winter ~particularly subject to observational error. A common problem is due to \-later freezing in the oceanographic bottle before samples are withdrawn. As the ice formed is almost pure the salinity of the remaining water is enhanced giving a false reading \-lhich, combined with temperature, giVes densities that are too high. More puzzling is the "warm winter" as indicated by the ~Brch 77 readings in A fjord. Waters beneath growing sea ice must usually be very close to their freezing point due to the vertical circulation induced by the salt rejected by the sea ice during growth but these data show waters nearly l~C above freezing point. If this is not an instrumental error it ~~uld appear to be possible only as a highly transient condition that might result from a sudden exchange of the waters in A fjord with those outside, a point that will be discussed later. As all of the other winters shOl" values close to freezing we will assume that freezing is the "normal" state, there being insufficient information to make any other comment on this curiosity. Summer data Sho~ no density inversions but a normal spread of values from year to year and indeed from day to day as the flow of fresh water into A fjord causes changes in the surface layers or strong winds mix in this fresh water downwards. Occasionally strong winds are sufficient to remove the fresh water layer completely; the September '74 data show4fa surface layer in A fjord of about 3J.~%o salinity at a temperature of about 2.7°C. Figure 6 shows salinity/temperature/density data out to Station 16. There is an increase in the surface salinity as one progresses a\'Iay from A fjord and the development of a distinctive mixed iayer down to a depth of about 20 m, almost certainly a result of the wind blowing over an increasing fetch. Helicopter operations from Narmorilik have produced an intermittent record of wind velocities and air temperatures in the latter half of 1977. It is clear that little significant cooling of the water takes place before October 20th when air temperatures, for the first time, drop below freezing. There~ a rise in air temperature ',.. 26 'c. -2 28 0 30 2 32 4 ' -2 r""=::::::-;/,> 0 2. 26 -2 4 1---1..---'-- 28 30 2 32 4 1=-=':'=:-:--:"'" St 10 SI 1 -20 2 26 4-2 1----'----'--7'> St 12 28 30 2 o 1-.L..-~?7 St 1(; E ....Q.I W " FIGURE 6 Temperature "and salinity variations in moving from the head of A fjord out over the sill and seaward in Q fjord. m " 32 4 Station positions defined in Figure 1. 13 St about November 15th, 1977 which remained moderate until the end of the month and then progressively got colder until December the 16th when a combination of calm and increased cold probably indicated freeze-up. Studies of water structure beneath sea ice depend greatly on knowing the initial conditions just before ice formation and for the present purpose we are only able to make an intelligent guess based on the September data and some knowledge of the rate of heat extraction. The wind records indicate that most of the winds were along the axes of A or Q fjords which is to be expected ) from a local topography and the location of the heliport. The stronger winds were towards the mouths of A and Q fjords and storms had typical long term values around 20 knots. A maximum of 37 knots was recorded on December 8th. As observations were taken only during helicopter operations it must be emphasized that the stronger winds may not have been recorded and that there may be a bias in the data because it is being taken at the same time of day rather than regularly throughout the 24 hrs. It has also been said that the winds in the center of the fjord are much stronger than at the helicopter site. Thus, the data should be taken as indication of a meteorological situation rather than a quantitative description. Dissolved oxygen values of around 9 mg/l in A fjord independent of depth in summer showed that the waters were not stagnant. As a whole the data gives "snapshots" of conditions in A and Q fjords without giving clues to the circulatory processes involved in water exchange. Of the very many relationships between energy input and water movement possible in the typical fjord, only a fet" are present usually and it is necessary to obtain indication of dominant factors from time series measurements before an adequate analysis of the system can be carried out. No such clues are available from the }~rrnorilik data and much of it is of questionable precision. Under these circumstances it is simply not worthwhile to apply sophisticated methods of data analysis; the best that can he done is to indicate probable processes involved in water movement and to see how these serve to explain the observed water structure. An enhanced observational program is required before the pollutant transport problem can be properly described and suggestions along these lines are the subject of Chapter 6. 14 3. Circulation in Arctic Fjords The outstanding feature of Arctic fjords compared to those at a more southerly latitude ~s the absence of significant run off from the land and the isolation of the waters of the fjord from wind mixing for a significant part of the year. The paper "Oceanography of an Arctic Bay" by Gade et al (1974) gives a good description of seasonal changes and mechanisms to be expected in a fjord at 700N which probably may be applied to the Q-A fjord system with a suitable change in dates for- freeze-up and melt. Some of the unusual features occuring during the winter under the ice cover are the subject of a paper by Perkin and Lewis "Mixing in an Arctic Fjord" (1978). Energy inputs to Arctic fjords occur from the wind, from the tides, from the melting of floating ice masses, from the fresh water input, and from the salt rejection by sea ice during growth. Floating ice masses are absent from A fjord but are present in Q fjord. To understand the circulation of A fjord)wind mixing and tides should be considered during the summer and tides and the salt rejection by sea ice during growth during the winter. A further possibility is exchange of waters over the sill of A fjord due to the density differences between waters in A and Q fjords. Such a density difference may be caused by a reduction in salinity of the lower layers of A fjord due to upward diffusion of salt as the fresh water run off proceeds seaward near the surface. Winds blowinq down the fiord transfer part.of the~r energy to and by shear forces cause a mixed layer to develop at the s'urface which increases :in depth as the wind fetch increase"s:- i'1hen fully developed this produces a two layer system, a near homogeneous surface layer above a layer with considerable density gradients. Assuming a uniform density gradient near the surface in early summer and following Pollard, Rhines and Thompson (1973) gives a wind mixed depth of 6 m at the sill for 20 knot winds over A fjord which is in good accord \yith observation. Eventually the density differences between these layers prohibits further mixing due to simple shear proc~spses aI:ld the wind.. _tnav_be .thQught~::tp_ '~s.et uP" the surface laver against a coastline so that it is deeper at the downwind termination and the interface between the two layers tilts upwards in the upwind di.rection~ Such "set-ups·' are accompartied"by-a,--s.urface current in: the wind direction balanced by a return flow beneath. These processes are indicated diagramatically in Figure 7 - a & b. Following Hellstrom (1941) a difference in height of about 3.8 m is calculated between the position of the interface at the end of A fjord and at the sill for a 20 knot wind. An extreme case of the wind "set up" phenomenon is shown in Figure ?