JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO. C1, PAGES 909-915, JANUARY 15, 1986

Salinity Intrusion in the River Estuary Following a Major Reduction of Freshwater Input

S. LEPAGE AND R. G. INGRAM

Institute of Oceanography,McGill University,Montreal, Canada

On July 19, 1980, 80% of the runoff of Eastmain River, a subarcticshallow estuary,was diverted into the for hydroelectricdevelopment. Consequent to the diversion the estuary, which had been mostly salt-free,was subjectedto a gradual saltwater intrusion along a 10-km section in its lower reaches.The adjustmentof the salinity regime to a new quasi-steadystate took place over a period of about 40 days. The circulation field reached a new quasi-equilibrium within 8 days. The 2-month interval following the river diversion,termed the transition period, was analyzed in terms of both time and space modifications of the circulation and salinity fields. A one-dimensional finite difference explicit scheme numerical model was applied to the estuary as a complementto observationaldata. Good agreement was obtained between the model results and the low-frequency current meter observations of salinity and longitudinal velocity. These results showed that the salinity intrusion was primarily governed by tidal dispersion.

INTRODUCTION Eastmain is a coastal plain estuary with a very large width- An increasing number of diverse projects require modifi- depth ratio (700 at the mouth) and an exponential shape, cation of river systems,such as reservoir regulation or fresh- widening from 600 m near Basil Gorge to 2.2 km at its mouth. water withdrawals. In these cases,upstream changes of the The lower reachesof the river are very shallow, generally less natural river runoff will inevitably have consequenceson the than 6 m, with a seriesof bars and rocks in the central por- tion. estuarineportion of the river. Although river managementhas commonly occurredover the past 50 years,few oceanographic Prior to its diversionin 1980, the mean annual dischargeof studies werz conducted, and little is known about the immedi- the Eastmain was 1190 m3 s-x, ranging from a monthly ate and long-term estuarine responseassociated with the river average maximum of 2460 in June to a minimum of 260 in managementschemes [Kjerfve, 1976; Kjerfve and Greet, 1978; March. The July 1980 diversion, preceded in April by the McAlice and daeger, 1983]. For these reasons,there is a need damming of the Opinaca River, a tributary of the Eastmain, cut the total river discharge by over 80%. The postcutoff for more field observationscollected in conjunction with any mean annual dischargeis expectedto be 250 m3 s-•, with a estuary-relatedmanagement. Analysis of thesedata will also monthly maximum of 510 and a minimum of 50 [Prinsenberg, further our basic understandingof the estuarineprocesses and 19803. provide suitable data for modeling studies.The development of the hydroelectricpotential in rivers on the easternshore of FIELD DATA gave an opportunity to study the estuarine re- During a 3-year program, oceanographicconditions were sponseof the Eastmain River to a drastic runoff reduction of monitored for open water as well as under an ice cover in late about 80%. On July 19, 1980, a spillway gate located some winter-early spring. Becauseof the remote location, emphasis 150 km upstreamof the river mouth was closedwithin min- was placed on data obtained with self-recordinginstruments. utes. The diverted water was fed into the La Grande complex A series of Aanderaa RCM-4 current meters were moored to the north. Becauseof the small drainage basin downstream within a 13-km section near the river mouth (Figure 2 and of the spillway, mean annual dischargeat the mouth of the Table 1) to monitor (10- or 20-min sampling) current speed Eastmain River was cut to 20% of its former value. and direction, temperature, conductivity, and pressure. An The present study includes a description of the estuarine Aanderaa tide gaugewas installed near Eastmain Village (sta- responseto the above modification of freshwater input and tion T2) for a 15-day period in September 1979 and a 40-day discusses the results of a one-dimensional numerical model of period in June-July 1981. the estuarine region for variable discharge. Conductivity, temperature, and depth (CTD) surveysof the

