Current and Innovative
Approaches for Quantifying
Recharge at the Watershed Scale
Bridget R. Scanlon
Jackson School of Geosciences
Bureau of Economic Geology
Univ. Texas at Austin
Talk Outline
•
What is recharge?
• Why is recharge important?
• How do we estimate recharge?
– Remote sensing
– Surface water approaches
– Unsaturated zone approaches
– Saturated zone approaches
QAc1766(b)c
What is recharge?
Addition of water to an aquifer from any
direction (down, up, laterally)
Types of Recharge
• Diffuse recharge: precipitation, irrigation
• Focused recharge: streams, lakes, playas,
reservoirs
• Induced recharge: capture of surface water
• Artificial recharge: surface, vadose, and aquifer
structures
• Enhanced recharge: remove vegetation to reduce
ET (e.g. brush control)
Why is recharge important?
• Quantify groundwater resources
• Evaluate contaminant transport through the
unsaturated zone
• Identify suitable sites for waste disposal
• Groundwater protection….aquifer vulnerability to
contamination
How do we estimate recharge?
• Remote sensing… microwave, gravity
• Surface water techniques... watershed modeling,
heat tracer
• Unsaturated zone techniques… chemical tracers,
numerical modeling
• Saturated zone techniques…chemical tracers,
numerical modeling
Scanlon et al., 2002
Scanlon et al., 2002
Scanlon et al., 2002
How do we estimate recharge?
• Remote sensing…. microwave, gravity
• Surface water techniques... watershed modeling,
heat tracer
• Unsaturated zone techniques… chemical tracers,
numerical modeling
• Saturated zone techniques…chemical tracers,
numerical modeling
Remote Sensing
(Water Budget Components)
Water budget
R = P + Qon - Qoff - ET - DS
Precipitation: NEXRAD radar
ET (visible and thermal infrared spectral radiances,
Bastiaanssen et al. (1998) algorithms)
DS (change in water storage)
microwave remote sensing
satellite based gravity measurements
Microwave Remote Sensing
•
•
•
•
•
Passive and active systems
Truck, aircraft, or satellite systems
Optimal frequency < 6 GHz
Independent of cloudiness
Spatial resolution m (ground based), 100’s of m
(airborne) to10s of km (satellite systems)
• Repeat cycles for satellites: daily to 10s of days
• Difficult to relate soil moisture in 2 to 5 cm zone
to recharge
Soil Water Content
Data
Washita’92
Oklahoma
ESTAR
Passive, airborne
microwave data
Spatial resolution
200 m
Jackson et al., 1995
Gravity Recovery and Climate Experiment (GRACE)
Gravity Recovery and Climate
Experiment (GRACE)
• Launched in March 2002, 5 yr lifespan
• Monitor terrestrial water storage (including
unsaturated and saturated zones)
• Reliable footprint > 200,000 km2 (Rodell and
Famiglietti, 1999, 2001)
• Monthly repeat cycles
• Ground-based gravity surveys can also be used
to complement the satellite data
Lacoste and Romberge
Relative Gravity Meter
Model D
Comparison of Water Storage from Gravity Survey
and Groundwater Levels in a Well, Garden Canyon, AZ
Ground-water Storage and Water-Level Change
Near Garden Canyon, 1995-2001
0
Oct. 2000
Precipitation,
Streamflow,
Infiltration,
and Recharge
Depth to Water, in Feet
20
-2
40
-4
60
-6
80
-8
Water Level
Storage
100
120
Dec-94
Dec-95
Dec-96
-10
Dec-97
Dec-98
Dec-99
Dec-00
Dec-01
-12
Dec-02
Storage Change, in Feet of Water
0
How do we estimate recharge?