-c where the constant density line terminat~in the surface under continuous high wind. Under these conditions very extensive mixing can occur, probably ~o the fjord bottom. On one occasion an observer measuring dissolved metal concentrations in A fjord on a very windy day observed high concentrations in the surface waters moving out into Q fjord and we presume that on that occasion some such happening occurred but the general form of the summer heavy metal concentrations as given in Table 1 indicate that this is not common. ~e. water 15 CALM ( a) <E'<_W:...c..1 ND (b) « WIND (c) FIGURE 7 A schematic of wind set up. Extreme winds (c) can remove the fresh surface layer entirely from part of the basin and cause strong vertical mixing. 16 Wind/water interactions can also give rise to transient phenomena that can be of great importance in considering vertical pollutant transport. For example the sudden cessation of a steady wind responsible for set-up often res.ults in abrupt changes in surface layer thickness being propagated down the fjord. Such wave fronts may be reflected and/or amplified by local topographic features. Whether or not such phenomena play an important part in water circulation in A fjord awaits analysis of time series data. Winds are rarely steady and a record taken over a long period will frequently show energy peaks at certain frequencies. A two layer system in the £j ord \'li11 allow internal waves to exist on the interface whose velocity is a function of the layer depths and the difference in density between them. Waves are partially reflected from the coast and dimensions between coastlines, across fjords, etc. define lengths, which in turn define resonant frequencies for the given interfacial wave velocity. With the usual complexity of headlands, embayments, etc. frequently many such resonant frequencies exist and if the wind spectrum contains energies at these frequencies large scale transfers from wind to fjord are possible. For example, with summer conditions and a 6 m deep surface layer in A fjord, as defined in Figure 4, one would anticipate an interfacial velocity of 39 cm/sec and a resonant period bebqeen the fjord head and sill !l/4 wave lenqth because the interface lies abov_e sill depthl of about 11.4 hours. An energy input near this period would be available from the tides; other possible resonances in the system might take energy from the winds. If we suppose that such energies are --- ~" '"'~ - -" aval..lable the internal vlaves may cause signl..ficant additional mixing betWeen water layers by partially breaking as they are reflected on the shores of the system and by direct shear mixing, typically just below the layer interface. The discharge of 6,000 cubic rn per day of water at 16 DC leaves another unsolved problem although as has been noted, no surface concentration of suspended tailings has been noted in the vicinity of the tailing discharge outlet. A calculation shows that this amount of heat entering the fjord would be sufficient to raise its temperature by .1DC in about a hundred days if the loss to the surroundings were minor. Data for winter water temperatures in A fjord often seem to be about .1DC above freezing point; this could be the cause. It should be noted that close to freezing point temperature only has a very minor effect upon density which controls vertical movement in the water column. In summer it would have a larger effect and steps should be taken either by control of salinity or temperature to ensure that these tailing waters cannot_cause vertical circulation. As the fresher water moves seaward in summer more salty waters are entrained into the lower part of this fresh layer and taken seaward so that the density of the basin water in A fjord is continually being reduced. Conditions outside the sill may be ~fected by a variety of conditions which are not consequent upon those within the sill to produce a situation where the sill separates two water masses of different density below a uniform-layer above -sill depth. This density structure is unstable under any perturbation, a situation that pertains to all fjords and causes exchanges of basin waters at AIR -POLyCi1'is-iALLINC AND-TRANSITION REGIONS COLUMNAR CRYSTALS l - - - I C E PLATeS WITH BRINE INClUSIONS AS ~SANDWJCH FILLING- '. FIGURE 8 ~he typical structure of sea ice. See text for discussion. 18 intervals. Fjords have been observed with annual exchange, exchange every three years, every thirty years, etc, etc., depending on conditions. Gade (1973) has given a formula for basin water renewal based on the arumal salinity red,uction and the variation characteristic of the water density profile outside the sill but this cannot be applied to Q & A fjords due to lack of data. In the case of a shallow fjord such as A fjord, it might be thought (there being no firm experimental evidence whatsoever) that such an exchange might occur every few years. Whether or not this can explain the "warm winter" of 1977 cannot be said; we may be looking at a measurement error only and whether this additional circulation is of importance in terms of pollutant transport will depend on the velocity with which the incoming exchange water rolls down the inside face of the sill and the depth to which it penetrates. It appears certain that the most important form of circulation with regard to the winter transport of dissolved pollutants into Q fjord is that associated with salt rejection by sea ice during growth. This occurs because ice does not form solid solutions except with a very few exotic materials and that all the constituents of sea water are unifor.mly rejected by the ice crystal lattice. As this process is of importance in tailing ponds in Arctic regions as well in the present context it is tol0rth discussing it in some detail. Although the salts are rejected by the growing ice inclusions of brine get trapped between ice platelets and this is illustrated in Figure 8. A poly crystalline region of small ice crystals with random orientation is above a region of columnar crystals where the axis of symmetry of large crystals lies very close to the horizontal. The "skeletal layer" illustrated is the lotl1er extremity of these crystals and the entrapped brine exists between small plates protrUding down into the underlying water. As the sea