THE ESTUARY coastal waters adjacent to the river mouth were executed within a 4-hour period centered on high and low tides in Oriented east-west, the Eastmain River originates in the August 1979 and July 1981. Transects along a single line ex- Canadian Shield and empties into the eastern portion of tending offshore from the river mouth were done on numerous JamesBay (Figure 1). With a natural drainage basin of 47,000 occasions. Data were collected with Beckman RS-5 and YSI km2, a length of about 500 km, and a mean slope of 0.0015, salinometers.During the summerof 1980,five Foxborolim- the river is in its juvenile stage prior to 1980. nigraphs were deployed along the estuary to measure water The estuarine portion of the river extends from the Basil levels. In addition, meteorological data (barometric pressure, Gorge rapids to James Bay, a distance of 29.5 km. Two small air temperature, and wind speed and direction) were taken at tributaries, Fishing River and Cold Water River, are located a site adjacent to station T2. Unfortunately, mechanical prob- within the estuarine area. In the geomorphologic sense,the lems with the instrument in 1980 led to difficulties in deter- mining a reliable time for each observation. Wind data were taken on ship at irregular intervals, usually three to four ob- Copyright 1986 by the American GeophysicalUnion. servationseach day. Hourly observationswere available from Paper number 5C0706. the Canadian AtmosphericEnvironment Servicestation at La 0148-0227/86/005C-0706505.00 Grande, 200 km north. The terrain in the area is relatively flat.

9O9 910 LEPAGEAND INGRAM: SALINITYINTRUSION IN THE EASTMAINESTUARY

78 ø 75 ø 72 ø 69 ø W I IJJUARAPIK I ! i HUDSON BAY i I LA

LG2 LGI ,G3 CHISASlBI

JAMES EMINDJI BAY :A R.

EASTMAIN

EMI FORT RUPERT

r-/73 STUDY AREA

0 km I00 ß PROPOSED OR EXISTING HYDRO DAMS I i • ß I I

78* 75ß 72ß 69ß Fig. 1. Map of the La Grande complexand studyarea.

The predominant winds were from the northwest or south- technical report, 1976). Thirty-three cross sections,with a west.Although these directions are at a 45ø anglewith respect 0.8-km spacing,were monitoredin an upstreamdirection from to the river axis,it is to be expectedthat waveeffects during the river mouth.These data wereused in conjunctionwith the periodsof strongwind will lead to someamplification of the model to be discussed later. velocities recorded on the Aanderaa current meters because of All current vectorswere decomposedinto along-channel their shallow mooring depths.Salinities calculated from the components(E-W), positiveinto the river (90ø true),and cross- current meter conductivity observations are assumed to be channelcomponents (N-S), positivenorthward (0øT). Current accuratewithin 0.5• becauseof the low valuesgenerally pres- meter data were then filtered with an a6a6a7/6,6,7 type ent. moving mean to reducethe relative frequencyof observations Bathymetric characteristicsof the estuary were taken from to hourly values. Smoothed hourly values were also filtered an acousticsounding survey of the lower 26 km of the river in with an a24a24a25/24,24,25 filter to eliminate diurnal and 1976 by the James Bay Energy Corporation (unpublished semidiurnalfluctuations [Godin, 1972]. Tidal analyseswere

178o40, I78ø30 ' 178ø20' 178o10,

o

152' 17' c=oo

o

15' EASTMAINRIVER 13 ßCl EASTMAIN T4 T3 COLD , WATERLEVEL RECORDERS ß CURRENT METERS I$' --BATHYMETRICSURVEY 1976 CROSSSECTIONS USED IN MODELS 0 km 5

Fig. 2. Map of the studyarea includingmooring location and crosssections. LEPAGE AND INGRAM: SALINITY INTRUSION IN THE EASTMAIN ESTUARY 911

•-RIVER performed on water height and current data [Foreman, 1977, I MOUTH 1978]. 18 •

