A CHARACTERISTIC SOLUTION to NITRATE TRANSPORT and FATE in GROUND
WATER in AGRICULTURAL WATERSHEDS
Mohamed M. Hantush,1 Associate Member, ASCE and Miguel A. Marino,2 Honorary Member,
ASCE
Abstract
*
Ground-water contamination by nonpoint source pollutants is recognized as a major
environmental threat to human health and the sustainability of healthy ecosystems. In this paper,
we present advective-reactive models that describe transport and fate of pollutants in inclined
unconfined aquifers subject to a variable source. The models have implication on nitrate
transport and denitrification at depth in aquifers, and the impact this has on attenuating nitrate
discharge to streams in agricultural watersheds. The objectives of the models are: (1) to predict
the impact of geochemical and physical controls on baseflow loadings in agricultural watersheds;
and (2) to develop an index that measures the capacity for nitrate removal in ground water in
agricultural watersheds. First, the steady-state ground-water flow problem is solved in two-
layered unconfined aquifers with impervious inclined beds. Approximate closed-form
expressions are also developed for ground-water fluxes in a two-layer inclined unconfined
aquifer. Second, the transport and fate problem is formulated for the two-geochemically distinct
layers, and solved in a Lagrangian framework using a regular perturbation technique. The
solutions are applied to paired agricultural watersheds in the Mid-Atlantic coastal plain, and
predictions show that denitrification at depth, aquifer geometry, and historical variations of
nitrogen at the sources may be profoundly impacting nitrate baseflow loadings in two creeks in
the watersheds.
Introduction
Ground-water contamination by nitrate is a major source of concern to the public,
decision makers, and surface water and ground-water..resources managers. Eutraphication in the
Chesapeake Bay, the United States' largest and most productive estuary, is linked to elevated
levels of nitrate in its waters. Nitrate drains from agricultural lands to ground water and
subsequently discharges to streams, which contribute up to 80% of the fresh water to the
estuaries in the Bay. Riparian zones are natural filters for NO3" transported in ground water from
upper croplands to streams, and denitrification (microbially-mediated reduction of NO3" to N2O
and N2) is a major mechanism responsible for the removal of NO3 in these areas. Many studies
have addressed nitrogen removal in shallow organic-rich riparian environments (e.g., PeterJohn
and Correll, 1984; Cooper, 1990; Groffman et al., 1996; and Cey et al., 1999). Fewer studies,
however, investigated denitrification at depth in ground water (Kinzelbach et al, 1991; Korom,
1992; Bohlke and Denver, 1995; and Mengis et al., 1999). Figure 1 illustrates a hydrogeologic
setting in which ground-water flow patterns are influenced by the regional dip of the strata, and
possibly affecting NO3" baseflow to the two shown creeks. Such a scenario is not uncommon to
the mid-Atlantic coastal plain where paired agricultural watersheds show remarkably different
nitrate levels in their baseflows (Bohlke and Denver (1995). Concentrations of nitrate (NO3") in
1	Hydrologist, Subsurface Protection and Remediation Division, National Risk Management Research Laboratory,
ORD, U.S. EPA, 919 Kerr Research Dr., Ada, OK 74820. Email: hantush.mohamed@epa.gov (Corresponding
Author)
2	Professor, Department of Land, Air and Water Resources, and the Department of Civil & Environmental
Engineering, University of California, Davis, CA 95616. Email: mamarino@ucdavis.edu
1

