WATER RESOURCES RESEARCH,

VOL. 20, NO. 2, PAGES 291-296, FEBRUARY

1984

Landfill Leachate Migration Through Shallow Unconfined Aquifers

DAVID W. OSTENDORF, RICHARD R. Noss, AND DAVID O. LEDERER
Department of Civil Enlineerin1, Universityof Massachusetts

We modelthe transportof a simplyreactive contaminant througha landfilland initially pure,underlying, shallow,one-dimensional unconfined aquifer with a plane, slopingbottom under steadyhydraulic conditions in the assumed absence of dispersion and downgradient dilution.The userpopulationand a presumedly constantcontaminantloading factor determinethe pollution input to the groundwater system, and we modelthe near field response as a singlelinear reservoir whoseoutput comprises the far field sourceterm at the downgradientedge of the landfill. The far field analysisyields a method of characteristics solutionvalid in the vicinity of the sourcelocation with frame speeds modifiedby recharge, head loss,bottom slope,and linear adsorption, and concentrations reflecting first-orderreaction kinetics. We calibrateand test the near and far field modelsagainstconservative chlorideand first-order reactivebicarbonatedata at the Babylon, New York landfill with accurateand physicallyplausible
results.

INTRODUCTION

We model the transport of a simply reactive contaminant through a landfill and initially pure, underlying,shallow,onedimensional,unconfinedaquifer with a plane, slopingbottom under steady hydraulic conditions in the assumedabsenceof dispersionand downgradientdilution. The resultingquantitative appreciation of the physical transport mechanismsand time scalesassociatedwith unconfinedaquifer pollution enablesus to identify the sourcehistory of existingplumesand predict trajectoriesof future contamination.This understanding is prerequisite for the assessment of the emergingevidence of subsurface water pollution downgradientof existinglandfills [Garland and Mosher, 1975] and the proper design and operationof the futurefacilities necessitated by ongoingwaste generation. We follow surfacewater quality modelers [Fischer et al., 1979] by distinguishing a near field region under the landfill, where mixing of leachate and groundwater is presumedto occur, and a far field region of fully mixed, one-dimensional flow downgradientof the landfill in an attempt to relate subsurfacecontaminantmigration to surfaceapplicationof solid waste.We describethe near field with an initially pure linear reservoirthat convertsleachateinput to a contaminatedbaseflow output which comprises the far field sourceterm at the downgradientend of the facility. The lumped parameter near field approach follows Gelhat and Wilson [1974] and Mercardo [1976], who successfully describe regionalpollution due to distributed inputs with the linear reservoir analysis.We completethe near field model by relating the input to the per capita contaminant generationrate and solve the case of a linearly increasing userpopulation,althoughthe near field-far field decomposition accommodates other input variation with the convolution integral responsecommon to linear systems analysis[Dooge, 1973]. We distinguishanalytical and numericaldescriptions of far field subsurface contaminanttransport in the literature. The numerical modelers [Bachmat et al., 1980] retain all terms in the conservation equationsgoverningthe process by simulating differential equationsand boundary and initial conditions

with numerical equivalentsover a temporal and spatial grid; the resultingmodelsproperly representphysicsat the expense of site specific computer programs with attendant documentation requirements. An analyticalapproach,whereproblem geometrypermits,solvessimplifieddifferentialequations, and boundary and initial conditions explicitly, obviates the computer, and yields a flexible, simple, and physically valid model appropriate in a preliminary planning and assessment context. We may cite several analytical studiesof advectivedispersive transportof a contaminantinjectedinto a uniform flow field on a spatially or temporally discontinuousbasis: Lenau [1972] postulates a steadystate,conservative injection from a recharge well, while Wilson and Miller [1978, 1979] consider unsteady pollution due to a constant vertical line source of reactive contaminant. Bear [1979] summarizesunsteady contaminant migration due to a series of onedimensional reactive source conditions, and Prakash [1982] modelssteadystate reactivepollution in three dimensions due to point, line, and volume sources. We suggest that the continuous, spatially distributed contamination generated by landfill leachateyieldssmall gradientsand consequent dominance of dispersionby the advectiveand reactive transport mechanisms. The resultingneglectof dispersion permits us to considera simply varying, one-dimensionalflow field in the vicinity of the sourcelocation by usingthe method of characteristics; as in the work of Wilson and Miller [1978, 1979], Bear [1979], and Prakash [1982], the reactivecontaminants under study are assumed to be linearly adsorptiveor exhibit first-order decay in the far field. We note that Wilson and Gelhat [1981] and Bredehoeft and Pinder [1973] useanalytical
and numerical versions of the method of characteristics to