ESTUARINE CHARACTERISTICS IN NATURAL CONDITIONS A detailed description of the circulation and salinity charac- 08-08\ • • teristics in the estuary and offshore plume for natural runoff I 28-08 conditions can be found in the work of Ingram [1982] and Lepage [1984]. Bathymetric data from the 1976 survey showed that most of the estuary was less than 8 m deep. Before the diversion the mean depth was 2.8 m. At the mouth (station T1) the dominant lunar semidiurnal tide (M2) had an average amplitude of 0.34 m. Four kilometers upstream at station T2, the M 2 component was 0.24 m. Tidal currents were typically 25 cm s-• betweenthese two locations.During August-September 1979, saline waters were detected at sta- I •8-o8 0 tions C2 and C3 (infrequently), mostly during the flooding -2 0 2 4 6 8 I0 tide. However,for dischargesexceeding 1500 m3 s-l, salt in- X (km) trusions were produced only by strong meteorologicalforcing. Fig. 3. Progressive salt intrusion following the river diversion at a Consideringthat the averagesummer (June-August) discharge depth of 2 m during summer 1980. was 1328m 3 s- 1 prior to modification,the Eastmaincould be data. Thus the ratio of salt dispersion by tidal diffusion to classedas an essentiallysalt-free estuary. In the coastal waters gravitational circulation was typically 0.90 [Hansen and Rat- adjacent to the river mouth, a large plume of more than 100 tray, 1966, Figure 1]. km2 in surfacearea was formed.Within the plume, vertical Offshore, at station C1, salinity values at 2 m rose gradually stratification was typical of a salt wedge configuration, the throughout the summer following the diversion (Figure 5). pycnocline varying between 1 and 2 m. The amplitude of the semidiurnal variability of the salinity POST-DIVERSION CONDITIONS field decreasedrapidly within 1 week of July 19. The overall estuarine responsecan be seen in the mean values listed in Following the diversion on July 19, 1980, the estuarine characteristics were drastically altered. As can be seen in Table 2. Although salinity and current data were collected Figure 3, the saline James Bay waters gradually intruded into during the natural low-flow period in March-April 1980, one the lower estuary in the following weeks. An example of the cannot compare these to those collected in the postcutoff evolution in time of middepth velocity and salinity fluctu- regime, becausethe dynamics are greatly modified by the ice ations at a singlelocation (for station C6) are shown in Figure cover during winter. Ice melt usually occurred in late April- 4. On average,the river flowsdecreased to about 5 cm s-• or early May. lesswithin a period of 8 days. The middepth semidiurnal tidal In regard to water level fluctuations, it was difficult to evaluate the averagelowering along the estuary using the field velocitiesincreased by 75% and 100% at the inner stations C4 data. Mechanical problems were encountered with the lim- and C6, respectively.During the summer of 1981, with the nigraphic instrumentsused in 1980, and neither the recorder river dischargefluctuating about 100 m3 s-X, salinityvalues installed in 1979 nor the one installed in 1981 was referred to averaged 8-9%0 at the river mouth and 2%0 at stations C4N a bench mark. Thus the absolute water level could not be and C4S. The upstreamlimit of the salt intrusion was situated determined. However, the records obtained at stations T2 and near station C7. In terms of stratification, CTD data showed T3 in 1980 indicated a relative lowering of 0.1 m and 0.4-0.5 that the salinity differencein the water column was generally m, respectively.Tidal analysisof the 1979 and 1981 data at T2 less than 1-3%0 within the estuarine area. Laterally, the few showed an increaseof 30% in the amplitude of the M2 com- data available indicated very little variation (lessthan 1-2%o). ponent and of 200% for M,•. An insignificant increase was For the postcutoff regime, the estuary could be consideredas found at station T1. partially to well mixed. In terms of the Hansen and Rattray [1966] scheme of estuarine classification, the Eastmain was MODELING THE TRANSITION PERIOD transitional between types 2a and 2b during late summer The data collected in the Eastmain estuary gave a good 1980, as was determined from CTD, current meter, and runoff picture of the evolution of the oceanographic conditions during the summer of 1980. Since the period following the TABLE 1. Current Meter Moorings From 1979 to 1981 river diversion, termed the transition period, was of particular interest, a numerical model was developed to complete the Depth, Aug.-Sept. June-Sept. June-July study. Model choice was dictated by the estuarine character- Station m 1979 1980 1981 istics. Since the estuary closely matched quasi-well-mixed con- ditions, a one-dimensional approach was used. The overall C1 2 X X objective was to describe the long-term variation of salinity in C2 4 X X C3 2 X X a shallow tidal estuary for variable discharge.The model used C4 2 X X was of the type formulated by Harleman and Thatcher [1974] 4 X X and Thatcher and Harleman [1972a, b, 1981]. C4N 2 The governing one-dimensional equations for unsteady C4S 2 hydrodynamics and salt transport are a set of nonlinear, C5 1 X X C6 2 X X second-order,partial differential equations: C7 3 the continuity equation C8 4