-------
ground-water discharge to the Morgan Creek were reported to be significantly lower than in the
Chesterville Branch. A combination of denitrification in reducing subsurface sediments at the
Moman Creek SW
Aye (years) » altered
NO^" [rng/ll as N) ¦ 2-3
Agw (years) ¦ 3*50
NO3" (mg/L as N) ¦ 3-2C
Chftxrervfc Branch SW
Aga (years) ¦ altwrnd
NOj" (mgA. as N) ¦ 9-tO
South
¦ Morgan Cwafc GW
Afl» {years) *25-40
¦¦NOj* (mgfl. as N) m 0
Excasft N£ (nrg/l. aft N) ¦ W
bjawyed'-
Fig. 1 Cross section of two agricultural watersheds in the Mid-Atlantic coastal plain. The redox
zone is shallower below the Morgan Creek relative to its depth below the Chesterville
Branch, due to the regional dip of the strata (adapted from Bohlke and Denver, 1995).
base of the surficial aquifer (shaded in Fig. 1) and historical input variations may be responsible
for the observed behavior. The influence of the aquifer-bed dip on ground-water flow pattern
may be responsible for greater reduction of NO3' baseflow loading to the Morgan Creek, due to
the relatively shallower depth of the redox zone below the creek as Fig. 1 shows. In this paper,
we model nitrate transport and fate in unconfined aquifers with inclined beds, and attempt to
resolve the impact of relatively deep denitrification, aquifer geometry, and historical variations at
the source on nitrate baseflow loadings.
Ground-water Flow In Inclined Unconfined Aquifer
Figure 2 depicts a conceptual stream-aquifer system in which the aquifer bed is inclined
with a positive angle 0 measured relative to a horizontal plane at a reference elevation (or datum)
of aquifer bed at the right-hand-side boundary. Ground-water flow toward the stream occurs due
to a uniform recharge of magnitude N [m/d] and a prescribed influx Qn [m3/d] at the right-hand-
side boundary at x = 0 m. A water divide at x = 0 corresponds to the special case of Qn = 0.
Invoking Dupuit's assumption and considering the aquifer to be homogeneous, the steady-state
flow in a two-layer unconfined aquifer may be described by the following quation (Bear, 1972):
in which h{x) = hydraulic head at steady state [m]; £(*) = steady-state water-table elevation
relative to the aquifer bed, or saturated flow depth [m]; K\ = hydraulic conductivity of the upper
layer [m/d]; K2 - hydraulic conductivity of the lower layer [m/d]; (3 = tan 0 = gradient of the
(1)
where
A(j0 = C00 + P*
T(x) = Ktfix)
T!(x) = C(*) + [(K2/£,)-!] 6
(2)
(3a)
(3b)
2

-------
aquifer bed; b = thickness of the lower layer [m]; N = uniform recharge [m/d]; and x = horizontal
distance. Equation (1) is valid for sufficiently small (3 so that Dupuit's assumption of horizontal
flow in the aquifer is applicable. We seek a solution to (1) subject to the following boundary
conditions (Fig. 2):
o)^=a.
ax
0
(4a)
- K{ T|(/) = K'W k{l) , H ,	x — I	(4b)
dx	b
in which A" = hydraulic conductivity of the streambed [m/d]; b' = thickness of the streambed
deposits [m] (i.e., semipervious boundary layer); W = wetted perimeter at the ground water-
surface water interface [m], at x = /; and H = stream stage relative to the datum [m]. Equation
(4b) is based on the assumption that ground water is intercepted entirely by the stream and that
no horizontal flow occurs below the stream. Also, in 4(b) the hydraulic head along the wetted
perimeter (A-B) in Fig. 2 below the streambed at x = / is assumed to be equal to /?/.
N C* I
i = h(x)
¦ aT(C)=Ne*Qo i C(*)
impervious
Boundary
Datum
N") = ?(*) ~ 4M
Fig. 2 Schematic illustration of an inclined phreatic aquifer hydraulically connected to a
stream.
The solution to the above boundary-value problem (1-4) is given by (Hantush and Marino, 2000)
T|z + (P/iV)r|(/V x + g0) + (1 /A^iVXA' x + Oq)2 _
TV + (P / N) n, (/V / + O0) + (1 / KtN)(N I + Q0f
expi
where
1
yN
tan
-.fn/ /(M + g0) + p/(2iV)
(5)
- tan
-i
q(jc)/(Afr + g0) + p/(2AQ'
y
P2
K, N 4/Y
(6a)
3

-------
and
(6b)
in which ^ = water table elevation above the stream bed at x = / [m], Equation (4) requires that y
> 0. In the case of net abstraction (e.g., evapotranspirative demand exceeding infiltration), y < 0
and the solution may be given by Eq. 5, but with the function {- coth"'(#)} replacing tan"'(*).
A useful approximate closed-form solution to (5) is also obtained for sufficiently small P
« 1 and N * 0 using a straightforward expansion in the form of a power series in f3 (Hantush
and Marino, 2000):
in which <}>o(x), <}>i(x), etc., are to be identified. Equation (7) allows for a closed-form solution to
the transport problem, as will be discussed briefly later.
In this analysis we separate the saturated flow thickness of the aquifer into two
geochemically-distinct layers, an upper oxic and a lower redox zones, and use the Dupuit's
assumption to derive expressions for ground-water flux in the lower layer, Qi(x), of saturated
thickness b (Fig. 2), and the flux in the upper layer, Qu(x), of saturated thickness of £-b.
Figure 3 shows cross section A-A' which traverses the two agricultural subwatersheds,
Morgan Creek and the Chesterville Branch. The study area (referred to as Locust Grove) is
located in Kent County (Maryland), northeast of the Chesapeake Bay. It is composed of two
agricultural watersheds with 10% of the area occupied mainly by riparian wooded areas. The
Fig. 3. Morgan Creek and the Chesterville Branch subwatersheds showing the stream-
flow gage stations and cross section A-A'.
area is underlain by a relatively shallow surficial (Columbia-Aquia) aquifer, which consists of
sand and gravel of fluvial origin. The inclination of the surficial Columbia-Aquia aquifer bed at
^(*) = 0o(*) + P^(X) + P22(>) +	> P«l
(7)
^1493492^
' ' /'49^Se20>^t*33497
/ ivK ^1>93498
/y	r
l JkjVwsco \	J
A
0 1 2 Kiiorreters
4