describecontaminant migration in the unsaturated and saturated zones,respectively.

FAR FIELD GOVERNINGEQUATIONS

The steadyconservation of water massin a one dimensional, unconfinedaquifer subjectto constantrechargee is simply
q--qs + ex (1)

Copyright1984by the American Geophysical Union.

Paper number 3W1936. 0043-1397/84/003W1936505.00
291

with horizontal distance x downstream of the pollutant source, where conditions are denoted by an s subscript,as indicated in Figure 1. The dischargeq per unit width and average linear velocityv are relatedby definition

v = /nn

(2)

292

OSTENDORF ET AL.' LANDFILL LEACHATEMIGRATION

fluids to be immiscible, so that the contaminateddischarge q' equalsits sourcevalue,i.e.,

d =0
' NEAR ' FAR
FIE LD

q'= qs

(10)

FIELD
SOURCE

and the plume is overlainby a freshwater lensas suggested by Figure 1. A common pressuregradient drl/dx drives the lens and plume at the samevelocityhowever,so that the contaminatedaquiferthickness h' will be givenby
h' = qs/nV (11)

x
Fig. 1. Definition sketch.

We postulate a first-orderreactivecontaminant with decay constant ,[ subject to linear adsorption with corresponding retardationfactor R givenby Freezeand Cherry [1979]

R= 1+ K p-n

(12)

with porosityn and aquiferthickness h givenby

where p is the solid massper unit solid matrix volume and K is the distributioncoefficient characterizing linear adsorption.

h = hs+ x tan fi - r/

(3)

Equation (3) reflectsthe assumptionof a plane, sloping,underlyingaquiclude of smallangle to the horizontal;r/is the water table elevation below its source position, and its horizontal gradientobeysDarcy'slaw when is small:
vkg drl
vn dx

The longitudinalcoefficient of hydrodynamic dispersion DL is the productof the aquiferlongitudinaldispersivity and the averagelinear velocity[Freeze and Cherry, 1979]
D. = OtLV (13)

so that, in view of (2) and (9), dispersion is negligible when

Xc>> (14)

(4)

with gravitational acceleration g, permeability k, and fluid viscosityv. We collect(1), (2), and (4) and discover
q
h

where xc is the distanceover which concentrationvariesappreciably.When (14) is satisfied, (9)-(11) yield a simpleadvection equation describingcontaminant migration in the far
field, valid for nonuniform, immiscible flow
0c v 0c - ;c

kg d/
v dx

(5)

which, in the vicinity of the far field source,yields the firstorder relation

+ a ax- a
which is subjectto specified sourceconditions
c = Cs x= 0 (16)

qsVX

rt-- khs

x <<

hs

tan 0' e

qs --

(6)

In view of (1), (3), and (6), we may approximate the average linear velocityby modifying(2) with the result

CONTAMINANT CHARACTERISTICS

V=Vs 1+7

(7)

We solve (15) and (16) with the method of characteristics [aleson,1970] which restson the chain rule
dc cc cc dx

where the second-orderfactor 7 reflectsrecharge,head loss, and bottom slopeeffects ehs qv

dt

ct

+ -- --

cx dt

(17)