ice grows the temperature around each brine inclusion drops as the freezing interface leaves it behind. Further freezing must then occur within each inclusion for the brine to stay in thermodynamic equilibrium with a solid phase surrounding it and the expansion associated with this freezing causes cracking. When very large numbers of such inclusions are discussed the overall effect is the production of brine drainage channels which look like vertical river estuaries complete with tributaries extending up into the sea ice to maybe two thirds of its depth. Brine drainage occurs both at the freezing interface and from these channels and the potential energy so released into the water column is the driving force for vertical convective circulation. It should be noted that ice will clear almost any impurity from the water and that on an average about 10% of the initial pollutant or salt concentration remains locked tl1i thin the inclusions in the ice during the winter. Once the snow melts from the sea ice surface in early summer the brine drainage channels act as foci'- for internal melting within the ice sheet. The walls of the channels'can melt at a lower temperature than can the surrounding ice due to the presence of salt and concentrated brine. The channels are also disrupt'ions in the crystalographic pattern of the ice sheet so the sun's incident , radiation is scattered and absorbed. Once the dense brine trapped in the channels is released by melting it moves downwards under the influence of DAYS AFTER INITIAL ICE FORMATION o 15 30 45 60 75 FIGURE 9 Increase of ~xed layer depth due to salt rejection by sea ice during growth. Assumptions: - • salinity of ice is 3 D / 10 DO - constant ice growth of 1 ern/day - one dimensional model ignoring sloping sides to fjord or bottom irregularities 20 - E - - - - ~"'- -SILL DEPTH initial profile has a 6 m thick wind mixed layer of 31.7%0 water at freezing point. The remainder of the profile is taken from the summer data ,(Figure. 4) . 30 -0---0--INITIAL - no advective exchanges PROFILE I - heat from the water column has negligible effect on the ice growth rate. I- 40 D... W o 50 GO • 70}---------------~--------------+_----------~~~---25.5 26.0 26.5 27.0 . SIGMA T ) 2Q gravity into regions of higher temperature en~ling it to cause further melting of the surrounding ice. While still remaining above its freezing point the di~uted brine will continue to descend causing further melting until it runs right out of the ice sheet into the underlying water. The salinity of this draining brine has h,een measured at over twice that of the water from which the sea ice .,.las originally formed and thus it penetrates to a-considerable depth below the now disintegrating ice sheet. For the immediate purpose freezing during the winter and the circulation associated with it is considered. This problem was first posed and solved by Zubov (1943) and experience indicates that his calculations are quite adequate in the context of the present data. Based an an educated guess of the density profile at the time that freeze-up commenced (around mid December) Figure 9 shows the increase of the mixed layer depth due to salt convection as a function of time assuming that 1 m of ice grolY'th occured evenly during the period Jan 1 to March 31, 1978. It is seen that convective mixing occurs right to the fjord bottom. On the basis of the existing observations it would appear certain that significant vertical transport of dissolved pollutants to depths where they become available for movement out over the sill by tidal circulation results from saline convection. The_only practical way to prevent this which carnes to mind is to arrange for retention of a fresh water layer from the summer of sufficient depth to allrnY' the ice to grow with little salt rejection and to confine the resulting circulation to trivial)depths by the very large vertical density change at the fresh salt/water boundary. Before considering various ~ethods of keeping a fresh water layer on the surface of A fjord throughout the year it is necessary to realize that one important consequence of this action cannot be predicted with any accuracy, the rate of change of the dissolved heavy metal concentration depth profile with time, when the annual winter outflow of heavy metals into Q fjord has been stopped. Taking the concentration of dissolved lead as an example, Figure 10 shows the profile measured in September 1976 at Station 4 in A fjord together with that from the following March. The more or less uniform March profile is about 140 ppb less than wofild have been obtained by mixing the concentrations from the September profile to produce uniform distribution of lead, and the real export of lead must have been in excess of this as further material must have dissolved into the water from the tailings during the September to March interval. The exi~ting data sh0wf the ~~~~ __ .~~eady st~te Sept~~.t: _profile re,sulting from an annual winter extraction but, as we have nO knowledge of the . rate at which lead will e.nter sO,lution fr'?gt the tai~~ngs when there is no annual extraction, it is irl!Possible to compute the shape of a ne\Y' September curve. This is illustrated schematically on Figure 10 showing that an enhanced lead concentration is anticipated just above the tailings and that the profile shape will change vlith time. We do not know whether suc,!;,.a change l-Till occur in years or decades. It is clear that the rate ~ which lead enters solution depends on the concentration of dissolved lead immediately adjacent to the tailings and any water movement in the area tending to reduce the local lead ,. LEAD; . parts in ~ \J1 o 109 \J1 o '"o o o o , : It, ,, 10 20 ® : MAR.1977 , -'-'-'-'-'-' -'-'-' ~ "'-- ------ -'-'-'-'-'-'-'SEPT. 1976 ---- --------_"'--.., '-:': -·-·-·_· __ .C-....A).-..... ... ® ~ .~ -~ 30· ;---. ~ E 40· r 50 : ~ -...... , " --~,~, "~~, '''\, , ~ L Ll 60 . \ i .I ) I I I • ~! 70 80' PIGtmE 10 Dissolved lead concentrations in A fjord. The curve labelled I'A" is speculative and is intended to show what might happen if the annual winter transport into Q fjord ceased. ) 22 concentration Nil! encourage further flow of lead from the tailings. In the limit the molecular diffusion coefficients for lead will be of importance as well as the levels of oxygen concentration and pH. There may be other factors as ,...el1. It has already been pointed out that the eventual exchange of the basin waters in A fjord is inevitable unless the sill is closed off by a solid dam. At that time a considerable proportion of the dissolved heavy metal above the tailings are bound to escape into Q fjord. 