4.4 Groundwater Flow Modeling
4.4.1 Model Construction
The Groundwater Modeling System (GMS) 10.2.4,a commerciallyavailable com-
prehensivegraphical userenvironment for performing groundwater simulations, was
usedtocreateaconceptualmodeloflowerScottsCreekusingGIS,field,andlaboratory
data.
EnvironmentalSystemsResearchInstitute(ESRI)ArcMap10.4was usedtocreate
GIS data including the model boundary and locations of wells, piezometers, and
exploratory boreholes. All GIS data were assigned the NAD 1983 California State
Plane Zone III FIPS 0403 (meters) projected coordinate system. GIS data were
imported into GMS and converted to featureobjects. In GMS,feature objectswere
used to build coverages (group of feature objects) including a sources and sinks
coverage and a coverage for each alluvial layer. All coverage data were entered in
units of meters and days.
The sources and sinks coverage included the irrigation wells and the boundary
conditions of the model. The screened interval of each well was specified using
information obtained from well completion reports. The Archibald Well screened
intervalwas from 4to -23.5 m,the Pump House Well screened intervalwas from-11
to -25.3 m, the VFDWell screened interval wasfrom -9.8 to -28 m, and the Queseria
Well screened interval was from 0.5 to -27.5 m. The grid around all four wells was
refined and setto abase size of 15m,a bias of 1.1, and amaximum size of 120 m.
Three alluvial layer coverages were created, each representing one hydrostrati-
General-Head Boundary
General-Head Boundary No-Flow Boundary
River Boundary
Figure 4.52: Plan view of the three-dimensional (3D)finite-difference groundwater flow model (MODFLOW 2000) grid of lower Scotts Creek showing the boundary and initial conditions (starting head contours (m)) of the model.
Monitoring Well
Bedrock Aquiclude
Pumping Well
Piezomete
r
Stream
Middle Aquita
rd
Lower Semi-Conf
ined Aquifer
Potentiomet
ric Surf
ace
Upper Unconfined Aquifer
Figure 4.53: Sc hematic of conceptua l mo del of the groundw ater system on lo w er Scotts Creek.
Figure 4.54: Schematic of general-head boundary (excerpted from the online user- supported help database for XMS software).
Figure 4.55: Schematic of river boundary streambedconductance components for an individualcell (excerpted fromMcDonald and Harbaugh, 1988).
interpolated from a 1.5 meter digital elevation model (DEM) of the Scotts Creek
watershed. The elevationof the bottomof layer 1, and the topand bottom elevations
of layers 2 (aquitard) and 3 (lower semi-confined aquifer), were interpolated from
exploratory borehole data and information from well completion reports that were
entered into GMS as2D scatter points. Similarly, hydraulic conductivities of layers
1 and 2 were interpolated from laboratory permeameter test data. The hydraulic
conductivity and specific storage of layer 3 were set at constant values of 9.16 m/d
and3.5 × 10−5 m−1, respectively, average values from pumping testcurve-matching
analyses. The specific storageof layer 2was set at a constantvalue of3.5 × 10−4 m−1
(one order of magnitude greater than layer 3) and the specific yield of layer 1 was
set at a constant value of 0.3. All three layers were isotropic. Starting heads were
interpolated frommeasured water table elevation data recorded inAugust 2017 (see
Appendix I for 2D scatterdata). The interpolation scheme used throughout was the
inverse distanceweighted(IDW) methodusingthe constantnodalfunction(Shepard’s
method) method which is givenby the equation
n
X
F(x,y) = wifi, (4.19)
i=1
where n isthe number of points used to interpolate, fi arethedataset values atthe
points, andwi arethe weight functions assigned at each point calculatedaccording to
equation −p r wi = n , (4.20) P h−jp j=1
where p isan arbitrary positive realnumber called theweighting exponent(default
mostby measured valuesclosest tothe predictionlocation andless by more distant
points. Theconstant nodalfunctionform of the IDWmethod was usedbecause of its
simplicity,while still retaining allthe functionality needed for the application.
The model boundary was split into four distinct arcs and each was assigned a
boundarycondition. Thenorthernmostarc(Archibald Creek)was definedas ageneral
headboundary,theeasternmostarc(SwantonRoad)wasdefinedasano-flowboundary,
the southernmost arc (Queseria Creek) was definedasa generalhead boundary, and
the westernmost arc (Scotts Creek) was modeled as a river boundary. All of the
boundaries (exceptthe no-flow boundary) were assigned a conductance of 0.34 m2/d.
Additionally, head-stage wasdefined atthe upstream and downstream nodes of each
generalhead boundary arc, and head-stage and bottom elevationwere definedatsix
locations (upstream,downstream, and the four instream piezometer locations)along
the river boundary using observed instream piezometer data. Table 4.11 provides
a summary of the head-stage and bottom elevation of individual nodes along each
boundary arc.
Table 4.11: Head-stage and bottomelevation in meters of nodesalong the boundaries of the MODFLOW model. Dash (–) indicates data are not applicable.
Node ID Head-Stage (m) Bottom Elevation (m)
Archibald Upstream 12.0 – Archibald Downstream 6.8 – Queseria Upstream 4.0 – Queseria Downstream 2.3 – Scotts Upstream 6.8 6.2 Scotts PH-up 5.4 4.8 Scotts PH-down 4.9 4.3 Scotts VFD-up 4.4 3.8 Scotts VFD-down 4.1 3.5 Scotts Downstream 2.3 1.7
62 columns, 22 rows, and 3 layers oriented in a north-south direction was created.
The grid generated included4,092 cells. The active zones of the modelwere defined
by activating cells in the coverages. The active grid included 2,398 cells. The 3D
grid model wasconverted to a MODFLOW 2000 numericalmodel. The MODFLOW
global optionswere inspected to ensure thatthe conceptual model data were assigned
to the appropriate cells. Starting heads representative of steadystateconditions were
generated by runninga transient simulationwith 75stress periods (75 days) and no
groundwater pumping. The model generated heads forthe 75th time step were set as
the starting heads forall subsequent transientsimulations.