7=q+k-- tan
Next we consider the conservation of contaminant

(8)
mass for

wheredc/dt represents temporal changein a frame of reference moving at speeddx/dt. Equations(7), (15), and (17) yield the frame speed

steady hydraulicsin the far field as a balance of retarded dt R storagechange,advection,recharge,reaction,and dispersion [Pinder, 1973;Freezeand Cherry, 1979] in which contaminantconcentration obeys

dx Vs 1+7
dc
dt

(18)
(19)

n h 'R-[+ xx (q 'c )- d c =- n h ', c+ nxxh 'D Lxx (9)

with time t and rechargeconcentrationc. We follow Wilson and Miller [1978, 1979] by postulating a shallow aquifer so that c representsdepth averaged plume concentrationand mild differentialdensityeffects, which tend to establish vertical concentrationgradientswithin the plume, are ignored. Field measurements[Kimmel and Braids, 1980] do suggestthat gross differential density separatesthe plume from lighter downgradientrecharged, and we consequently assumethe

c9( c9c)

2c

The paths, or characteristics, of the moving framesfollow upon integrationof (18) from a startingtime t at the source x = 0 to any subsequent place x and time t in the far field, subject to the constraintsof (6). Recalling the smallness of Vx/hs, we have

__ i) s t fi(1-Y)dZ f,d

(20)

OSTENDORF ET AL.' LANDFILL LEACHATE MIGRATION

293

so that the contaminant characteristics in the vicinity of the sourceare approximatedby

TABLE 1. Population SegmentParameters,Babylon Landfill

t -- ts = -- 1 -V s

x <<,tan fi e

(21)
tsi,s x 10s G,cap/sx 10- ' Pi, capx 10 ' ti, s x 10 s ci,kg/m 3 x 106 Csi , kg/m 3 x 106

The frame starting at t = 0 marks the arrival %, t of the contaminantplume; (21) suggests that
t -1(22)

0 1.06 5.44 5.13 5.25S 0

4.10 7.11 9.79 -2.72 - 18.7S 6.93S

5.68 3.05 21.0 1.21 3.53S 11.6S

Equation(19) indicates that the reactivenature and source strength of the contaminant determine its temporalbehavior betweensegments of linear growth in the moving frame; we integrate from Cs,ts starting con- Csi ditionsto any subsequent c, t with the result Cs = 0 ts = 0

(28a)
(28b)

This nonhomogeneous, linear, first-orderordinary differential The far field sourcestrengthand starting time also represent equationwith constantcoefficients has the solution[Rainville near field output and accordingly link our two model reons and Bedlent, 1969] together, asindicatedschematically by Figure 1. .... + 0 < ts < ts
NEAR FIELD SYSTEM

c= Cs exp (ts - t)

(23)

C s = Csi

ts = tsi

c s c,{(1 '--"1-' exp( t)]t}

t/L

(29a)

The landfill constitutes a distributedinput of width b and

length in the directionof groundwater flow to an initially Cs= ci 1 + pure,linear reservoir whoseoutput comprises Cs. The conservation of contaminantmassper unit width is givenby

ti

k, Ci

- 1

dcs SP nhsR +qsCsb

(24)

exp [(tsits)/t] } tsi <t s <tsi + (29b)

Cs = Csd exp [(tsd -- ts)/t] ts < ts (29c)
BABYLON LEACHATE PLUME

where,in keepingwith the simplenear field system approach, we assume that the pollution input is simply related to user populationP by a constantcontaminantloading factor per capita S. The loading factor constancy reflectspresumedly rapid contaminant generation due to precipitation and solid waste interaction, in contrast to the slower time scale governingcontaminant decay. This decaytime scale mustbe comparableto the far fieldtime scale if concentrations are appreciable in the far field; the first-order reactions are therefore

with sourceconcentration Csa at the time of shutdowntsa.We illustrate near and far field calculationswith a casestudy.