23 4. Hethods of retaining polluted waters within A fjord and their probable COnsequences An obvious method of controlling the discharge into Q fjord is to build a dam across the A fjord sill complete with a sluice gate to control the rW1 off of waters. It is understood that this is im- practical on two grounds; cost, and the fact that the ore carriers docking facilities are in A fjord. Nevertheless it will be considered because, in effect, A fjord is then a tailings pond and similar considerations would apply to any alternate location. Suppose then that the dam were built and the gate were controlled to produce a fresh water layer 4 or 5 meters thick on the surface of A fjord at mid summer. It has been seen that the usual depth of wind mix in A fjord with a typical summer density profile Nas about 6 meters, but with the retention of fresh water this would be significantly reduced, and it is probable that little mixing would take place along the salt/ fresh water interface. Winds blowing across the surface would cause set-up and calculation indicates that a steady 20 knot (10m/sec) wind blowing down A fjord would cause a difference in level of 52 cm between the two ends of the fresh/salt water interface - a matter of little consequence. The period of oscillation of internal waves on the densi ty interface is shown in Figure 11 a,s a function of the fresh water layer depth. These will be excited if there is .significant energ}r~ In the wind field" at these frequencies-~ It- Is- ant-:LcJ.pate.d that in the late fall, just befo-re -freeze-·up;--t'he-s-urface 'layer salinity should not exceed 5% 0 on average. Vertical convective circulation associated with surface cooling will be greatly restricted even within the upper layer itself as the maximum density of water of this salinity occurs at temperatures above freezing point. Thus, the sensible heat available from the water column will be greatly reduced and one would visualize that freeze-up could occur up to two weeks earlier than if the surface layer were not controlled. Ice would begin to form and eventually develop the structure as shown in Figure 8. Even if all the salt were rejected from the ice during growth the salinity of the upper layer would still only increase to an average of about 6 or 7%°' far below that of the underlying waters and no downward interlayer miXing would occur. Sea water for use in the extraction process would be taken from outside A fjord·at a depth where the salinity would at 1east equal that at the bottom of A fjord and care taken that it was ejected into A fjord with an enhanced salinity and at a temperature not far above freezing. The 6000 m3/day tailing 'inflow constitutes about aIm layer laid over A fjord during a year so that 'vithout wind mixing the fresh/salt interface would move up by this amount annually. By adjusting the thickness of the upper layer the top meter of the lower layer could be wind mixed into the upper layer which would then have to be replaced each year to avoid a cumulative salinity build up. This, in turn, sets a minimum required elevation of the top of the sluice gate above the high \vater mark in Q fjord - that of the desired fresh water layer, say 3 m as a compromise to allow some mixing by wind in summer but none by vertical saline convection in 'vinter. RESONANT PERIOD OF INTERNAL MODE l-+-----A/2·4km ,.' --~---------1-------r_----- 7 0 h 6 ,0 - 1.027 gm/,m 3 SOm 0 5 - 0 ~ lfl L :J 0 4- .J:: o ~ 3- o 0 0 n:: o o w 2 o Q FIGURE 11 1 Resonance periods for internal waves on the interface between seawater and a fresh water layer trapped behind a dam (solid or floating). o~--------+---------~---------+----------~------~1 5 2 3 4 HEIGHT h (m) 25 The overall level of the pond or fjord would then be dropped as far as possible just before the main snow melt ,.,hile checking the electrical conductivity of the discharged water and raising the sluice gate if it exceeded some predetermined level. During this period the rotting sea ice would prevent wind mixing to any appreciable extent while the fresh surface layer was being replaced. The significant environmental difference between a dammed A fjord and a tailings pond is probably the volume of run off but as winter vertical convection is not really of great importance when pond and sea are totally isolated)ice growth itself may be used to purify the surface waters. If it may be assumed that the annual ice g:~~~_~xceeds the annual increase in ~ater height due ,to discharge of tailings, in theory it is always possible to run off that amount of '-'neai:'ly pure 1jah~-r' every--year so it-s to--return- to t.he"same level for the start of the next years cycle. Year by year, concentration of dissolved metals will, of course, build up in the pond and it has already been noted that these profiles are not predictable on the basis of existing knowledge. However, impurities will be excluded from a surface water laye~,. at least ~qual t9 .. ~~, maximl.Ull ice thicknes!=i: __ ,_ ,E~~.e:~tial fac;.:t,ors for success in ss=~e;~~~._~£, ._thi~ __ ti"pe are~q;t.low!?:_-:-. 1) that a sufficiently thick layer of fresh water can be retained each year on the surface to reduce wind mixing to small proportions 2) that internal wave systems existing on the interface between the two layers do not exceed a certain amplitude where shear mixing may occur. This, in turn, depends on the natural frequencies of the basin and the energy content of the wind. 3) the volume of tailings discharged converted to a water level increase over the year does not exceed the annual ice growth thickness. The sluice gate would be dropped just after freeze-up to allow the escape of possible pure water, then raised by 3 or 4 meters. Ice growth in the winter would purify another meter layer and hopefully the snow melt would once again fill up the pond to where \"ind mixing was effectively eliminated. Of the blO water control proposals available for use with a tight dam, a fresh water layer summer and winter, or summer only using ice growth to purify surface.water, the latter is preferable as it eliminates the need to remove salt annually to compensate for the tailing input. In addition it is not necessary to know the rate of solution of the heavy metals. Only the former proposal is available for non water tight systems. Because of difficulties associated with bringing vessels through locks in a solid dam it has been suggested that a "dam with gap" should be built leaving a passage for ships in and out of A fjord. This would retain a thicker fresh water layer than existed before through th,e greater part of the year, but at the critical time just before freezeup when we anticipate that flow from the glacial streams ...sill have reached a very low level it would do nothing to prevent the eventual escape of the remaining fresh water; no t~...o layer system would be possible and saline convection would, once again, occur to the bottom 26 of A fjord. A dam of this type may have positive disadvantages as i t would increase the velocity of flow of tidal waters in and out of A fjord to the level where it is quite possible that significant entrainment of the lower waters might occur and, as a result, heavy metal discharge might occur at all seasons of the year. The pollutant concentration above the tailings ''''culd be reduced at all times of