Estimates of Ground-Water Estimates of Ground-Water Recharge in MinnesotaRecharge in Minnesota
Dave Lorenz and Geoffrey DelinUSGS Water Science Center of Minnesota
Research supported by the USGS, Office of Ground Waterand DNR Waters
Study ObjectivesStudy Objectives
Quantify recharge to unconfined sand and gravel aquifers in Minnesota using multiple methods representing different time and spatial scales.
Compare and contrast the results.
Estimation Methods UsedEstimation Methods UsedMultiple regression analysis relating recharge to precipitation, ET, and soils data (Regional Regression Recharge)
Ground-water level fluctuation (water-table fluctuation)
Unsaturated-zone water balance (zero-flux plane)
Ground-water age dating
Regional Regression Recharge Regional Regression Recharge MethodMethod
Recharge based on the Rorabaugh method that estimates average recharge in a drainage basin from streamflow records.
Rorabaugh Method—TheoryRorabaugh Method—Theory
Rorabaugh Method—ComputationRorabaugh Method—ComputationL
OG
AR
ITH
M O
F S
TR
EA
MF
LO
W
PEAK 1
PEAK 2
RECESSION RATE = K
TIM
E O
F P
EA
K 1
BE
GIN
OF
GW
RE
CE
SS
ION
TIM
E O
F P
EA
K 2
Stream Gaging Station SelectionStream Gaging Station Selection
Criteria reviewed: length of record, common periods of record, missing data, size of watershed, (maximum of 3,000 mi2), and existence of control structures (dams or diversions).
40 stations selected based on these criteria
Stream Stream Gaging Gaging Stations Stations Used in Used in RORA RORA
Baseflow Baseflow Recharge Recharge AnalysesAnalyses
Landscape CharacteristicsLandscape Characteristics
Several landscape characteristics were considered originally:
Soil characteristics; Percent sand, percent clay, porosity, bulk density, permeability, and specific yield.
Other landscape characteristics: percent various classes of geologic deposits in basin, basin slope, stream slope, and percent lake area in basin.
Landscape Characteristics—Final Landscape Characteristics—Final
Decided to use specific yield (SY) as the landscape characteristic in the model:
Direct measure of the capacity of the material to hold and release water under gravity. This is a linear property. That makes it possible to project back to the land surface.
Highly correlated with other properties that affect recharge—permeability and hydraulic conductivity.
Specific Yield Specific Yield
Specific Yield Specific Yield
Several methods to estimate SY were used. The method described in Rawls (1982) was used in the final regression equation. It uses percent sand, clay and organic matter. Data from STATSGO.
Shown is average precip. 1971-2000
Regression used decadal average going back through 1940.
PrecipitationPrecipitation
inches
Shown is average ET. 1961-1990
Regression used decadal average of growing degree days.
Evapotran-Evapotran-spiration (ET)spiration (ET)
Regression EquationRegression EquationDecadal averages for recharge and precipitation were used—reduces serial correlation between precipitation and recharge and smoothes out the variability in precipitation and recharge.
Generalized least squares regression was used to account for the correlation between decadal data for each basin. Recharge = 14.25 + 67.63(SY) + 0.6459(P) - 0.02231(GDD*)
GDD* is the minimum of GDD or 1350 degree days above 10 degrees celsius.
Average Average Recharge Recharge through through soils in soils in
MinnesotaMinnesota1971-20001971-2000
Water-Table Water-Table Fluctuation (WTF) Fluctuation (WTF)
MethodMethod
Data from 38 wells equipped with Data from 38 wells equipped with data loggers at five different sitesdata loggers at five different sites
Temporal variability in rechargeTemporal variability in recharge
Water-Table Fluctuation MethodWater-Table Fluctuation Method
Δh
Recharge = SY Δh
Multiple WTF Multiple WTF Methods UtilizedMethods Utilized
Graphical method
RISE program (Rutledge, 2003)
Master Recession Curve
Correlation Between Graphical Correlation Between Graphical WTF Recharge and UZ WTF Recharge and UZ
ThicknessThickness2003 data 2003 data from 23 from 23 wells at 3 wells at 3 different different sitessites
0
10
20
30
40
50
60
70
0246810
Unsaturated zone thickness, meters
Rec
har
ge,
cm
/yr
Williams Lake
Glacial Ridge
Anomalously high recharge for UZ
thicknesses > 3.5 m
Bemidji
Effects of Measurement Interval on Effects of Measurement Interval on WTF Recharge EstimatesWTF Recharge Estimates
1993 datalogger data from MSEA well R2 near Princeton, MN1993 datalogger data from MSEA well R2 near Princeton, MN
4
5
6
7
8
9
10
11
12
13
14
0 5 10 15 20 25 30 35
Measurement frequency, days
Est
imat
ed r
ech
arg
e, c
m/y
r
Hou
rly /
daily
No change No change in estimated in estimated recharge recharge going from going from hourly to hourly to daily daily measuremeasure
Recharge estimates based on WTF method (RISE program)
0-54 % under-0-54 % under-estimation of estimation of the recharge: the recharge: from daily to from daily to weekly weekly measurementmeasurement
Weekly
(- 23%)
18-60 % under-18-60 % under-estimation of estimation of the recharge: the recharge: from daily to from daily to monthly monthly measurementmeasurement
Mon
thly(- 48%)
Measurement interval, days
Est
imat
ed r
ech
arg