Our case study is the well-documentedleachate plume downstreamof the sanitary landfill in Babylon, New York. The contaminantflows into the upper glacial aquifer in southern Long Island, an unconfinedaquifer of porosity n = 0.27

and permeability k = 6.34x 10- TM m2 underlain by the rela-

negligibly slowin the fastflow fieldunderthe landfill,and we

tively imperviousGardiners Clay and Magothy formations, consequentlyset ; = 0 in the near field. We also assume which intersectthe water table a distance6300 m upstreamof the landfill [Collins et al., 1972] with a local bottom slope linked linear segments of userpopulation,i.e., tan fi = 0.0027 [Kimmel and Braids, 1975, 1980]. Kimmel and P = Pi + Gi(ts-- tsi) tsi< ts< tsi+ (25) Braids[1980] presentwater table elevations under the landfill, and we accordingly estimate h s = 22.5 m and (dl/dx)s = with populationPi at time tsiand growthrate Gi valid for the

0.00161;(2)and (4) thenyieldVs = 3.37x 10-6 m/s and qs= ith segment of time. We combine(24) and (25) and derivethe near field response 2.05 x 10-5 m2/s,wherev = 1.1 x 10-6 m2/s.We consider the upstreamrechargearea in light of this discharge and estiequation
mate e = 3.25 x 10-9 m/s so that (8) yields 7 =0.00248.
Kimrneland Braids [1980] suggest that the landfill openedin
tc tc 1 dt-- +--=+

(26)1947 and serveda 1973 population of 287,000; the userpopulation data of Table 1 reflect this value and census[U.S. De-

with landfill response time tc and segmentconcentration and time ti obeying

tc = R/vs ci =
S(P i - Gitsi) bqs
Pi

(27a) (27b)

partmentof Commerce, 1977] figuresfor Suffolk County. The anomaloussurge of growth in the early 1960'scited by the table forces us to consider three growth segmentsfor the Babylon area. Kimrnel and Braids [1975] also specifylandfill
dimensionsb = 505 m and = 689 m. Kimrneland Braids [1980] suggest that the chloride ion is conservative in the relatively pure upper glacial aquifer and

ti= - tsi

presentthe 1974 depth averagedconcentrations Cm presented (27c) in Table 2, so that we specifyt = 8.52 x 108 s, R = 1 and

We solve(26) subjectto pure initial and matching conditions J,= 0. Wells 1-3 and 8 lie at the downstreamboundary of the

294

OSTENDORF ET AL.' LANDFILL LEACHATE MIGRATION

TABLE

2.

Observed Chloride Concentration and Error, From Kimmel and Braids [1980]

TABLE

4.

Observed Bicarbonate Concentration and Error, From

Kimrneland Braids [1980]

x,

cm,

ts,

Cs,

(5,

x,

Cm,

ts,

Cs,

c5,

Well
127 6 10 12 124 118 122 35 29

m
360 900 920 1570 1580 2180 2230 2810 3190

kg/m3
0.245 0.190 0.170 0.175 0.058 0.055 0.048 0.057 0.044

s x 108
7.47 5.98 5.93 4.26 4.24 2.83 2.71 1.47 0.71

kg/m3
0.256 0.180 0.176 0.100 0.100 0.073 0.071 0.044 0.023

%
5 -6

i
3 3 3 2 2 1 1 1 1

Well
127 128 6 10 124 118 122 35 29 " '1973

m
360 630 900 920 1580 2180 2230 2810 3190 3320 Data.