-------
this particular cross-section was estimated to be (3 = 0.0052, - 0.0052 (0.3°) at the Morgan Creek
and the Chesterville Branch sides, respectively, and the average thickness of the redox layer at
the cross section is estimated to be b = 3.2 m. This information was synthesized from local
hydrostrategraphic information and the GMS model (Hantush and Cruz, 2000). An estimate to
the recharge in each subwatershed is obtained by baseflow separation at each drainage point,
using 1997-1999 Water Year data.	,
Figures 4(a) compares the exact (5) and approximate solution (7) with observed water-
table elevations along Morgan Creek transect length, based on the estimated recharge N = 0.71 x
10~3 m/d and calibrated values of K = 60 m/d and a = 2.0 d"1. Figure 4(b) compares exact and
approximate estimates of Qu(x) and Qfoc) at the Morgan Creek side. Figure 5 shows similar
results for the Chesterville Branch. The recharge is estimated to be N = 1.19 x 10"3 m/d and the
calibrated parameters are Ki = K2 = K = 8 m/d and a = 2.0 d"1. Hantush and Marino (2000)
applied a different approach in which they assumed a single K = 16.76 m/d for the two
subwatersheds (from Reilly et ai, 1994) and calibrated for the recharge to yield jV = 2.4 x 10"4
m/d at Morgan Creek and 2.0 x 10~3 m/d at the Chesterville Branch. The approach adopted here
assumes that annually-averaged baseflow discharge from each subwatershed is balanced by
ground-water recharge. Note that the baseflow at Morgan Creek is dominated by ground-water
flux from the lower (redox) zone (Fig. 4(b)), whereas at the Chesterville Branch, baseflow is
dominated by the discharge from the upper layer (Fig. 5(b)), This implies that denitrification at
depth may provide a greater control on nitrate loadings to Morgan Creek.
(a)	(b)
Morgan Creek
Water divide
N= 0.71 x 10 m/d
K - 60 m/d
cr = 2.0 cfl
b = 3,2 m
Qu{x) - approximate
- exact
£
O)
0.6
H
0.4
VJLx)
^ approximate
^exact
0.2
0.0
1000
1500
2000
2500
0
500
Morgan Creek
Water divide
25
Exact
20
Approximate
Observed head
15
10
Streambcd
elevation
5
o
o
2000
2500
500
toco
1500
x (m)	x (m)
Fig. 4 (a) Estimated versus observed heads at Morgan Creek transect and (b) estimated ground-
water fluxes in the upper and lower layers.
5

-------
(a)
Water divide
Chcstcrville
Branch
-o_
Observed Estimated
(exact & approx.)
N= 1.19 x 10~3 mid
K = 8 m/d
<7 = 2.0 d1
b = 3.2 m
(b)
Water divide
OJ
d
Chesterville
Branch
(exact & approx.)
0/to
(exact & approx.)
X (m)
X (m)
Fig. 5 (a) Estimated versus observed heads at the Chesterville Branch and (b) estimated ground-
water fluxes in the upper and lower layers.
Nitrate Transport and Fate
In this section, partial differential equations, which describe stratified advective-reactive
transport of nitrate, are developed using the concept of conservation of mass and then solved
analytically in a Lagrangian framework. The solutions, however, are based on two assumptions:
(1) the effect of dispersion is negligible; and (2) the lower layer above the confining unit is a
redox zone where ground waters are entirely denitrified.
Figure 6 shows a unit-width control volume of length Ajc and height ^(x), tapping the
aquifer at distance x. The upper layer is of thickness C,(x) - b and the lower layer is of thickness b.
N C*(l)
Q^x+Ax.tJC^x+Ax.t)
Q,(x + Ax, t) C ,(* 4-A*, t)
Reducing
Sediments