X denotes stations sampled. b•7+•xx-q =0 (1) 912 LEPAGE AND INGRAM: SALINITY INTRUSION IN THE EASTMAIN ESTUARY

I•r

......

,-4%,.... W""-T ...... 'l" ...... P?lUlin!lillUtl]iq,?qqqlHlllltPlipi" l., , I..I.I II Illlll,,i.lI ..... ,I ,,i. JUNE JULY uusT SEPT i9eO TIME (E.S.T.) Fig. 4. Station C6 (2-m depth) salinity and velocity data. the longitudinal momentum equation To resolve this problem, a coupling of the estuarine model to a "plume model" was done so as to enable calculation of a OQ &Q &u &rl Q IQI (2) more realistic entrance salinity. This techniquewas inspired -- + U•xx + Q•xx + •7A •xx + •7A C2Rh from the plume length concept discussedby Harleman and the salt balance equation Abraham [1966]. The tidally averagedfresh water velocity themaximum flood velocity u.... anda quantityreferred to as (3) the tidal prism ratio lippen, 1966] are calculatedin the estuar- &(AS)&(QS) &( &S) ine model every tidal cycle.The averagelength is given by the equation of state for salinity B = C•Rø'SuvT (5) p = 0.75s + 1000 (4) where R = nu•/Umaxis the tidal prism ratio, T is the tidal period, and C• is a calibration constant.Superimposed on this where average is the tidal component represented by a cosinusoidal b channel width, m' function of maximum value uoT/r•, where Uo is the maximum r/ surface elevation, m' tidal velocity (flood velocity for B minimum and ebb velocity Q cross-sectionaldischarge, m 3 s-•; for B maximum). The ocean salinity So was taken at the ex- q lateral inflow per unit length,m e s- •' tremity of the seaward distanceB. The river mouth salinity S• u averagecross-sectional velocity,m s -'•' was then calculated using the solution of the one-dimensional •/ accelerationdue to gravity,m s-•; dispersion equation A total cross-sectionalarea, me' p fluid density,kg m- 3; S,/So = exp (urB/Kt,) (6) r wind stressterm, m s-2' C Chezy coefficient; whereur and K•,, the dispersioncoefficient, are evaluatedfor R h hydraulic radius, m' the plume length B. K,• longitudinaldispersion coefficient, m2 s-•; S cross-sectionalsalinity, 9/00. LONGITUDINAL DISPERSION Sincethe coefficientKx in (3) variesas a function of time t The solution techniquechosen to solve the above equations and longitudinal position x, it must include the local dynamic was the finite differenceleapfrog explicit scheme. conditions at any location in the estuary. Tides, meteorologi- cal conditions, freshwater runoff, and estuarine geometry all THE OCEAN BOUNDARY play an important role in estuarinemixing. Thus it is of in- A major difficulty in modelingtransient salinities is often terest to formulate a relationship for K,, in terms of these encounteredin prescribingthe outer boundary.The usual ap- variables. proach is to fix the river mouth salinity equal to the ocean In the present application, K,, values calculated in this salinity. This approximationis unsatisfactoryfor the present model are related to local values of the tidal prism ratio. A application.From analysisof the current meter salinity data, number of different relationshipswere used.A formulation of it was evident that oceanlike conditions were never attained at the type used by Thatcher and Harleman [1981] was retained the estuaryentrance, even when the river dischargewas at its for calculation of K: lowest. Moreover, before the river diversion, there was almost no salt water present. It was thus not possibleto set the Kx = C2R-ø'31lmaxD (7) downstreamsalinity equal to the JamesBay salinity (17-18%o) as a boundary condition. where D is the local estuarine width and C2 is a calibration