kg/m3
0.540 0.277 0.665 0.154 0.158 0.086 0.138' 0.054 0.020 0.023

s x 108
7.47 6.72 5.98 5.93 4.24 2.83 2.19 1.47 0.71 0.47

kg/m3
0.487 0.416 0.341 0.336 0.190 0.140 0.115 0.083 0.044 0.030

%
- 16 33 -57 83 - 10 10 -44 -4 31 -24

i
3 3 3 3 2 1 1 1 1 1

4 -43 72 34 49 -23 -47

Babylon landfill and, as suggested by Table 3, their average value accordingly yields a source concentration Cs--0.305

kg/m 3 at time ts-- 8.52x 108s. We consult (27)-(29)to comgradientcc'/cy through the voidsat the pute tc= 2.04 x 108 s and the ti, ci, and Csvaluescited verticalconcentration in Table 1; the chlorideloadingfactor So = 1.40x 10-8 freshwaterinterface so that, equating this transport to the
kg/cap-s (where "cap-s" denotes"capita seconds") calibrates the near field model with the 1974 data point. Table 2 also listsfar field model error 6 definedby
reactionterm in (9), we deduce

Oc' - nh'2c = - nDr interface (32)

Equation (32) yieldsa rough decayconstantestimate

6 c- c,
Cm

(30)

We predict far field concentrationat x, t by identifyingthe startingtime ts of the frame of reference occupying x, t from (21); the ts value also identifiesthe appropriate population segment constants in Table 1. The startingstrengthCs follows from (29) and far field concentrationC(Cs, rs) from (23). We

hs 2

(33)

where c/hs characterizesthe vertical concentrationgradient and the transverse dispersivityr is analogousto the longitudinal parameterin (13). Bear [1979] discusses a similar apassess model accuracy with the mean error cand standard proach describingdiffusivecontaminant transport into deaddeviationa definedby [Benjaminand Cornell,1970] end pore space.

$=1 6
J

(31a) at ts= 8.52x 108s so that the bicarbonate loading factor,in

Wells1, 3, and 114specify a source valueCs = 0.580kg/m 3

viewof Table 1 and (29),is Sco3 = 2.66x 10 -8 kg/cap-s.

Table 4 indicates that the calibrated decayconstant 2co3= 6.7 x 10-xo s-x zerosthe meanfar fielderror,with starting

conditionsCs,ts computedin accordance with the previous chloride procedureand 1974 predictedconcentrations reflecindicatea goodfit of far field modeland data and accordingly ting first-order decay in the moving frame, as suggested by endorsethe conservative contaminanttransportanalysis. (23).The errorstandard deviation aco= 39% is particularly A similar conservativeanalysisof the Kimrnel and Braids encouraging in view of the possible samplingerrorsin wells6, [1980] bicarbonate data of Table 4 yields a consistently 10, and 122. strongfar field overprediction, and we accordingly look to a The calibrated bicarbonate decay constant and postulated simple account of the reactive nature of the bicarbonateion. mechanismof (33) yield a transversedispersivityestimate Kimrneland Braids [1980] suggest that high gaseous carbon 0r = 0.10 m for the upper glacial aquifer. This value, while dioxide pressureat the landfill generateshigh bicarbonate considerablyless than Pinder's [1973] numerical model estileachateconcentrations; we postulate falling carbon dioxide mate of 4.2 m obtained from upper glacial aquifer contamipressures downgradientof the landfill due to transverse dis- nation in nearby Nassau County, is in keepingwith Kimrnel persivetransportto the freshwater lens,wherepressures equil- and Braids' [1980] reported concentrationcontoursat Babyibrate with low atmospheric values. A transverse hy- lon, which retain their original source widths.The discrepancy drodynamicdispersioncoefficientDr sendsthe gas down a may in part reflect the presenceof silt lensesin the upper glacial aquifer at Pinder's [1973] site, which would act to increase dispersion[Freeze and Cherry, 1979]. Pinder [1973] TABLE 3. SourceData, From Kimmeland Braids[1980] also estimates longitudinal dispersivityto be equal to 20 m at Well co, kg/m 3 CHCO3, kg/m 3 Nassau,and we satisfy(14)'s neglectof dispersion even with his high parametervalue. 1 0.625 0.898

The meanerror Cl= 5% and standard deviation ac= 38%

2 3 8 114 '1973 concentration 0.342.

