Transcript Integrated Modeling of Regional Basins: Thirteen Years of

Integrated Modeling of Regional Basins:
Thirteen Years of Hard Lesson Learned
Mark A. Ross, Patrick D. Tara, and
Jeffrey S. Geurink
University of South Florida
Integrated Model: Coupled
Comprehensive SurfaceGroundwater Model
(Specifically: Experience with
coupled HSPF-MODFLOW model
known as FHM, ISGW, or IHM)
FHM (HSPF/MODFLOW) Integration Pathways
Process
Coupling
Hydrologic Cycle
1. DTWT
Interception
2. Hydr. heads
(streams,
wetlands &
Transpiration
Precipitation
lakes)
Interflow
Infiltration &
Impervious Lens
Percolation
3. Base flow Evaporation
Evaporation
Surface water
4. SW/GW ET
Runoff
storages & stream
flows
5. Irrigation
Groundwater Flow
Water Table
Baseflow
Leakage
Bays, Gulf,
fluxes
Unconfined Aquifer
and Oceans
6. Variable SY
Aquitard (Confining Unit)
(vadose moist.)
Confined or Artesian Aquifer
FHM Chronology
1989 FHM development
 1993 FHM ver 1.2 used in mine reclamation
 1995 ISGW proprietary version used for public water
supply investigations, FHM District-scale application
 1995-1997 FHM peer-reviewed, adapted for USFWS
water rights investigations
 1995-1998 period of W-C Fl water wars, numerous
models
 1997-2000 several FHM version updates for regional W-C
Fl regional investigations, ISGW peer reviewed
 2001-2002 major re-write of FHM/ISGW->IHM (Interra,
Aquaterra, USF)

Early Problems
(Limitations)
 Computers
(286s)
 Model components (HSPF & MODFLOW)
 Data utilities (GIS, database programs)
 Digital data (no digital quads)
 Client perspectives/interest (“recharge
generator”)
 User acceptance (“too much time & money”)
Consequence: Limited Discretization
Bill Williams National Wildlife Refuge
R 18 W
R 16 W
R 17 W
95
Lake Havasu
T
11
N
River
R 18 W
R 17 W
N
R 16 W
R 21 E
R 24 W
R 23 W
T
Cibola National
Wildlife Refuge
1
S
T
2
S
N
IMPERIAL
NATIONAL
WILDLIFE
REFUGE
R 16 E
R 17 E
T 16 N
Las Vegas
National Wildlife
Refuge
T 15 N
N
Rattlesnake Creek Study
Model Simulation Results Rattlesnake Creek
USFWS Applications
Lessons Learned
 Need
does not justify model where there is no data
 Streamflow separation (runoff/baseflow) very
important to do but problematic
 Component pre-calibration very important
 Size and discretization barriers remained
 Need much better data utilities
Mid 90s “Water Wars”
West-Central Florida
 Many
model applications, similar stream flow
performance & gross ET, wildly varying resultant
models (recharge), conclusions
 Wide variability in model parameterization
resulted from inadequate data, understanding and
characterization of processes
 Need for detailed basin-scale study, tie down
internal fluxes and storages
Saddle
Creek
Study
Near Field Model
(Data
Collection,
Far-field
and Nearfield
Models)
10
0
10
20
30
40
Distgen.shp
Saddle Creek Basin Outline
No Flow Boundary
General Head Boundary
Hydrography
50
60 Miles
N
W
E
S
Saddle Creek
Gauging
Stations
Station 17b Calib
120
100
Flow(cfs)
80
O bs .
S tre a m fl ow
Station 17 Calibration
S im.
S tr e a m flow
60
40
20
0
Oct-96
A pr-97
Oct-97
A pr-98
Oct-98
A pr20
D ate
Obs.
Str eamfl ow
Obs.
Basefl ow
Si m.
Str eamfl ow
Si m.
Basefl ow
Inches
15
10
5
0
Oct - 96 Ap r - 97 Oct - 97 Ap r - 98 Oct - 98 Ap r -
D ate
Saddle Creek Study
Lessons Learned
 Extensive
basin-scale data collection helped refine
model calibration and resultant internal fluxes
 Real important to characterize time/space scale of
rainfall
 Needed to understand the mechanism of runoff,
especially role of variable saturated areas (VSAs)
 Variable specific yield (SY) very important process
Stream/Lake
Basins
Ungaged
Gaged
Coastline
SWFWMD
Southern
District Model
HSPF,
MODFLOW pre-calibration,
1st phase of integrated model
N
30
0
30
60 Miles
SWFWMD Southern District Model
Lessons Learned
 Importance
of including all hydrography explicitly
 Strong parameterization and model performance
constraints by DTWT resulted in greatly improved
calibration (streamflow and aquifer behavior)
 Indicated much higher GW ET fraction in shallow
watertable settings than previously considered –
resulting in model concept changes
 Importance of irrigation fluxes and deep aquifer
discharges zones
Alafia Subbasins and Land Use
Alafia Subbasins
Subbasins
Land Use
Urban
Ag/Rec Irrigated
Grass/Pasture
Forrested
Open W ater
Wetlands
Mining/Other
Statecounties_poly.shp
N
W
E
S
9000
0
9000
18000 Meters
Alafia Model
Lesson
 Need
to explicitly characterize connected and
unconnected hydrography for each basin
 Depart from basin calibration, move to land use
calibration
Alafia Micro-Scale Field Study
Preliminary Results
Runoff dominated by saturation excess (VSAs)
 Air entrapment plays a strong role
 Water table fluctuations are very rapid
 Baseflow timescale may be controlled by ET timescale not
water table drainage
 SY is highly variable (.2 – 2 m) controlling water table
fluctuation
 Vadose zone moisture maintenance pronounced, thus
significant GW ET (watertable depths < 2 m)