20 27 4 II 18 25 I 8 15 22 29 5 12 JUNE JULY AUGUST SEPT 1980 TIME (E.S.T.) Fig. 5. Station C1 (2-m depth) salinity data. LEPAGEAND INGRAM' SALINITY INTRUSION IN THEEASTMAIN ESTUARY 913

TABLE 2. MonthlyMean Salinity and Velocity Values' Before and After Cutoff

Local Instrument Mean Mean M2 Tide Depth, Depth, Day Salinity, Velocity, Direction, Velocity, cm s -I Station m m Year Interval %o cm s- • øT

CI 6 2 1980 172-201 9 5.4 235 42 4 1980 172-201 13 2.1 105 26 2 1980 201-232 14 1.2 270 42 4 1980 201-232 16 3.8 095 20 2 1980 232-257 17 2.8 020 38 C2 4 2 1980 172-201 1 20.1 275 29 2 1980 201-232 8 4.2 275 33 2 1980 232-257 13 2.9 280 29 C3 4 2 1980 172-201 0 31.1 235 22 2 1980 201-232 8 3.2 180 33 2 1980 232-257 13 2.5 130 32 C4 5 2 1980 171-201 0 45.5 270 26 4 1980 171-201 0 26.1 270 23 2 1980 201-232 1 9.7 270 37 4 1980 201-232 2 4.4 280 13 2 1980 232-266 6 6.0 270 46 4 1980 232-266 7 1.8 345 15 C6 6 2 1980 172-201 0 40.0 325 16 2 1980 201-232 0+ 5.7 330 35 2 1980 232-266 3 ......

The majordiversion occurred on day 201, 1980.

constant.Equation (7) wasalso used for the K v coefficientin Tidal elevations at the river mouth were introduced as max- (6), D beingreplaced by B. imum and minimum heightscalculated from the Canadian Tide and Current Tables. A cosinus fit was used to calculate MODEL CALIBRATION AND APPLICATION the intermediateheight values. This methodwas used because The estuarywas dividedinto 15 segments(16 sections)of the water level observationsin 1980 were of insufficientquali- 1.6 km (Ax) for a total lengthof 24 km (Figure 2). The odd ty to be usedas input data. James Bay salinity was set equal cross sectionsof the 1976 bathymetric survey were schema- to 17.5%o. tized as trapezoids.Channel angle was given in radians at The tidal hydrodynamicswere calibratedby determining eachsection, zero beingtrue east.A commonreference datum roughnessvalues (Manning's n introducedin (2)) whichbest was arbitrarily given. representedthe estuarinecharacteristics and by givinga value The stabilitycriterion relating the time increment(At) and to Cxin (5)for theplume length calculations. The longitudinal the spaceincrement (Ax), the so-calledCourant-Friedrich- dispersionwas calibratedto obtainK,, and K v valuesin (7) Levy condition,satisfies that producedsuitable salinities over the extremesof dis- charge.Best fit was obtainedwith a C2 = 0.35 for K,, values Ax and0.0023 for Kv values.The model was run in "quasi-steady at _< (8) u+c state"using a proceduredescribed by Thatcherand Harleman [1972b].The runswere executed for dischargevalues of 1200 wherec 