• Remote sensing…. microwave, gravity
• Surface water techniques... watershed modeling,
heat tracer
• Unsaturated zone techniques… chemical tracers,
numerical modeling
• Saturated zone techniques…chemical tracers,
numerical modeling
Watershed Models
Advantages:
– Can be used to evaluate effect of climate and land use
on watershed response
– Few assumptions about mechanism of water movement
Disadvantages:
– Large number of parameters to be measured or
estimated
– Recharge estimates accumulate errors in other terms
– Large uncertainties when R is small compared with
other variables
Uncertainty in Recharge Estimates
using Watershed Modeling
Simple example:
• R = P - ET
• If assume error in each parameter is  10%:
• R = 500 ( 50 mm) - 450 ( 45 mm)
= 50  95 mm
• This corresponds to 200% uncertainty in
recharge estimate
Common Watershed Models
• HEC HMS: US Army Corps of Engineers
• Precipitation Runoff Modeling System (PRMS):
USGS, Leavesley
• Deep Percolation Model (DPM): USGS, Bauer and
Vaccaro
• Hydrologic Simulation Program Fortran (HSPF):
USGS and US EPA, water flow, solute transport
• Soil Water Assessment Tool (SWAT), USDA,
Arnold
• TOPMODEL - Beven, UK
Model Input
#0 Adaven
Tonopah
#0
Goldfield
#0
#0
Death Valley Regional
Hydrologic System
#0
Cities and towns
#0
Yucca Mountain
Nevada Test Site
Death Valley National Park
Playas and wide channels
Ground water discharge (ET)
Amargosa River
Death Valley GW basin
Ground water flow model
Beatty
#0
#0
#0
0#
Pahrump
#0
Las #0
Vegas
DEM (shaded relief)
Trona
#0
N
W
E
S
#0 Baker
50
0
50 Kilometers
Hevesi, 2002
Deterministic Watershed Model:
Daily Root Zone Water Balance Model
Hourly Net Radiation Energy Balance Model
Rain
Runoff
Evapotranspiration
Snow-melt
Run-on
Top Soil Layer
Soil
Middle
Soil Layers
Bedrock
Root Zone
Rock Layer
Net Infiltration
Lower
Soil Layer
Infiltrated
Run-on
6.0
meters
Ñ
Ñ
Ñ
Ñ
1950 - 1999 Modeled
Net Infiltration
( millimeters/year )
Ñ
Ñ
GW Discharge Zone
Ñ
0 - 0.1
0.1 - 2
2-5
5 - 10
10 - 20
20 - 50
50 - 100
100 - 500
> 500
Ñ
Ñ
Ñ
not modeled
N
W
E
50
S
Ñ
Ñ
Ñ
Ñ
0
50 Kilometers
How do we estimate recharge?
• Remote sensing…. microwave, gravity
• Surface water techniques... watershed modeling,
heat tracer
• Unsaturated zone techniques… chemical tracers,
numerical modeling
• Saturated zone techniques…chemical tracers,
numerical modeling
Heat as a Tracer
(Constantz, 2002)
Frequency and Duration of Ephemeral Stream Flow
1. Delay in peak flows down channel
2. Decrease in amplitude of streamflow down channel
3. Decrease in duration of streamflow down channel
(Constantz, 2002)
Streambed Temperature Variations
(Constantz et al., 2001)
40
35
PS - 2
30
25
20
15
10
5
0
1 0 /9 /9 8
1 2 /8 /9 8
2 /6 /9 9
4 /7 /9 9
6 /6 /9 9
8 /5 /9 9
6 /6 /9 9
8 /5 /9 9
Thermographs
Bear Canyon
(New Mexico)
40
35
PS - 3
30
25
20
15
10
5
0
1 0 /9 /9 8
1 2 /8 /9 8
2 /6 /9 9
4 /7 /9 9
40
35
PS - 4
30
25
20
15
10
5
0
1 0 /9 /9 8
1 2 /8 /9 8
2 /6 /9 9
4 /7 /9 9
6 /6 /9 9
8 /5 /9 9
2 /6 /9 9
4 /7 /9 9
6 /6 /9 9
8 /5 /9 9
40
35
PS - 5
30
25
20
15
10
5
0
1 0 /9 /9 8
1 2 /8 /9 8
Constantz et al., 2001
Subsurface Temperature Monitoring to
Estimate Recharge
Rio Grande near Albuquerque
Monitoring T in groundwater
Simulated T fluctuations using
VS2DH and PEST
Recharge rates: 10 – 14 m/yr
(Bartolino & Niswonger, 1999)
How do we estimate recharge?