the year r not just during the ,,,,inter so that more would go into solution and the annual discharge of heavy metals would increase. Allother possibility is a "leaky dam" formed by a loose rock fill across the mouth of the fjord and sill. Such a dam ~7ou1d do Ii ttle to retain the fresh water layer but would act as a diffuser for the tidal currents so that if, due to the density structure, any high velocity levels existed with a potential for downward mixing, they would be eliminated. A more useful idea would be a rock darn faced_or cored for the upper three or four meters to be water tight and fitted with a sluice gate/lock system for water control and entrance to the ships. This would appear to provide most of the necessary factors and the reduction in current velocities both by control of the outward flowing current in the upper layer and of tidal diffusion to the lower layer would probabl~ extend the period of the inevitable basin water exchange greatly. A further possibility, and much cheaper to construct, would be an ice boom across the mouth of the fjord with a hanging curtain 3 or 4 m deep with a removable section or sections for the entrance of shipping. Structures of this type exist on the St. Lawrence River to protect sensitive areas from ice pressure. Such a floating structure could not, of course, have a sluice gate but it would appear that the most suitable design would allow escape of fresh water over the curtain rather than underneath it in order to avoid comparatively high velocity water corndng into contact with the lower . laver. _At toe ~ame time it would b_e ,undesirable to "lose" large quantities of fresh water due to surface waves,' etc •. "tn"'the al:iove- .. _surface portion the curtain might be a type of diffusing screen. This is the author's presentl~i-favoured solution. The only signIficant disadvantage that is apparent is enhanced tidal flow velocities due to the restriction in a cross sectional area available for tidal exchange between Q and A fjords. An entirely different form of solution would be to pump fresh water beneath the growing ice, shortly after freeze-up, so as to prevent vertical convective circulation. This would certainly work providing always the temperature of the water discharged was a few tenths of a degree above freezing point. As the bottom of the water in Arctic lakes must be around 4°C this would become a problem in insulation during transit from its source to discharge in A fjord. After the ice had grown sufficiently to support the operators a slot could be easily cut in the ice over the sill and a short plastic skirt hung through to restrict the loss of fresh water into Q fjord. Weekly monitoring would then be sufficient to ensure that a fresh water layer was kept there at all·times. ) 27 Should any of these schemes be implemented the question arises as to what situation would exist upon the termination of mining activities, maybe 10 years hence. What further inspection and control of the polluted fjord would then be necessary? There is a hope that the tailing would be overlaid by a layer of rock flour which must materially reduce their ability to go into solution. Another concern is what would happen if the floating boom herein suggested as an optimum solution ruptured and the 4 or 5 m layer of fresh water discharged into Q fjord? The floating boom solution will naw be investigated in detail •. ) 28 5. A Floating Boom Dam Any structure placed over the sill of A fjord to retain fresh water will have to withstand ice pressures when break-up occurs in the spring and it is fortunate that the problems posed in designing ice booms have heen quite well explored. Perham (1974 and 1977). 'give's design details of structures used on the St. Lawrence River. The engineering detail of a floating dam of similar design is beyond the scope of this report but in theory it liOuld appear that a hanging curtain might be added to an ice boom which would then be built in hinged sections that could be individually transported without difficulty. Suppose then such a darn were installed across the mouth of A fjord, conplete with a "lock" section for ore carrier transit with an impervious "sail" extending a distance h above the free water surface and a weighted "keel" hanging dowmvards to an indefinite depth. Fresh water flowing into A fjord would then displace salt water near the surface until a fre"sh water layer, approximately SOh thick had been b~ilt up when further fresh water would flow over the sail. If sail height and keel depth were limited to the values required to produce the designed fresh water layer thickness there would be losses due to surface waves over the sail and interfacial waves under the keel. It is desirable that the main flow of fresh water be over the sail to avoid comparatively high currents at the keel bottom with possible entrainment of saline ~qater from beneath. Nevertheless, once the ice has formed it may well be that no'flow route over the sail would exist and any fresh water entering the fjord (possibly from underground flows) would have to escape beneath the keel. This might occur suddenly . <luring !i ,storm ~n?- would be p,art,iC;::,uf,?-rly undesirable. Poss;ibly the best solution is that sail and keel should bot~_consist of solid sections surm:>unted by leaky diffusing plates. The 'upper diffUsing -plate would allow a steady flow outward into Q fjord to balance input into A fjord and normally prevent large scale water loss from waves. The lower plate would not normally carry a steady flow but would tend to smooth out any escape events. This is illustrated schematically in Figure 12. The mechanical response of the system to surface waves would need to be investigated. with a floating darn the escape of waters to'compensate for the annual I'm layer increment associated with the deposition of tailings' would no longer be necessary. Saline ~laters would move out over the sill beneath the keel at miniscule velocities to achieve a volume balance. The fresh water layer would be automatically renewed and there would be no sluice gates to require attention. The layer thickness could be adjusted in a practical design by blocking off apertures in the sail diffusing plate. From Figure 11 it appears :that a suitable layer thickness to prevent wind mixing might give a two hour period to the longitudinal. mode of the fjord and it is thought unlikely that energies at this frequency would be dominant in spectra of the wind or tides. The major uncertainty in this design is the time rate of change of the pollutant concentration profile - a topic that has already been discussed and is illustrated in Figure 10. Whatever rates of profile change are observedJthey can be minimized by lowering the tailing discharge outlet level to near the bottom of the fjord and taking precautions to ensure that no vertical circulation of the convective plume type' can rise from the discharge, specifically the water should be cooled and its salinity -- _.-_...--.