e, c
m/y
r
Unsaturated-Zone Water Unsaturated-Zone Water Balance Balance
(zero-flux plane) (zero-flux plane) MethodMethod
Bemidji, Williams Lake, andBemidji, Williams Lake, andPrinceton MSEA sitesPrinceton MSEA sites
Temporal variability in rechargeTemporal variability in recharge
UnsaturatedUnsaturated ZoneZone Water Balance MethodWater Balance Method
Lowland Sites
0
10
20
30
40
50
60
70
Upland Upland Upland
MSEA R2 Bemidji 9014 Williams Lake
0
10
20
30
40
50
60
70
Lowland Lowland Lowland
Bemidji 981 Bemidji 9015 MSEA R1
Rec
har
ge
as a
per
cen
t o
f p
reci
pit
atio
n
Upland Sites
Bemidji well 981
Bemidji well 9015
MSEA well R1
MSEA well R2
Bemidji well 9014
Williams Lake site
Rec
har
ge,
per
cen
t o
f p
reci
pit
atio
n
Ground-Water Age Ground-Water Age Dating MethodDating Method
Average recharge, spatial variabilityAverage recharge, spatial variability
Ground-Water Age Dating MethodGround-Water Age Dating MethodD
ep
th b
elo
w w
ate
r ta
ble
, in
met
ers R1 - Low land Site
CCl F -model recharge age, in years before the sampling date
2 2
V v o = 0.7 0.2 m /yr+
R1-B
0 10 20 30 40 5012
11
10
9
8
7
6
5
4
3
2
1
0
1993, 7°C
1994, 7°C
1993, 9°C
1994, 9°C
W ater table (Z = 9.5 m)
R1-10
From Delin et al. (2000)
Example from Example from Princeton MSEA site Princeton MSEA site using CFC datausing CFC data
Recharge = Recharge = vertical GW velocity vertical GW velocity x porosityx porosity
SFSF66 and and 33H-H-33He He
techniques can techniques can also be used; min. also be used; min. time resolution of time resolution of ~1 year BP~1 year BP
Method Method ComparisonComparison
0
10
20
30
40
50
60G
laci
alR
idg
e
Des
Mo
ines
Riv
er
Wil
liam
sL
ake
Bem
idji
MS
EA
Selected USGS Research Site
Ave
rag
e g
rou
nd
-wat
er r
ech
arg
e, c
m/y
r
RISE program
Graphical
MRC
UZWB
GW age dating
RRR Method
GW flow model
35% of precipitation
Comparison of Average Recharge Comparison of Average Recharge Rate Computed at Each SiteRate Computed at Each Site
Shallow depth to Shallow depth to water table results water table results in WTF recharge in WTF recharge rates being too large rates being too large for Glacial Ridge, for Glacial Ridge, Des Moines River, Des Moines River, and Williams Lake and Williams Lake sitessites
Pretty good Pretty good agreement agreement between regional between regional estimates at most estimates at most sitessites
WTF Method
RISE program
Graphical
GW flow model
35% of precipitation
MRC
UZWB
GW age dating
RRR Method
Other site-specific Methods
Regional Methods
Similarity in Similarity in recharge rates for recharge rates for some methods at some methods at some sitessome sites
Of the WTF approaches, Of the WTF approaches, MRC estimates generally MRC estimates generally are the greatest; RISE are the greatest; RISE program lowestprogram lowest
Methods are scale dependentMethods are scale dependent
AlmostAlmostthe the endend
Statewide Statewide Analysis Analysis
WTF WTF MethodsMethods
45 wells with weekly
data available
from DNR database
45 wells with weekly
data available
from DNR database
Datalogger site (36 wells total)
Bemidji
Williams Lake
Des Moines River
MSEA
Glacial Ridge
Graphical MethodGraphical Method
Manual method for estimating recharge. Developed in the late 1950s.
Baseline recession that would have occurred in the absence of recharge projected to the time of peak in the hydrograph.
The value of Δh determined manually.
Graphical Calculation for WTF MethodGraphical Calculation for WTF Method
From Delin (1990)
RISE ProgramRISE Program
Simple program that calculates the daily rise of water level in an observation well.
The program makes no allowance for the baseline recession that would have occurred in the absence of recharge.
The input data can be read right out of NWIS Web or can be created from data logger files.
Rutledge (2003) electronic communication
RISE Calculation for WTF MethodRISE Calculation for WTF Method
From Delin (1990)
Master Recession Curve MethodMaster Recession Curve Method
First step is to define a Master Recession Curve from “typical” recessions for a well. This is accomplished by a nonlinear regression that estimates the recession rate and recession asymptote. Other methods for estimating a master recession curve have also been developed.
Program calculates the daily recession of water level in an observation well and the rise from the difference between the theoretical recession and the actual water level.
MRC Calculation for WTF MethodMRC Calculation for WTF Method
From Delin (1990)
Recharge Estimates - WTF MethodRecharge Estimates - WTF Method
Williams Lake examples: Williams Lake examples: PrecipitationPrecipitation and and rechargerecharge in in cm/yrcm/yr
UZ thickness: 5 m 9 m 2 mUZ thickness: 5 m 9 m 2 m
16% 18% 11 % 13% 9% 10% 91% 120% 101%
Unsaturated Zone Water Balance Unsaturated Zone Water Balance
From Delin and Herkelrath (in press)
Time
Zero-flux plane
Wells Wells Sampled Sampled for SFfor SF66
GW age GW age dating.dating.
Also used Also used CFCs for CFCs for
datingdating
22 wells sampled for
SF6, including
2 nests
SF6 sample site (18 this study)Other GW age-dating site (6)
Bemidji
Williams Lake
Des Moines River
EXPLANATIONMSEA
Glacial Ridge
Rock River
Perham
Prairie Island