Fig. 6 Mass balance across an aquifer column of an infinitesimal area. The upper layer
depicts the oxic zone and the lower layer denotes the reducing sediments.
6

-------
Mass balance of a contaminant through the upper layer in the control volume, after taking the
limits Ax —» 0 and At —> 0, can be shown to yield (Hantush and Marino, 2000)
nuft-b)^ + Qu^ + {N + nu 0l -b)ku)cu=N C\x, t)	(8)
dt	ox
where b = (K21 Kx)b\ Cu = concentration of nitrate in the upper layer [g/m3]; nu = porosity of
upper layer; C\x,t) = concentration of nitrate at the recharge [g/ m3]; and ku = first-order rate loss
coefficient [d"1], which accounts for partial denitrification. In the above equation, we assumed
uniform concentration throughout thickness £(jc) - b. Equation (8) ignores dispersion and
concentration's variations across the layer thickness, and it is sufficient to describe nitrate
reduction baseflow assuming unlimited capacity for denitrification at depth (i.e., in the
subsurface reducing sediments). A closed-form characteristic (Lagrangian) solution to (8) can be
obtained using asymptotic expansions of the form (7) and for Cu(x,t) (Hantush and Marino,
2000),
C„(jr,O = V0(*,O + Vi(*,OP+ •••¦, P«1	(9)
in which integral expressions to vj/o(-V) and \J/](x,/) are obtained by direct integration. We report
only the solution to the characteristic time approximated to the second order:
N	2N
In
+ o(-r)
- In
+ (t)o(xo)
-o('ro)
J j
(10)
I
01 (m)
2o.+ Nu
¦du
, • 2 (N I + Qq)2
where a =r\, + -	
Kl N
The function t(x) may be interpreted as the average age of
ground-water mixture discharging at a given point in time across section x in the upper layer. It is
a function of Qo, N, and the geometric and hydraulic properties of the aquifer. Nete-Jthat=&e
impact of aq4iifer-bed inclination on-the-average age is approximated to thgJj4^oF<4er-hy-fh{rfer
(ffltegral)-term on-the-righyiand-side of Eq. (1-0). In particular, we focus on the integral effect of
denitrification at depth as a geochemical control, and the regional dip of the stratum as a physical
control on nitrate baseflow loadings. Because oxic waters dominate in the upper layer,
denitrification of NO3" is assumed to occur exclusively in the lower layer (redox zone). Hence,
averaged NO3' conccntraction C(x,t) at section x is given by C(x,t) = Qu(x)Clt(x,t) / QT(x),
where Qt{x) = N x +Q0. This represents the dilution effect caused by mixing with NCV-free
ground waters.
Figure 7 shows regressed record of nitrate NO3" (mg/L) loading to the water table based
on regionally averaged annual tonnage of fertilizer apportioned to Kent County (1945-1991).
Two scenarios are considered: (1) the average nitrate concentrations at the recharging waters C
after year 1992 remained constant at 17 mg/L (1992-2040), and (2) discontinuous application
after year 2000. Figure 8 depicts baseflow NO3" concentrations predicted by (9) and (10) at
Morgan Creek and the Chestcrville Branch at cross section A-A', for the two scenarios of C
shown in Fig. 7. Baseflow NO3" concentrations at Morgan Creek are reduced by approximately
7

-------
80% (by comparing with the maximum level of 17 mg/1 at the source), as near steady-state levels
indicate, whereas denitrification at depth resulted in about 15% reduction at the

C
o
c
4>
u
G
o
U
O
C (discont. source)
1940
1960
1980 2000
Year
Fig, 7 Historical variations of nitrate at the source, regressed from 1940-1991 records, and
assumed from 1992-2040 using continuous and discontinuous source scenarios.
Chesterville Branch. This may be explained on the basis ofaquifer bed inclination, as the former
resulted in a relatively shallower denitrification zone below Morgan Creek ((3 = 0.0052) as
opposed to the relatively deeper depth below the Chesterville Branch (P = - 0.0052). The
discontinuous-source scenario indicates that observed low baseflow nitrate levels at year 2040
due to historical variations of the source should not be confused with losses by denitrification.
to
c
o
c
o
CJ
c
o
U
O
z
Morgan Creek
P = - 0.0052 (hypothetical)
p = 0.0052 (actual)
Discont. source
Chesterville Branch
P = -0.0052 (actual)
12 ¦
Discont, source
// P = 0.0052
(hypothetical)
1960
Year	Year
Fig. 8 Predicted baseflow NOi" concentrations at both creeks due to the two scenarios of
historical variations at the source.
8