-----;z-:-n..r -.. - ... - ..... ~.~ .. ,0'. ......... ~.~.;; .:. .... FLOATION "LOGS" FLEXIBLE SKIRT I I 00000000 o STEEL . ROPE IGHTS FIGtffiE 12 Sche~atic· of an . floating d l.ce boom .retention : . ~orfJord. .fresh wate r 30 checked so that it is at least as dense, preferably denser, than the ambient water at the point of discharge. In the normal course of events an exchange of basin water in A fjord would be inevitable as has already been discussed. In fact, restriction of vertical circulation by the fresh surface layer would prevent the basin getting its annual "salt supplement" and so hasten this exchange. If, how- ever, the tailing waters were returned at an enhanced salinity, significantly greater than that existing outside, the exchange might be delayed indefinitely. It is understood that part of the water extracted from the fjord is distilled to provide fresh water and thus the salinity of the discharge water could be increased materially above that of its source. This would have the additional effect of creating a greater stability in the bottom waters of the fjord so that the diffusion of pollutants upward from the tailings might almost reduce to those levels predicted by the molecular diffusion coefficients~ In addition, if the salinity of the lower waters of the fjord could be increased sufficiently the winter vertical convective circulation, the main cause of the present problems, would cease ~vi thin the basin~ From the Zubov calculations shown in Figure 9 it may be predicted that the winter convection below 30 m would cease if the salinity of the basin \-later exceeded 35.5% 0 at DoC or 37% 0 at 10°C. The floating dam is now seen as a temporary solution to be continued tmtil the basin salinity has been increased to a suitable level. Even if mining operations make the salinity enhancement of the tailing water difficult, presumably the enpty inco~ng ore carriers could bring rock salt which could be crushed and fed into the discharge outlet~ The total weight of rock salt required would be of the order of 10 5 metric tons~ Recommendation of an optimum salinity for the discharged water will require considerable further study as well as a much better data base. An increase in basin water salinity would therefore reduce upward diffusion of heavy metals, prevent winter time convective mixing and postpone basin water exchange indefinitely. Hopefully the tailings, when covered with natural sediment after the mine is closed will no longer go into solution and it even may be possible that the rock flour ~.,ill tend to precipitate the remaining dissolved heavy metals as has been postulated to happen in Q fjord. Even if this is wishful thinking, the enhanced density of the basin waters ~ill cause the release of pollutants over the sill into Q fjord to occur very slowly. How slowly cannot be predicted with the present data base and is bound up with recommendation for discharge salinities. A final consideration, the effect of rupture of the floating dam is dealt with in the appendix~ A small rupture of 5-10% of the length of the dam, would cause a 4 m surface layer to be reduced to I m over the course of a couple of days, which would give plenty of time to get busy on repairs (spare sections might be stocked for the purpose). Sudden removal of the entire dam would produce less easily predictable consequences, but there are reasons to expect that the transient respOnse time of the fjord would be 20 hours or more. Entrainment and exchange would be lim! ted to the \...ater above sill depth and most likely to the upper 10 m. Realizing that significant dam breaks must occur in open water conditions Table 1 indicated that this loss should not be serious. Finally and obviously the removal of the fresh layer leaves the fjord sUbject to wind mixing hut probably not more so that it is at present. 31 6. Suggested programme of environmental observations The fjord responds to the energies in tides, winds, surface run off and ice growth, and it is essential to gain information about the time scales of these inputs and of the response of the system. Most of the necessary data can be obtained from recording instruments but one essential parameter, salinity, cannot yet be measured automatically at the level of accuracy necessary for useful density determination. Specific recommendations follow. in each case, explaining the use of the proposed measurement. 1) Meteorology 'An automatic weather station should be installed to give wind speed and direction, air temperature and, if possible, air pressure throughout the year. The sensors should be located so that the values recorded are not~ffected by the immediate buildings, rock outcrops, etc. They should also be as close to the shore of the fjord as possible but it is probable that the recording unit woUld need to be kept indoors during the winter. The measurements would be utilized to estimate the maximum depth of shear wind mixing and to obtain the wind energy spectrum and examine it for peaks when optimizing the fresh water layer thickness in terms of the resonances possible within the fjord. 2) water Density Profiles A simple "conductivity-temperature-depth" (CTD) system should be purchased and records taken every two weeks at Stations 4 and 10. A member of staff from the mine should be trained to make these readings and contracted to do so. He should be also required to go through a standard calibration procedure prior to making each set of measurements, water density is the most important of all the parameters to be measured. It controls eve~ type of water movement discussed in this report from internal waves to vertical convection due to sea ice growth. 3) Water Movement It is recommended that two recording current meters, complete with sensorS for conductivity and temperature be installed just on the A fjord side of the sill where the water is, say 30 m deep. One current meter would be placed at mid water depth and the other l'lithin two meters of the bottom: they would record for a year. The data would give information on basin water exchange processes, if any, seasonal changes in density as well as provide a record of any long period internal waves existing at these depths. 32 4) Temperature Profile To supplement info~tion taken with the "CTO" survey a recording thermistor chain should be installed close to site 4 taking readings at five minute intervals over a period of a fe~" weeks in summer conditions. The instru- ment could be installed and recovered during the usual September survey. These mea5u~ements are required to record the amplitude and frequency of internal waves existing in A fjord. It should also be deployed in the vicinity of the tailing discharge outlet during the winter survey trip to record the associated water movement. 5) Tides A tide gauge should be put just outside the sill in Q fjord during both summer and winter surveys and set to record at 5 minute intervals to look for high frequency water height fluctuations. These often exist in complicated fjord systems and are a major energy source for driving internal waves. 