• Surface water techniques... catchment scale
modeling (water budget), heat tracer
• Unsaturated zone techniques…chemical tracers,
numerical modeling
• Saturated zone techniques…numerical modeling,
chemical tracers
Focused Flow Beneath Playas, Southern High Plains
Multipeaked Tritium Profile Beneath a Playa
Scanlon and Goldsmith, 1997
Playa vs Interplaya Recharge, High Plains, Texas
Scanlon and Goldsmith, 1997
Subsurface Distribution of Bomb Pulse Tracers
Chloride Mass Balance
Chloride input = chloride output in subsurface
P  CCl P  R  CClu z
where P is precipitation, R is recharge rate, and
CClP and CCluz are Cl concentrations in
precipitation and uz pore water respectively.
R
P  CCl p
CCluz
Cl Profiles Interdrainage Settings, Chihuahuan Desert, Texas
Scanlon, 2000
Chloride Bulge Profile
(Chihuahuan Desert, Texas)
Scanlon, 1991
Evaluation of Upward Flow in Interdrainage
Semiarid Regions
Mojave
Desert
AD
#
High
Plains
#
HP
AD: Amargosa Desert
HB: Hueco Bolson
EF: Eagle Flat
HP: High Plains
##
HB EF
Chihuahan
Desert
Scanlon et al., in press
Upward Flow (simulated vs. measured)
Matric potential (m)
-1000
0
-750
5
0 yrs
1 kyr
2 kyr
5-9 kyr
Field
Depth (m)
10
15
-500
-250
0
-1000
0
5
High
Plains
10
15
20
20
0
0
5
10
15
20
0 yrs
5 kyr
10 kyr
13 kyr
Field
10
20
Hueco
Bolson
30
40
50
-750
-500
-250
0 yrs
5 kyr
12 kyr
Eagle
Field
Flat
Field-OP
0 yrs
5 kyr
10 kyr
16 kyr
Field
Lab
Amargosa
Desert
0
Upward Flow…Downward Diffusion
Simulated vs. Measured Chloride
Chloride (mg/L)
0
2500
5000
0
1 kyr
5 kyr
9 kyr
Measured
10
High
Plains
15
20
5
10
10000
0
20000
0
5
10
20
Eagle
Flat
15
0
15
10000
0 kyr
5 kyr
12 kyr
Measured
20
0
10
5000
0
5
Depth (m)
0
Hueco
Bolson
5 kyr
10 kyr
13 kyr
Measured
5000
20
30
40
50
Amargosa
Desert
10000
5 kyr
10 kyr
16 kyr
Measured
Time Required to Flush Chloride
Chloride (mg/L)
0
10000
20000
Depth (m)
0
5
0 yr
20 yr
100 yr
500 yr
10
15
Flux 1 cm/yr
20
30000
How do we estimate recharge?
• Surface water techniques... catchment scale
modeling (water budget), heat tracer
• Unsaturated zone techniques…chemical tracers,
numerical modeling
• Saturated zone techniques…chemical tracers,
numerical modeling,
(Plummer and Busenberg, 1999)
Tracers (Chlorofluorocarbons)
• CFCs work well in shallow, aerobic, sand aquifers
• CFCs cannot be used in areas of septic tanks or
near urban regions
• Recharge rates using CFCs
– 30 – 60 cm/yr Delmarva Peninsula (Dunkle et
al., 1993)
– 13 cm/yr Sturgeon Falls, Ontario Canada (Cook
et al., 1998)
– History of nitrate loading, Locust Grove,
Maryland (Bohlke et al., 1998)
Tracers (Tritium/Helium)
t
•
3H/3He
H / 3 He
3
  3 H 

 ln 
 1
1  3 He

 
age gradients near water table -- vertical
velocity -- groundwater recharge rate
• Need rapid vertical flow velocity to confine
3He…recharge rate should be greater than
3 - 5 cm/yr
• Can be used to date water in contaminated areas
Vertical Cross Section 3H/3He Ages, CA
(Hudson et al., 2002)
Hudson et al. (2002)
How do we estimate recharge?