2 = gh, c beingthe shallowwater wave speed. To de- m 3 s- x and 100 m 3 s-x. The Manning coefficientswere ad- scribeproperly the tidal forcing,the timeincrement must be justedby comparingthe computedtidal rangeat different smallin comparisonwith the tidal period. pointsof the estuarywith the availabledata. The valuesgiven Sincethe maximumsection depth was 7.5 m (section15) rangedbetween n = 0.025and n = 0.042.Methods to search and a value of 0.4 m s-• was taken as representativeof the for optimumn valueswere described by Davidsonet al. [1978] cross-sectionalvelocity, this led to a At smallerthan 178 s. A and Fread andSmith [1978]. Thesetechniques required a con- 150-s increment was chosen. sistentset of tidal data, which was not availablein this case. The problemof dispersioninduced in an explicitscheme has For plume length calibration,station Cx salinity data beendiscussed by Fisheret al. [1979].An apparentdiffusivity (Figure5) wereused as an indicatorof "oceanicconditions." K' resultingfrom thisinfluence has a maximumvalue of 0.125 This station,located some 4 km off the river mouth,matched (Ax)2/(At).K' valuesgenerated by the numericalprocedure JamesBay conditions (very small salinity fluctuations) in mid- were shownto be small (lessthan 15%) with respectto the Kx August,when the river dischargewas minimum. From these values in (3). data,the constantC• in (5) wasset equal to 2.0.This provides Becausea natural fall marked the upstream limit of the for plumelengths B of 7-10 km duringhigh discharge and 2-5 estuary,it was treated as a closedend with all freshwater km duringlow discharge.These lengths are reasonableesti- inflows introduced as lateral inflows q. Daily discharge rates matesfor precutoff[Ingram, 1982] and postcutoff(B. d'Angle- measuredat BasilGorge station(km 34.6)were utilized for the jan, unpublishedmanuscript, 1981, and CTD data, 1982)con- main inflow and were inserted between model sections 15 and ditions. 16. The first rapid is located I km upstreamof station T5 The model was applied to the transitionperiod of 1980. (Figure2). Since no datawere available for thecontribution of Figure6 showsthe dischargevalues introduced as input data. Cold Water and Fishing Rivers to the total discharge,they The missingdata (July 20 to August5) wereevaluated from were estimated(from current velocitydata) to accountfor the observedupstream velocities. The model was run with 2.5% (each)of the Eastmainnatural discharge. z = 0 over 175 tidal cyclesfor a 3-monthperiod from June18 914 LEPAGE AND INGRAM' SALINITY INTRUSION IN THE EASTMAIN ESTUARY