6) Run Off It is essential that the fresh ''later run off into A fjord and its seasonal variation be known in order to predict filling rates behind the dam, entrainment of the lower layer waters at various times of the year, etc. etc. If and when such time series are av~~able it will enable selection of those processes that are truly important in terms of vertical water movement in A fjord from the very large number of· possibilities that exist. It is estimated that the total value of equipment to be used would be about $35,000. In addition to these field studies it is necessary to Obtain further information about the tailings themselves; maybe in a flume tank. What water velocities are re.quired to keep the tailings in suspension? What velocities are required to resuspend them? (usually quite different). What are the rates at which the heavy metals will go into solution as a function of the concentration immediately above the tailings? The need for this information has already been pointed out in detail. If all the information listed above comes to hand the present report can be rewritten and the optimum tailing disposal system designed on the basis of the observed physical processes with little need for speculation. lfuether or not this work is undertaken it is recommended that the tailings discharge point be lowered and the salinity of the tailing water increased to 40%°' at DoC as soon as is practi caL 33 Appendix HYDRODYNMIIC EFFECTS OF A BREAK IN THE FLOATING DAN • From the engineering point of view two practical considerations arise if the dam breaks: 1) How rapidly will the trapped fresh-w'ater layer disappear? 2) What are the dynamic processes that are likely to draw pollutants from the deeper layer as a result of the darn break? Background This is a cOng;llex problem of the "selective withdrawal" type studied in reservoir dyanamics. Due to the poor data base, it has not been thought useful to carry out a full analysis alo?9 these lines in the present report. There are also other unpredictable influences due to the wind and ocean.ographic processes outside A fjord. In view of these difficulties the following calculations should be considered as no more than first order estimatesi they can only be relied upon to give a ro,ugh estimate of likely events following a break in the dam. Further study as well as much better field data and possibly some laboratory measurements are required to resolve the consequences of a dam break in detail. The basic and conservative assumption is made that the water layer . . , interface remains level at all times following the dam break. In practice, a small decrease in depth near the mouth of the fjord would be expected. Any error involved in this assumption would have the effect of underestimating the time required to drain the surface layer. ,, A model for withdrawal The maximum transport of a two-layer fluid through a constriction occurs when 2 Fj + (1) = 1 h and h1' h2 and U1' U2 are the layer depths and were Fi = Ui ( g ~.)-Y2 P l. velocities of each layer. This problem was first investigated by Stommel and Farmer (1952, 1953) in connection with overmixed estuaries. An important exception occurs in the presence of barotropic forcing (stigebrandt, 1976), but this is not expected to be relevant in the present case. For a shallow surface layer in which the lower layer is essentially motionless, critical conditions limit the flow when the flow velocity equals the internal wave speed; approximately: u. Uj • (g ~j) ~ p A higher velocity would imply supercritical flow in one layer with hydraulic transitions to subcritical flow and internal wave breaking ,along the interface. It seems unlikely that such conditions could be maintained, so that equations (1) and (la) seem useful -for estimating an upper limit to the depth mean velocity. Wind shear could introduce further conplications but is not considered here. (I-a) 34 From the fluid dynamical point of view there is an important difference between a relatively small break, for example a break between two segments of the floating dam, and the more catastrophic case of the entire dam being carried away. In the first case.the area of high shear in which entrainment or mixing will occur is confined to the region immediately adjacent to the break so that estimates of the loss of surface water may be based On the physical reasoning underlying equation (1) or (I-a). For a large break, or the removal of the entire dam, quite different physical processes will apply and a simple two-layer model would give misleading results. In what follows we shall model the simpler and perhaps more likely case of a limited break. Some of the additional physical considerations applicable to the full-scale darn break are also briefly discussed, but no attempt has been made to model this latter example. This is an aspect that would require further study prior to detailed planning of a floating dam. Consider the events immediately following a lirni ted r,upture of the floating dam. The lighter surface water would rapidly accelerate through the break, radially spreading out to a thin, slow moving layer beyond the dam. Beneath the dam salt water would move inwards to replace the lost surface layer, but this inflow could occur throughout the full width of the channel. Thus no rapid opposing current immediately beneath the outflow is e~ected. Cqnsider a break of width b in a dam of total W~h B at ~he mouth The rate" of change of volume of a channel of length L and uniform breadth. of fresh \-rater is then given by d dt (BLh,) "'" - bU,h,. (2) But from (I-a), substituting for U, and henceforth dropping the subscript, we obtain d: (BLh) = - bh (gl!.~ h)~ (3) or h-~ db = - bg':!:;: dt (4) BL where g' = g ~ p is the reduced gravity. Integrating (4) and evaluating the constant of integration from the requirement h = h at t = 0, gives o (5) or rearranging (6) 3S As an illustration, consider the example of a break of 5% of the dam width. ~p Taking the channel length L = 4000 m, = 0.02 and an initial depth h = 4 ro, equation (6) shows that it W~Uld take about o 50 hours for the surface layer to drop to 1 m thickness. This calculation almost certainly overestimates the withdrawal rate since no account is taken of shallowing of the interface near the break due to acceleration, or of the retarding influence of interfacial shear stresses and entrainment. It is assumed that the scale time of surface layer response was significantly less than the scale time for withdrawal. This is justified a posteriori in the present example if the time required to reduce the surface layer to 0.25 h is accepted as a suitable scaling. As indicated above, the assumptiogs will tend to overestimate the rate of withdrawal producing a "worst case" condition since the surface layer thickness at the break wil~ tend to be less than that in the fjord. Dynamic Processes associated with the dam-break 1) When the