Comparison of Recharge Estimation Methods Used in Minnesota by Geoffrey Delin, Richard Healy, David Lorenz, and John Nimmo Minnesota Ground Water Association Spring Conference Methods for Solving Complex Ground-Water Problems St. Paul, MN, April 19, 2007 U.S. Department of the Interior U.S. Geological Survey Funding provided by the USGS Ground Water Resources and National Research Programs and the Minnesota Department of Natural Resources
Take-Home Messages High-quality, long-term, continuous hydrologic and climatic data are important for accurate recharge estimation Use of multiple recharge estimation methods is beneficial Local and basin-scale scale recharge estimates can be regionalized using the regional regression recharge (RRR) model RRR map provides regional estimates for uses such as initial ground-water flow model calibration
USGS Study Objectives Quantify recharge to surficial materials in Minnesota: (a) using multiple methods, (b) representing different time and spatial scales Compare results of the methods Attempt to regionalize the site-specific specific or basin-scale scale estimates
Recharge Estimation Methods Used Site-Specific Specific Methods Unsaturated-zone water balance (UZWB) Water-table table fluctuation (WTF) Age dating of ground water Basin-scale scale Method RORA analysis of streamflow records using a recession-curve curve displacement technique
Other Methods Used to Estimate Recharge in Humid Regions Local-Scale Methods Lysimeters (unsaturated zone) Chloride, isotopic, and environmental tracers Applied tracers Darcy flux (using Darcy s Law) Seepage meters Water-balance equations Numerical modeling Basin-Scale and Regional Methods Streamflow hydrograph separation (baseflow) Water-balance equations Numerical modeling GIS techniques Remote sensing techniques Scanlon et al. (2002)
Unsaturated-Zone Water Balance (UZWB) Method Spatial scale: : 1 m 2 Temporal scale: : Event based / seasonal Temporal variability in recharge
Unsaturated-Zone Water Balance Method Assumptions Based on premise that soil water moves upward in response to ET above a boundary in the unsaturated zone and that Water below that depth moves downward to the water table as a result of each recharge period Delin and Herkelrath (2005)
UZWB Equipment Needs Soil moisture and soil tension measured hourly at multiple depths in the unsaturated zone UZ instrumentation Bemidji research site Datalogger & multiplexers
Unsaturated-Zone Water Balance Depth below land surface, centimeters Water table Volumetric moisture content Conceptualized diagram
Unsaturated-Zone Water Balance Depth below land surface, centimeters Water table Volumetric moisture content Conceptualized diagram
Unsaturated-Zone Water Balance Precipitation Depth below land surface, centimeters Integration between these two profiles approximates the infiltration amount for this event However, due to ET, not all of this water reaches the water table Water table Volumetric moisture content Conceptualized diagram
Unsaturated-Zone Water Balance Precipitation Depth below land surface, centimeters Evapotranspiration Percolation ET/drainage boundary Water table Volumetric moisture content Conceptualized diagram
Unsaturated-Zone Water Balance Precipitation Depth below land surface, centimeters Evapotranspiration Percolation ET/drainage boundary Water table Volumetric moisture content Conceptualized diagram
Unsaturated-Zone Water Balance Precipitation Depth below land surface, centimeters Evapotranspiration Percolation UZWB recharge ET/drainage boundary Water table Volumetric moisture content Conceptualized diagram
Limitations of UZWB Method Small scale of influence Intensive collection of soil- moisture and soil-pressure data is required Probes installed in wall of trench Delin and Herkelrath (2005)
Water-Table Fluctuation (WTF) Method Spatial scale: : 1 to 100s m 2 Temporal scale: : Event based / seasonal Temporal variability in recharge
WTF Method Assumptions Based on premise that rises in ground-water levels in unconfined aquifers are due to recharge, calculated as: Recharge = Sy x (dh( t ) where Sy = specific yield, dh t = water-level rise (difference between peak rise and low point of extrapolated recession curve at the time of the peak) Healy and Cook (2002)
Three Approaches Used to Estimate dh for t WTF Method Graphical extrapolation (manual) Master Recession Curve (MRC) (automated) RISE program (Rutledge, 2003) [does not account for hypothetical recession] Delin et al. (2007)
Graphical Approach to Estimate dh t 5.6 Depth to water, in meters below land surface 5.7 5.8 5.9 6.0 Recharge = (water-level rise) x (specific yield) = 21.0 cm x 0.23 = 4.83 cm 21.0 cm This approach involves more subjectivity than the other WTF approaches. Different users would produce different results. 6.1 Example graph from Delin (1990)