New IHM Model Developments
 New
code integration structure, HSPF and
MODFLOW run concurrently, greatly enhanced
capability and enhanced run speed
 Integration and all other timesteps completely user
defined
 Landforms explicitly modeled (more distributed
parameterization)
Old Model Structure
New Model Structure
Legend
Land Use
Urban
Irrigated/Rec
Grassland/Pasture
Forested
Open Water
Wetlands
Mining/Other
TBW
Connected &
Unconnected
Hydrography
DICRETIZATION
MODFLOW:
20,000 grids (1/4 mi)
150,000 river reaches
3 aquifer layers
HSPF:
172 non-connected reaches
172 storage attenuation
73 routing reaches
172 basins
5 landform categories
320,000 landuse polygons
Three Layer Soil Moisture Model
zo = 0
z
Three-Layer Soil Moisture Model
a)
Zone 1: Upper Constant
Moisture Region
b)
c)
Equilibrium
Profile
D1
zcz
Zone 2: Intermediate
Capillary Zone
Wet
Profile
Dry
Profile
D2, icz
(z)
zcf
Zone 3: Lower
Capillary Fringe
D3, cf
zwt
zrz
Important Spatial & Temporal Scales
(Shallow Aquifer Coastal Plain Systems)
ET: Surface
Vadoze
Water table
Rchg: Vadose zone
Surficial
Confined
1 km2
102 m
102 m
102 m Horizontal
.1 – 3 m Vertical
3mV
0.1-2 m
102 mH, 1 mV,
103 mH, 10-102 mV
5-15 min.
5-15 min.
5-15 min.
hourly
daily
Daily
1-24 hrs.
1-24 hrs.
Wkly
Other: Stream base flow
GW Pumping sens.
Landuse change sens.
103 m
Vars.
Vars.
daily – season
daily
1-5 years
Rainfall
Runoff flow plain
Infiltration
Overall Conclusions &
Recommendations
 Understand
the hydrologic processes and water
budget magnitudes before beginning
 Ensure adequate data to support model
 Commit to the data pre-analysis
 Understand the limitations and long-term
commitments
 Attention to internal fluxes and storages will
ensure fully constrained, unique solution
Integrated Model Commitments
Enormous data requirements
 Complete surface water dataset
 Complete groundwater dataset
 Data pertaining to the integration
 Different timescales and space scales
 Considerable data analysis prior to calibration
 More difficulty in calibration
 Users must possess both SW and GW expertise

Root Capillary Zone
ZLS
ZRZ
ZCZ
ZCF
ZWT
ZLS
ZCZ
ZCF
Z
ZRZ
WT
ZCZ
ZRZ
ZCF
ZWT
(a) Deep water table (ZWT < ZCF 
ZRZ) (b) Capillary interaction (ZWT < ZCF Zrz)
(d) Root zone water table
( ZRZ < ZWT; ZCZ ZLS)
ZLS
(c) Root zone capillary fringe
(ZWT ZRZ < ZCF)
ZLS
ZCZ
ZCF
ZWT
ZLS
ZCF
ZWT
ZLS
ZWT
ZRZ
ZRZ
ZRZ
(e) Capillary zone at land surface
(ZCF ZLS ZCZ)
(f) Capillary fringe at land surface
(ZWT ZLS ZCF)
Figure 1 a-f: Six cases for vadose zone soil moisture condition.