-------
Also, Fig. 7 indicates that the model would have predicted baseflow NO3" concentration of 13.75
mg/1 at Morgan Creek in year 2040 as opposed to 2.75 mg/1, assuming a hypothetically deeper
redox zone, (3 = - 0.0052. Bohlke and Denver (1995) reported values of zero mg/1 of NO3" at the
discharge at Morgan Creek and 3-5 mg/1 at the Chesterville Branch (see Fig. 1), and the location
where these values were measured is in close proximity of section A-A'. Our predictions at year
1995, based on the continuous source scenario, were 1.75 mg/1 and 14.5 mg/1 at ^lorgan Creek
and the Chesterville Branch, respectively. Figure 8 also shows that the predicted window of
baseflow concentrations at Morgan Creek ((3 = 0.0052) starts at year 1970. This means that
ground waters in a baseflow mixture at Morgan Creek in year 1970 are as old as 30 years (recall
that the historical source record starts at year 1940, as Fig. 7 indicates).
Nitrate Removal Capacity Index
We define the removal-capacity index as the ratio of the rate at which the constituent
mass is lost out of the system to the rate of input, at steady state, and is given by (Hantush and
Marino, 2000)
,	1 + n, (AT, / K2 )(k, / -b'K2IKx)
£ RC — [	^	lllj
[1 + nu(ku /N)(i\(x) - b)] [1 + n,(tf, / K2)(k, / N)t\, ]
in which k/ = first-order rate of denitrification in the lower layer [d"]; «/ = porosity of the lower
layer; and (T|-Z>)= average saturated thickness of the upper layer [m]. If r|i = b K2/K1 (i.e., £(/)
= b) and assuming that ground waters are instantly denitrified in the redox layer (k/ —> °°), we
have Irc =1. When ku = 0 and kt —» °°, I/?c = b (K2/K1)/ T|| , which may be applicable to the study
site. At Morgan Creek, assuming K\ = Kj, we have I/?c = 3.2/4.3 = 0.74, and at the Chesterville
Branch, I/?c = 3.2/24 = 0.13.
Conclusions
An analytical framework was presented, which describes ground-water hydraulics and
transport and fate of distributed-source pollutants in aquifers with inclined beds. The solutions
have implications on nitrate transport and denitrification at depth in agricultural watersheds. The
characteristic solutions to the transport problem were applied to paired agricultural watersheds in
the Mid-Atlantic coastal plain, where geochemical and physical controls, and historical
variations of nitrogen at the source may be responsible for observed differences in baseflow
nitrate loadings to two creeks. The models predicted the impact of aquifer bed inclination on the
fraction of ground-water fluxes through the redox zone, and the effect this and denitrification
might have on nitrate baseflow concentrations at each creek. The impact of historical variations
at the source were also examined with two scenarios of nitrate input. An index for nitrate
removal capacity was developed, and its utility in estimating the potential for nitrate removal at
the site by denitrification was demonstrated.
Acknowledgment'. This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for presentation and
publication. We thank Dr. J. Cruz for the baseflow separation calculations and production of Fig.
3. The research reported in this paper is part of Project 4940-H of the Agricultural Experiment
Station of the University of California, Davis.
9