Master Recession Curve Approach Recharge = (sum of rises) x (specific yield) = (R1+R2+ +R n ) x Sy R1 R2 R3 R4 R5 R6 R7 R8 R9 This approach: involves multiple steps; easy to apply; avoids subjectivity; but may include rises unrelated to recharge Delin et al. (2007)
RISE Program Approach to Estimating dh t 31.2 31.4 Recharge = (sum of rises) x (specific yield) = (R1+R2+ +R n ) x Sy Depth to water, meters 31.6 31.8 R150 32.0 32.2 R1 32.4 32.6 R75 32.8 R270 EXPLANATION Daily water-level measurements Water-level rise (example) 33.0 Jan Feb Mar Apr May June July Aug Sept Oct Nov Dec This approach: simple, avoids subjectivity; easy to apply; but makes no allowance for hydrograph recession Al Rutledge, USGS, electronic communication, 2003
Limitations of WTF Method Small scale of influence ET, river fluctuations, or pumping may affect hydrograph Does not account for constant recharge Specific yield estimates uncertain
Ground-Water Age Dating Method Spatial scale: : 1 to 1000s m 2 Temporal scale: : 1 to 50-year, average CANNOT evaluate temporal variability in recharge
Ground-Water Age Dating Method Assumptions Assumes ground-water age and well-depth information can be used to obtain a ground-water velocity (V) Recharge = V x φ where V = ground-water velocity, and φ = average saturated porosity Piston flow is assumed Delin et al. (2000)
Ground-Water Age Dating Age must be determined to within about 1 year before present Estimated using chlorofluorocarbons (CFC s), sulfur-hexafluoride (SF 6 ), and tritium/helium ( 3 H- 3 He) techniques (+ others) Ground-water age is defined as the time elapsed since water entered the aquifer as recharge Equip. used in 3 H/ 3 He sampling Method of sample collection for SF 6 and CFC analysis is easy
Vertical Velocity Determination Ground-Water Age Dating Method Depth below water table, in meters 0 1 2 3 4 5 6 7 8 9 10 11 12 Delin et al. (2000) R1-10 o V v = 0.7 + 0.2 m/yr Water table (Z = 9.5 m) R1 - Lowland Site Data point R1-B 1993, 7 C 1994, 7 C 1993, 9 C 1994, 9 C 0 10 20 30 40 50 CCl2F2 -model recharge age, in years before the sampling date Linear age-depth profile used where only a single ground-water age is available Exponential age-depth profile developed from multiple ages
Limitations of Ground- Water Age Dating Method Ignores horizontal movement below water table Analyses costly: $895 for 3 H- 3 He; $112 (CFCs( & SF 6 ) per sample Specialized labs Delin et al. (2000 and 2007)
RORA Method Spatial scale: : 100 to 1000s km 2 Temporal scale: : monthly to period of record Limited temporal variability in recharge
RORA Method RORA is an automated method for estimating average recharge in a basin Analyze streamflow records using the recession-curve curve-displacement method of Rorabaugh (1960, 1964) RORA accounts for ET effects, underflow, and other losses following a precipitation event Rutledge (1998 and 2000)
Gaging Station/Basin Selection for RORA Method Primary Basin Selection Criteria: - no missing data (1 year required), - basin size less than about 1,300 km 2 (500 mi 2 ) - flow not affected by control structures For Minnesota study: - basin size less than 5,000 km 2 (~2,000 mi 2 ) - 10+ years of record, - evaluated records from 340 basins, - 38 basins met our criteria Rutledge (1998 and 2000)
RORA Recharge Estimate Recharge = 2( Q)K / 2.3026 where Q = difference in the theoretical flows at the critical time, Slope = 1/K K = recession index, time required for ground water discharge to recede by one log cycle after recession becomes linear Rutledge (1998 and 2000)
RORA Limitations Assumes that the streamflow recession is caused by ground-water discharge Slow runoff from snowmelt could be confused for ground-water discharge High-quality, daily streamflow data are required Rutledge (1998 and 2000)
Regionalization of Recharge Estimates Using the Regional Regression Recharge (RRR) Model Spatial variability of recharge
RRR Model Assumptions Spatial variability in recharge can be estimated from: (1)climate, (2)local or basin-scale scale recharge rates, and (3)landscape characteristics Although local-scale recharge estimates could have been used RORA recharge estimates are best suited to regionalization at a State scale Lorenz and Delin (2007)
RRR Methodology Regression equation developed based on: - precipitation data (P),( - growing degree days (GDD( GDD) - recharge (R)( ) from RORA analysis of streamflow, - specific yield (SY Rawls ) derived from STATSGO soils data, as the landscape characteristic R = 14.2 + 0.6459P 0.02231GDD + 7.63SY SY Rawls Final step: : create recharge map of Minnesota using GIS based on a regression analysis of the data sets Lorenz and Delin (2007)
38 Basins met RORA Recharge Selection Criteria Limited coverage imposes some uncertainty in the RRR recharge estimates However, these basins were representative of variability in the State