0.054 0.385 0.157 -"

"' 0.470 ". 0.371'

CONCLUSIONS

We model near and far field migration of a simply reactive contaminantthrough a landfill and initially pure, underlying, shallow, one-dimensional,unconfined aquifer with a plane,

OSTENDORFET AL.: LANDFILL LEACHATE MIGRATION

295

sloping bottom under steady hydraulic conditions in the as rechargevelocity,m/s. sumed absenceof dispersionand downgradientdilution. The landfill length, m. near field is construed as a single linear reservoir with an r/ water table elevationbelowsourceposition,m. input governed by a constant contaminant loading factor re2 decayconstant,1/s. flecting per capita waste generationand an output specifying v waterkinematic viscosity, m2/s. the depth averaged source term for the far field model. We p solidmatrixmass perunitvolume, kg/m 3. a standard deviation. analyze the advective transport in this latter region in the vicinity of the source with the method of characteristics and Subscripts discover that recharge,head loss, bottom slope, and linear adsorptionmodifythe reference framespeeds, while first-order a ari'ivaltrajectory. reactionsinfluencepollutant concentration in the frames.Obc characteristicquantity. served 1974 chloride and bicarbonate concentrations at the C1 chloride property. d shutdown condition. Babylon, New York landfill calibrate the near field linear reservoir model with contaminant loading factors Sc= HCO3 bicarbonateproperty. 1.40x 10 -8 kg/cap-s and SHco3 = 2.66x 10 -8 kg/cap-s, rei populationgrowth segment condition. spectively. We test the far field advectivetransport model L longitudinaltransportproperty. against downgradient conservative chloride measurements m measuredquantity. with a mean error of 5% and standard deviation of 38%, s conditions at far field source. indicativeof good accuracy. A postulatedfirst-orderbicarbonT transverse transport property. ate reaction zeros the far field model error with a decay con rechargeconditions.

stant/Hco3 = 6.7x 10-xo s-x,which implies a transverse displume behavior'; the 39%bica-bonate errorstandard devi-

persivity r = 0.10 m consistent with observednondispersive Superscript

contaminateddischarge property. ation suggests reasonablemodel accuracy as well. Future researchmay proceedon severalfronts. We suggest Acknowledgments.The Commonwealth of Massachusetts Division that the contaminantloading factor include temporal varia- of Water Pollution Control funded this research under grant MDWPC 80-32; we acknowledgeand appreciatetheir support. bility representative of precipitation-solidwasteinteractionat the landfill surface, the near field parameterestimationreflect REFERENCES a distributedparameter analysisof contaminationunder the
landfill, and the far field model accommodate more realistic

reactions. Thesemodel improvements must considerthe long time scale of the subsurface environmentand preservethe simplicity of the presentapproach; in the latter regard, the near field-far field decompositiondecouplesthe improvements as well, facilitating separateanalytical investigations. Additionally, the calibrated loading factors and decay constant representing Babylonconditions shouldbe comparedto comparableplumes in geologicallysimilar aquifersas data
becomes available.

Bachmat,Y., J. Bredehoeft, B. Andrews,D. Holtz, and S. Sebastian, GroundwaterManagement:The Use of Numerical Models, Water Resour.Monogr. Ser.,vol. 5, AGU, Washington,D.C., 1980. Bear,J., Hydraulics of Groundwater, pp. 263-276, McGraw-Hill, New
York, 1979.