• Surface water techniques... catchment scale
modeling (water budget), heat tracer
• Unsaturated zone techniques…chemical tracers,
numerical modeling
• Saturated zone techniques…chemical tracers,
numerical modeling,
Estimation of Recharge Using Groundwater Models
• If you only have head data, you can only estimate
the ratio of recharge to hydraulic conductivity.
This is a result of Darcy’s law:
R  K
h
x

h  h0 
R
x
K
• As long as R/K is the same, calculated h will be
the same: R/K = 10/5=2; R/K=100/50=2
Estimation of Recharge Using Groundwater Models
• Additional information: flux data, travel times
– Flux data from spring discharge, baseflow
discharge to streams
– Travel times from environmental tracers such
as 3H, 3H/3He, CFCs, and 14C
• Trial and error calibration and automated inverse
procedures can be used to match heads, fluxes,
and travel times
Automated Inverse Modeling
• Codes: UCODE, PEST, MODFLOW2000
• Minimize objective function using nonlinear
regression
• Example objective function for heads and flows:
nh
nq
i 1
i 1
S (b)   wi (hi  hˆ)2   w j ( q j  qˆ j )2
S is sum of squared weighted residuals
b is the vector of parameters being estimated
w are the weights on the measured heads and flows
h, q are measured heads and fluxes
^ indicates simulated
Inverse Modeling Death Valley
d’Agnese et al. (1999)
• Steady state, predevelopment model
• 3D, 100,000 km2
• Recharge based on modified Maxey Eakin
including elevation, slope, aspect, parent
material, and vegetation
• Automated inverse modeling (MODFLOWP) using
heads (n=500) and spring discharge (n=63)
• Matched heads ± 45 m; spring flows ± 46%.
• Composite scaled sensitivities used to determine
parameters to be estimated.
COMPOSITE SCALED SENSITIVITIES
12
10.8
10
8.94
8
7.85
7.53
6
4
3.98
3.32
2.82
1.77
2
0.963
0
K1
K2
K3
K4
K5
Kfmtn
PARAMETERS
ANIV3
RCH2
RCH3
Inverse Modeling Middle Rio Grande
Basin (Sanford, 2000)
• Observations: heads (n=200) and 14C ages
(n=200)
• Weighted observations
• Flow (MODFLOW), Ages (ModPath), Inverse
modeling (UCODE)
• Recharge mountain front and streams
• K from geologic data
• Recharge estimates: 55,000 af/yr (20,000 af/yr
from Rio Grande)
• Previous recharge estimates: 124,000 af/yr
(Tiedeman et al., 1998)
Estimation of Recharge Using Groundwater Models
• Joint inversion using heads and ages
• Heads are sensitive to the ratio of recharge to
hydraulic conductivity (R/K)
• Ages from tracer data are sensitive to the ratio of
recharge to porosity (R/n)
• Use of both head and age data provide
constraints on recharge, hydraulic conductivity,
and porosity
• Because recharge, hydraulic conductivity, and
porosity are highly correlated, porosity is
generally estimated for the system.
Boundary
of the
Middle
Rio-Grande
Basin
Groundwater flow
paths and travel
times
Edge of the
finite-difference
model grid
Albuquerque
Three path lines
from different
recharge zones to
obs. wells
Sanford, 2000
Summary
• Remote sensing offers promise for qualitatively estimating
recharge rates at large scales
• Surface water techniques….important in semiarid
regions…estimate potential recharge
• Unsaturated zone data … interdrainage semiarid
regions…no recharge…upward flow since Pleistocene
• Saturated zone techniques
– CFCs and 3H/3He…higher recharge rates (humid
regions)
– modeling based on head, age, and flux data
• Use as many different approaches as possible
Theme Issue:
Groundwater Recharge
Vol. 10, No. 1, February 2002
14 papers; methods, issues, case studies