Mississipp ississippi i Climate Spatial Data Sets Used in RRR Analysis 850 900 Minnesota River GDD, in degrees celsius above 10 C - days less than 800 800-900 900-1000 1000-1100 1100-1200 1200-1300 1300-1400 1400-1500 greater than 1500 0 50 Miles Average annual precipitation, 1971-2000 Average annual growing degree days, 1971-2000
Anoka sand plain EXPLANATION Water Specific yield dimensionless 0.20 to 0.25 0.16 to 0.20 0.11 to 0.15 0.06 to 0.10 0 to 0.05 Unclassifiable 0 50 Miles Landscape Spatial Data Set Used in the RRR Analysis Specific yield from RAWLS analysis of STATSGO soils data
RRR Model Anoka Sand Plain EXPLANATION Water Recharge, in centimeters per year Greater than 330 25.01 to 30 20.01 to 25 15.01 to 20 10.01 to 15 5.01 to 10 0 to 5 Unclassifiable 0 50 Miles Average Annual Recharge to Surficial Materials Lorenz and Delin (2007)
Results and Methods Comparison Delin et al. (2007)
Ground-Water Recharge Normalized as a Percent of Precipitation WTF RISE program results; Bemidji well 310D Delin et al. (2007) Fair correlation R 2 = 0.2696
Relation of RRR to Other Recharge Rates Age-Dating of Ground Water Water-Table Fluctuations Fair correlation R 2 = 0.3277 Very poor correlation R 2 = 0.0008 Delin et al. (2007)
WTF Method Example Plots of Graphical vs. MRC and RISE Approaches 25 140 MRC and RISE approach recharge, cm/yr 20 15 10 5 1:1 correlation line Williams Lake Well wt19 MRC and RISE approach recharge, cm/yr 120 100 80 60 40 20 1:1 correlation line Williams Lake Well wt22 0 0 0 5 10 15 20 25 0 20 40 60 80 100 120 140 Graphical approach recharge, cm/yr Graphical approach recharge, cm/yr Rech. % of precip: 15 % ~100 % UZ Thickness: 5 m 2 m MRC method RISE method
Relation Between WTF Recharge and UZ Thickness Recharge / precipitation, percent Bemidji wells Anomalously large WTF recharge for UZ thicknesses less than 3.5 m Glacial Ridge wells Williams Lake wells 2003 data Graphical approach 23 wells total Delin et al. (2007) Unsaturated zone (UZ) thickness, meters
Effects of Measurement Interval on WTF Recharge Estimates Datalogger, hourly Monthly Bi Monthly Question: How does reduced measurement frequency affect recharge estimates based on the WTF method? Delin et al. (2007)
Effects of Measurement Interval on WTF Recharge Estimates No change in estimated recharge going from hourly to daily measure Hourly / daily Weekly (- 23%) 0-54 % under- estimation of the recharge: going from daily to weekly measurement 18-60 % under- estimation of the recharge: from daily to monthly measurement (- 48%) Monthly Delin et al. (2007) WTF graphical approach; 1993 datalogger data
Temporal Variability in Recharge
Recharge / precipitation, percent Temporal Variability in Recharge, % of precip. 100 80 60 40 20 0 100 80 60 40 20 0 Williams Lake Well wt65 1998 1999 2000 2001 2002 2003 Delin et al. (2007) Princeton Well R1 1992 1993 1994 1995 UZWB method Recharge / precipitation, percent 100 80 60 40 20 0 100 80 60 40 20 0 EXPLANATION BemidjiWell981 1999 2000 2001 2002 2003 Des Moines River Well E04 1999 2000 2001 Water-table fluctuation method: RISE approach MRC approach Graphical approach UZWB results anomalously large most years WTF results fairly consistent Shallow depth to WT causes anomalously large recharge rates at some sites
Temporal Variability in Annual RORA Recharge Knife River near Mora R 2 = 0.52 Delin et al. (2007)
Summary Recharge rates to unconfined aquifers in Minnesota typically are about 10-35 % of precipitation Recharge based on the 3 water-table fluctuation (WTF) approaches are similar, however: MRC estimates are generally greatest RISE estimates are generally lowest Recharge estimation using the WTF method is challenging / inaccurate in areas of shallow depth to water table (< 3.5 m)
High-Quality, Long-Term, Continuous Data are Important Climate Streamflow Ground-water levels
Use Multiple Methods Use Multiple Methods RORA UZWB WTF Groundwater Age Dating
RRR Map RRR Model Apply at different scales based on data availability: Anoka Sand Plain EXPLANATION Water Recharge, in centimeters per year Greater than 3 25.01 to 30 20.01 to 25 15.01 to 20 10.01 to 15 5.01 to 10 0 to 5 Unclassifiable 0 50 Miles Valuable Not appropriate for regional for analyses: use at local e.g. scales input for ground-water flow models (1) climate, (2) local or basin-scale scale recharge rates, and (3) landscape characteristics
Reports Summarizing this Research http://mn.usgs.gov/publicationindex.html Delin, Healy, Lorenz, and Nimmo, 2007, Comparison of localto regional-scale estimates of ground-water recharge in Minnesota, USA: Journal of Hydrology, v. 334, no. 1-2, p. 231-249. Lorenz and Delin, 2007, A regression model to estimate regional ground-water recharge in Minnesota: Ground Water, v. 45, no. 2 Delin and Falteisek, 2007, Ground-water recharge in Minnesota: USGS Fact Sheet 2007-3002, 6 p. http://pubs.usgs.gov/fs/2007/3002/ Delin and Risser, 2007, Ground-water recharge in humid areas of the United States A summary of ground-water resources program studies, 2003-06: USGS Fact Sheet 2007-3007.