NOTATION

Benjamin,J. R., and C. A. Cornell, Probability,Statisticsand Decision for Civil Engineers, p. 12, McGraw-Hill, New York, 1970. Bredehoeft,J. D., and G. F. Pinder, Mass transport in flowing groundwater, Water Resour.Res.,9(1), 194-210, 1973. Collins, M. A., L. W. Gelhar, and J. L. Wilson, Hele-Shaw model of Long Island aquifer system,J. Hydraul. Div. Am. Soc. Civ. Eng., 98(HY9), 1701-1714, 1972. Dooge,J. C. I., Linear theory of hydrologicsystems, Tech.Bull. 1468, pp. 20-25, U.S. Dep. of Agric.,Washington,D.C., 1973. Eagleson, P.S., DynamicHydrology,pp. 337-344, McGraw-Hill, New
York, 1970.

landfill width, m.

contaminant concentration, kg/m 3. dispersion coefficient, m2/s.

populationgrowth rate, cap/s.

Fischer,H. B., E. J. List, R. C. Y. Koh, J. Imberger, and N.H. Brooks, Mixing in Inland and Coastal Waters, pp. 10-11, Academic,New
York, 1979.

Freeze,R. A., and J. A. Cherry,Groundwater, pp. 388-408, PrenticeHall, Englewood Cliffs, N.J., 1979. Garland, G. A., and D.C. Mosher, Leachate effectsfrom improper land disposal,J. Wasteage,6, 42-48, 1975. Gelhar, L. W., and J. L. Wilson, Groundwater quality modeling,

gravitational acceleration, m/s 2.

aquiferthickness, m.

distribution coefficient, m3/kg. permeability, m.

porosity. userpopulation,cap.

Groundwater, 2(6),399-408,1974.
Kimmel, G. E., and O. C. Braids, Preliminary findingsof a leachate study on two landfills in Suffolk County, New York, J. Res. U.S. Geol Surv.,3(3), 273-280, 1975. Kimmel, G. E., and O. C. Braids, Leachate plumes in groundwater

discharge of waterper unit aquifer width,m/s.

retardation factor.

fromBabylon and Islip landfills, LongIsland, New York, U.S.

Geol.Surv.Prof Pap. 1085, 1980. Lenau, C. W., Dispersion from rechargewell, J. Eng. Mech. Div. Am.

contaminantloadingfactor,kg/s cap.
time, s. averagelinear velocity,m/s.

far field horizontal distance downstream of pollutant

source, m.

Soc.Civ. Eng.,98(EM2), 331-344, 1972, Mercardo, A., Nitrate and chloride pollution of aquifers:A regional study with the aid of a singlecell model, Water Resour.Res., 2(4),
731-747, 1976.

vertical distance,m. dispersivity, m. underlyingaquiclude slopeangle. velocitymodificationfactor.

error.

Pinder, G. F., A Galerkin-finite element simulation of groundwater contamination on Long Island, New York, Water Resour. Res., 9(6), 1657-1669, 1973. Prakash, A., Groundwater contaminationdue to vanishingand finite size continuous sources, J. Hydraul. Div. Am. Soc. Civ. Eng., 108(HY4), 572-590, 1982.

296

OSTENDORF ET AL.: LANDFILL LEACHATE MIGRATION

Rainville,E. D., and P. E. Bedient,Elementary DifferentialEquations, Wilson, J. L., and P. J. Miller, Two-dimensionalplume in uniform pp. 110-125, MacMillan, New York, 1969. groundwater flow-closure,J. Hydraul. Div. Am. Soc. Civ. Eng., U.S. Department of Commerce,County and City Data Book, p. 330, 105(HY12), 1567-1570, 1979. Bureau of the Census,Washington,D.C., 1977. Wilson, J. L., and L. W. Gelhar, Analysisof longitudinaldispersion in D. O. Lederer, R. R. Noss, and D. W. Ostendorf,Civil Engineering unsaturatedflow, 1, The analytical method, Water Resour.Res., Department, University of Massachusetts, Amherst,MA 01003. 17(0, 122-130, 1981. Wilson, J. L., and P. J. Miller, Two-dimensional plume in uniform (Received July 12, 1983; revisedNovember 22, 1983; groundwaterflow, J. Hydraul. Div. Am. Soc. Civ. Eng., 104(HY4), 503-514, 1978. accepted December6, 1983.)