-------
References
Bear J., 1972. Dynamics of Fluids in Porous Media. American Elsevier, New York,
Bohike. J.K. and J.M. Denver, 1995, "Combined use of groundwater dating, chemical, and
isotopic analyses to resolve the history and fate of nitrate contamination in two agricultural
watersheds, Atlantic Coastal Plain, Maryland." Water Resour. Res., 31, pp. 2319-2339.
Cey, E. E., D. L. Rudolph, R. Aravena, and G. Parkin, 1999. "Role of riparian zone in
controlling the distribution and fate of agricultural nitrogen near a small stream in southern
Ontario." J. Contam. Hydro I., 37, 45-67.
Cooper, A. B., 1990. "Nitrate depletion in the riparian zone and stream channel of a small
headwater catchment." Hydrobiologia, 202, 13-26.
Groffman, P. M, G. Howard, A. J. Gold, and W. M. Nelson, 1996. "Microbial Nitrate
Processing in shallow groundwater in a riparian forest." J. Environ. Qual. 25, 1309-1316.
Hantush, M. M., and J. Cruz. Geohydrologic Foundation In Support Of Ecosystem Restoration:
Modeling of BasefJow Loadings of Nitrate In Mid-Atlantic Coastal Plain, EPA/600/R-99/104.
In press.
Hantush, M. M. and M. A. Marino. 2000. Analytical modeling of variable-source nitrate
transport and fate in groundwater drainage from agricultural watersheds. Submitted to Water
Resour. Res.
Kinzelbach, W., W. Schafer, and J. Herzer. 1991. "Numerical modeling of natural and enhanced
denitrification processes in aquifers." Water Resour. Res. 27(6), 1123-1135.
Korom, S. F., 1992. "Natural Denitrification in the Saturated Zone: A Review. " Water Resour.
Res. 28(6), 1657-1668.
Mengis, M., S.L. Schiff, M. Harris, M. C. English, R. Aravena, R. J. Elgood, and A. Mac Lean.
1999. "Multiple geochemical and isotopic approaches for assessing groundwater NOj"
elimination in riparian zones." Ground Water, 37(3), 448-457.
PeterJohn, W, T., D. L. Correll. 1984. "Nutrient dynamics in an agricultural watershed:
Observations on the role of a riparian forest." Ecology. 65(5), 1466-1475.
Reilly, T. E,, L, N. Plummer, P. J. Phillips, and E. Busenberg, 1994. " The use of simulation and
multiple environment tracers to quantify groundwater flow in a shallow aquifer." Water
Resour. Res. 30(2), 421-433.
10

-------
NRMRL-ADA-00222
TECHNICAL REPORT DATA

1. REPORT NO.
EPA/600/A-01/053
2.

4. TITLE AND SUBTITLE
A CHARACTERISTIC SOLUTION to NITRATE TRANSPORT and FATE in GROUND
5. REPORT DATE


6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
'Hohamed H. Hantush
'Miguel A. Marino
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
"USEPA, ORD, NRMRL, SPRD, P.O. BOX 1198, ABA, OK 74820
3Dept of Land, Air, £ Hater Resources and Dept of Civil & Environ Eng
University of California, Davis, CA 9S616
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
In-House
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. EPA
NATIONAL RISK MANAGEMENT RESEARCH LABORATORY
SUBSURFACE PROTECTION AND REMEDIATION DIVISION
P.O. BOX 1198,* ADA, OK 74820
13. TYPE OF REPORT AND PERIOD COVERED
Symposium Paper
14. SPONSORING AGENCY CODE
EPA/600/15
15. SUPPLEMENTARY NOTES
PROJECT OFFICERS Mohamed M. Hantueh 580-436-8S31
16. ABSTRACT
Ground-water contamination by nonpoint source pollutants is recognized as a major environmental threat to human
health as well as the sustainability of healthy ecosystems. In this paper, we present advective-reactive
models that describe transport and fate of pollutants in inclined unconfined aquifers subject to a variable
source. The models have implication on nitrate transport and denitrification at depth in aquifers, and the
impact this has on attenuating nitrate discharge to streams in agricultural watersheds. The objectives of the
models ares 1) to predict the impact of geochemical and physical controls on baseflow loadings in agricultural
watersheds; and 2) to develop and index which measures the capacity for nitrate removal in ground water in
agricultural watersheds. First, the steady-state ground-water flow problem is solved in two-layers unconfined
aquifers with impervious inclined beds. Approximate closed-form expressions are also developed for ground-
water fluxes in a two-layer inclined unconfined aquifer. Second, the transport and fate problem is formulated
for the two-geochemically distinct layers, and solved in the Lagrangian viewpoint using a regular perturbation
technique. The solutions are applied to paired agricultural watersheds in the Mid-Atlantic coastal plain, and
predictions show that denitrification at depth, aquifer geometry, and historical variations of nitrogen at the
sources may be profoundly impacting nitrate baseflow loadings in two creeks in the watersheds.
17.
KEY WORDS AND DOCUMENT ANALYSIS
A. DESCRIPTORS
B. IDENTIFIERS/OPEN ENDED TERMS
C. COSATI FIELD, GROUP



18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS(THIS REPORT)
UNCLASSIFIED
21. NO. OF PAGES
10

20. SECURITY CLASS(THIS PAGE)
UNCLASSIFIED
22. PRICE
EPA FORM 2220-1	(REV.4-77) PREVIOUS EDITION IS OBSOLETE

-------