dam breaks a wave of elevation change will travel back along the interface towards the inlet head. If this wave tended to steepen due to non-linear effects, short period waves with relatively high shears could be generated and these might lead to the generation of turbulence and mixing (c.f. Farmer & Smith, 1978). This possibility may be checked by consideration of a first order approximation to the non-linear equations of motion. For a two layer flow' it can be shown that the coefficient of the non-linear" term p is proportional to Jh2-hl)/hlhi, where hI and h2 are the surface and lower layer thicknesses respectively. Moreover a wave of slope a, in a coordinate system moving with the wave will al\qays tend to steepen and break if both ~ and p are greater than zero. (For a particular example, see Farmer, 1978). In the present example, the surface layer withdrawal causes an elevation of the interface. In a coordinate system moving with the wave this implies that a>o. Since hI < h2 we have p > a and this combination of parameters means that non-linear effects will decrease the slope and so reduce the possibility of short period waves' breaking following the in~tial darn break. When the waVe is reflected it will eventually reach the fjord mouth. Although reflection might be expected this \oIi11 be modified to some extent by interaction with the critical conditions at the break. "In this connection it should be noted that non-li!1ear effects are enhanc;ed when the \oIave travels into an opposing shear, an effect that would work in opposition to the effect of relative layer thickness described above. '\ 36 2) Since the flow is hydraulically constrained at the location of the dam break the existance of significant shears is implied. If the flow really is two-layer in nature, the bulk Richardson Number Rio ~ g'hl/U2 = 1 and with a thin, possibly turbulent layer, some /degree of entrainment must be expected. The entrainment rate E is hard to estimate. This value of Rio lies between the experimental values determined by Lofquist (1960) and by Ellison and Turner (1959). Nevertheless, by interpolation E may be estimated (7) where w is the vertical velocity and U is the relative velocity between the two layers. The surface layer, being thin and more turbulent, corresponds to the entraining layer represented by the laboratory experiments referenced above. There are considerable difficulties in extrapolating model results, especially interpolated model results, to geophysical scales. It must be emphasized that these are only first order estimates of what might occur. From (I-a) and (5) u = gl~ 2BL/(2BLh:~ + b91~t) • (8) For the example considered above, equations (7) and (8) would imply an initial vertical velocity of 0.018 ms- 1 dropping down to .005 ms- l when the surface layer thickness had reached 1 rn. These entrainment velocities are initially quite large; however it.is important to note that much of the entrained Hater will be dra~m from inflowing water entering the fjord at, or just beneath the foot of the dam. 3) As mentioned above, the simple model used here will only be applicable for limited rupture of the dam. In the event of sudden loss of the entire floating dam, the simple flow model used above would almost certainly break down and equation (1) would no longer be applicable. since the area of rapid flow would no longer be confined to the immediate vicinity of the break, significant entrainment, together with corresponding interfaced shear stresses, would occur over a ·large area. This would have the effect ·of slowing down the seaward movement of surface water and introducing a more gradual salinity gradient across the interface. The problem of estimating an exchange rate now becomes very compl2x and highly nonlinear. Moreover the reduction in density gradient would have the ~ffect of increasing the characteristic transient response time of the inlet. It is possinle that oscillatory effects would dominate the exchange process, a possibility that might usefully be investigated with laboratory models. There is little observational evidence of transient surface layer movements in fjords that can give guidance when estimating the time scales to be expected in a large scale exchange. Observations we have made in Alberni Inlet of large scale wind induced 37 motions emphasise the dominant effects of shear stresses on the response. A strong wind effect had a relaxation time of three or four days over the upper reaches (10-15 km) of the inlet. The relative density difference and layer depths were comparable to the present example. Allowing for the shorter length of fjord A and assurrdng the influence of shear stress to be similar~ one might expect a response time of perhaps 20 to 30 hours. It seems likely that a simple flushing out of surface water will not occur over this period; a more probable consequence would be the loss of a significant fraction, followed by the formation of a continuously stratified surface layer. Weak density currents would then continue to replace this new surface layer with outside water but at a more gradual rate. DAVID M. FARMER 38 REFERENCES Ellison, T. H. and J. S. Turner, (1959) Turbulent entrainment in stratified flows. J. Fluid Mech. ~, 423-48. Farner, D. M., (1978) Observations of long nonlinear waves in a lake. J. Phys. Oceanog.·.!!, No.1, 63-73. Farmer, D. M., and J. D. Smith (1978), Nonlinear internal waves in a fjord, Hydrodynamics of estuaries and fjords, ed: J.C.H. Nihoul, Elsevier, pp 465-493. Gade, H. G. (1973) Deep water exchanges in a sill fjord: a stochastic process. J. Phy. Oceanog. Vol 3, No.2, pp 213-219 Gade, H. G., R. A. Lake, E. L. Lewis Oceanography of an arctic bay 5; E. R. Walker (1974) Deep Sea Research, Vol. 21, pp·547-571 Hellstrom, B (1941) Wind effects on lakes and rivers Ingeniorsvetenskapsakaaemiens Handlingar nr 158 Stockholm Lofquist, K. (1960) Flow and stress near an interface between stratified liquids. Phys. Fluids, ~, 158-75. Perham, R. E. (1974) Forces generated in ice boom structures Special report no. 200 U.S. Army CRREL Hanover, New Hampshire Perham, R. E. (1977) St. Mary's River ice booms design force estimate and field measurements U.s. Army CRREL Report 77-4, Hanover, Ne'''' Harrpshire Perkin, R. G. and E. L. Lewis (1978) Mixing in an arctic fjord J. Phy. oceanog. To be published (tentatively scheduled for Vol. 8 No.5) pollard, R. T., P. B. Rhines, R. Thompson (1973) The deepening of the wind-mixed layer. Geophy. Fluid Dynamics, Vol. 3, pp 381-404. Stigebrandt, A (1976) On the effect of barotropic current fluctuations on the two-layer transport capacity of a constriction. J. Phys. Oceanog. Z, 118-122. Stommel, H. and H. G. Farmer (1952) Abrupt changes in width in two-layer open channel flow. J. Mar. Res., II, 205-214. Stommel, H. ahd H. G. Farner (1953) Control of salinity in an estuary by a transition. J. Mar. Res., ~, 18-20. Zubov, N. N. (1943) Arctic Ice (English translation) u.s. Navy Electronics Lab., San Diego, Calif. pp 491.