2000 - data and interpolated values showed low-frequency salinity variations of amplitude and occurrence similar to those ob- • MEASURED ...... ESTIMATED served (Figure 8). Thus these results indicate that the inclusion 1500 of wind stress into the model could account for the 3-5 day variability noted in the current meter observations. _.•••%•EA STMAIN RIVER A tidal current harmonic analysis was performed on both I000 ß the modeled velocity values and the current meter data. The ß

ß analysis revealed a general agreement between the modeled

ß

ß and the observed modifications of the lunar constituents 5OO ß COLD WATER a ß consequentto the runoff reduction. In the first month follow- FISHING RIVERS ß ing the July diversion, the M 2 and M,• tidal current constit- ..... e ..... •• ..... • ..... • .... ! ..... :...... ! ..... , .....• • ß .. q ..... , .... .,,.; ,-,-,•-• ...... •...... ! .... uents increased by 45-140% and 200-800%, respectively 16 2 18 3 19 27 12 28 JUNE JULY AUGUST SEPT 1980 (model sections1, 4, and 6). In the secondmonth following the TIME (E.S.T.) diversion,the M 2 values remained comparable,while the M,• Fig. 6. Dischargegiven as input data for the transition period. amplitudes decreased to values closer to prediversion con- ditions. The same pattern was observed in the current meter tidal analysis, although increases were somewhat smaller. to September 17, 1980. Data collected at three different moor- Some discrepancyshould be expected by comparing cross- ing stations (C3, C4, and C6) were used for comparison with sectionallyaveraged quantities and thosemeasured at a specif- the model results obtained for sections 1, 4, and 6. ic location. The current at station C6 has an important N-S component becauseof the N-W orientation of channel in this RESULTS part of the river. Since the current was decomposedinto N-S Model results and observed data over the transition period and E-W components,the values shown in Figure 4 represent were compared over the period from June 23 to September 15, only a portion of the total current. 1980. The computed salinity data are shown in Figure 7. In terms of mean water level modification, the consequence These values can be compared with the field observations of the runoff reduction was demonstrated by the low- shown in Figure 4. However, in order to follow the salinity frequency modeled water height data. An average decreaseof intrusion over the transition period, the data were filtered with 0.1 and 0.3 m was predicted by the model at sections4 and 6, the low-pass filter described in the field data section. These respectively.Although water level data collected with the lim- data may then be directly compared by superposition on a nigraphic instrumentswere of insufficientquality to allow for common axis, as is shown in Figure 8. In general, the model a detailed comparison, they did, however, indicate that the resultswere satisfactory.They demonstratedthe ability of the model values were realistic. model to reproduce salinity changes as a function of time- Calculated values of B from (5) showed a bimonthly vari- varying dischargeand tidal height. The numerical values com- ation caused by the tidal forcing in addition to the runoff puted, in agreementwith the current meter data, showedthat influence. The K,• values obtained with (7) ranged between 25 the most important factor governing the salinity intrusion, at and 40 m 2 s- 1 beforethe river diversion,and between100 and low discharge regime, was the dispersion caused by tidal 200 m2 s- 1 subsequently.Like thoseof B, calculatedvalues of action. The variability of the observed salinity signal in late K x include a bimonthly variability. August and early September (Figure 8) was thought to be caused by wind forcing. As was discussedin the field data CONCLUSIONS section, winds were obtained from three sources and were not The responseof the Eastmain estuary to a major reduction reliable enough to be used as input to the full model run. of freshwater input has been described. Overall, the low- However, a short run (not illustrated) using the available wind frequencyresponse of the estuary to the rapid changein dis-

18

iI,'• .... I ...... i ...... i ...... I ...... ,,lli ,...fiJJil,lJhhmlii i i i i i i 23 30 7 14 21 28 4 II 18 25 I 8 15 JUNE JULY AUG SEPT 1980 TIME (E.S.T.) Fig. 7. Modeled salinityvalues at (top) section1, (middle)section 4, and (bottom) section6. LEPAGE AND INGRAM: SALINITY INTRUSION IN THE EASTMAIN ESTUARY 91 $

[8,- • CURRENT METER

...... i; -,-';-;...... ,• rMODE ......

03 ...... i .... i i i i i i

0(;--:-; , , , i ...... i ...... i ...... i ßß "";';'" i ,--7--,,, , • i ...... I ...... i ...... i ...... I ...... i ...... I 23 30 7 14 21 28 4 II 18 25 I 8 15 JUNE JULY AUG SE PT 1980 TIME ( E. S.T.)

Fig. 8. Low-passedsalinity valuesat (top) section1, (middle) section4, and (bottom) section6 from model results. charge occurred much more rapidly for the velocity regime Pac. Mar. Sci. Rep. 77-10, Inst. of Ocean Sci., Patricia Bay, B.C., Canada, 1977. than for salinity, with salinity intrusion occurring over ap- Foreman, M. G. G., Manual for tidal currents analysis and predic- proximately 1 month. Changes to the velocity regime were tion, Pac. Mar. Sci. Rep. 78-6, Inst. of Ocean Sci., Patricia Bay, accomplished within a week. In regard to tidal velocities, a B.C., Canada, 1978. significantincrease of the M2 and M,• constituentswas ob- Fread, D. L., and G. F. Smith, Calibration technique for 1-D un- steady flow models, J. Hydraul. Div. Am. Soc. Civ. Eng., 104, 1027- served after the cutoff. Conditions in the near offshore of 1044, 1978. James Bay showed a marked increaseof salinity in the week Godin, G., The Analysis of Tides, 264 pp., University of Toronto following the diversionas the river plume collapsed,with the Press, Ont., Canada, 1972. velocity regime following the coastal circulation rather than Hansen, D. V., and M. Rattray, New dimensionsin estuarine classifi- the divergentmotion usually associatedwithin a plume. In its cation, Limnol. Oceanogr.,11, 319-326, 1966. Harleman, D. R. F., and G. Abraham, One-dimensional analysis of present state the estuary is highly variable: both velocity and salinity intrusion in the Rotterdam Waterway, Publ. 44, Delft Hy- salinity fields respond rapidly to changes in wind and tidal draul. Lab., Netherlands, 1966. forcing. Harleman, D. R. F., and M. L. Thatcher, Longitudinal dispersionand Modeling of the Eastmain estuary during the major transi- unsteadysalinity intrusion in estuaries,Houille Blanche,29, 25-33, 1974. tion has demonstratedthe capability of the model to follow Ingram, R. G., Mean and tidal circulation of the Eastmain River satisfactorily the observed salinity changes.Overall, the com- (James Bay), Nat. Can., 109, 773-743, 1982. puted values of salinity, velocity, and water elevation were in Ippen, A. T., Estuary and Coastline Hydrodynamics,McGraw Hill, good agreement with the actual observations. The critical New York, 1966. period following the river cutoff (July 10-27) showed that the Kjerfve, B., The Santee-Cooper:A study of estuarine manipulations, model responseto sharp fluctuations was acceptable.During in Estuarine Processes,vol. 1, edited by M. Wiley, pp. 44-56, Aca- demic, Orlando, Fla., 1976. the model runs the river discharge values introduced ranged Kjerfve, B., and J. E. Greer, Hydrography of the Santee River during between 1245 and 55 m 3 s-1. moderate discharge conditions, Estuaries, 1, 111-119, 1978. Numerical diffusion was found to account for only a 15% Lepage, S., Salt intrusion and circulation changes in the Eastmain addition to the calculated dispersioncoefficients and might River estuary, James Bay, subsequentto a large reduction of the fresh water discharge, Master's thesis, Inst. of Oceanogr., McGill explain the higher than observedupstream salinities.A more Univ., Montreal, Que., Canada, 1984. realistic computation would have been possible had actual MeAlice, B. J., and G. B. Jaeger,Jr., Circulation changesin the Sheep- wind values at different locations been introduced. scot River estuary, Maine, following removal of a causeway, Es- tuaries, 6, 190-199, 1983. Prinsenberg,S. J., Man-made changesin the fresh water input rates of Acknowledgments.The authors thank all those people who as- Hudson and James bays, Can. J. Fish. Aquat. Sci., 37, 1101-1110, sisted in this research program: J.-C. Deguise (McGill Institute of 1980. Oceanography) in regard to computing and S. de Margerie and M. Thatcher, M. L., and D. R. F. Harleman, A mathematical model for Huot for assistancein the field. This study would have not been possiblewithout the financial support of the Soci6t6d'Energie de la the prediction of unsteady salinity intrusion in estuaries,Parsons baie James, the Groupe interuniversitaire de recherches Lab. Tech. Rep. 144, Mass. Inst. of Technol.,Cambridge, 1972a. oc6anographiquesdu , and the Natural Sciencesand Research Thatcher, M. L., and D. R. F. Harleman, Prediction of unsteady Council of Canada (R.G.I.). salinity intrusion in estuaries: Mathematical model and user's manual, ParsonsLab. Tech. Rep. 159, Mass. Inst. of Technol., Cam- briitge,1972b. Thatcher, M. L., and D. R. F. Harleman, Long-term salinity calcula- REFERENCES tion in Delaware Estuary, J. Environ. Eng. Div. Am. Soc. Civ. Eng. EEl, 11-27, 1981. Davidson, B., R. Vichnevetsky,and H. T. Wang, Numerical tech- niques for estimating best distributed Manning's roughnesscoef- R. G. Ingram and S. Lepage, Institute of Oceanography, McGill ficients for open estuarial river systems,Water Resour. Res., 14, University, 3620 University Street, Montreal, Quebec, Canada H3A 777-789, 1978. 2B2. Fischer, H. B., et al., Mixing in Inland and Coastal Waters, Academic, Orlando, Fla., 1979. (Received May 17, 1985; Foreman, M. G. G., Manual for tidal heightsanalysis and prediction, acceptedSeptember 9, 1985.)