1 of 32

Improved understanding of groundwater age and nitrate trends in southeast Minnesota

2 of 32

Kuehner, K.J.¹, Runkel, A.C.² & Barry, J.D.³Informing nitrate concentration trends: estimating groundwater residence time in a karstic, multiaquifer system using anthropogenic tracers (Minnesota, USA). Hydrogeol J (2025). https://doi.org/10.1007/s10040-024-02871-2

1. Minnesota Department of Agriculture, Pesticide and Fertilizer Management Division

2. Minnesota Geological Survey, University of Minnesota

3. Minnesota Department of Natural Resources, Division of Ecological and Water Resources

3 of 32

Acknowledgements

Co-authors Tony Runkel-MGS and John Barry-DNR
DNR- Holly Johnson, Susan Montgomery
UM-John Nieber, Joe Magner, Kerry Holmberg, Bruce Wilson
USGS- Jared Trost, Andrew Berg, Colin Livdahl, Kirsten Faulkner
MDA Monitoring Unit
Root River Field to Stream Partnership

MGS-Bob Tipping, Julia Steenberg, Andrew Retzler
Reid Christensen, Mitchell Schaps
SE MN Water Analysis Laboratory
Caitlin Meyer, Kim Crawford, Terry Lee
MPCA and DNR monitoring programs

4 of 32

Key questions

1. Where are nitrates trending?

2. How do we interpret…..especially within the context of the complex hydrogeology of southeast Minnesota?

3. How soon can we measure water quality improvements while also acknowledging challenges?

Springs

Streams

Wells

5 of 32

Nitrates in private wells�cropland and vulnerable groundwater

Source: MDA 2013-2019

Statewide

9.1% above 10 mg/L

32,217 wells sampled

Southeast Driftless Area

13.6% above 10 mg/L

6 of 32

Nitrate trends in SE MN�Six-county area

1,156 sites, nearly 35,000 nitrate samples
Primarily reflects years 2000-2021
Significant when p value <0.05 using nonparametric methods

Median nitrate (mg/)

8.2

5.9

3.8

7 of 32

Nitrate trends in springs

Differences are not attributable to land use alone
Hypothesized that variable mixing of local and regional groundwater with different ages is a key factor.
Poor understanding of groundwater residence time

Avg. 5.5 mg/l

4% increase(0.19/yr)

Avg. 10.0 mg/l

No trend

Avg. 9.6 mg/l

-1.2% (-0.13/yr)

Thiel-Sen (Ktau)

Lowess smoother

Hwy 76 (28A076)

PdC/Jordan

Fountain West (23A037)

Galena Fm. Water table

Amherst (23A539)

PdC water table

8 of 32

Big Spring�Beaver Creek Valley State Park

“Water bubbling out of the ground is a natural wonder. Where does it come from? How long has it been in the ground?”

“The water coming from Big Spring has not been in the ground for a very long time- most likely less than a year”……

Prairie du Chien and Jordan aquifers

9 of 32

Alachlor ESA in springs

9

Alachlor has not been used for over 25 years, but is still increasing or near peak.

Goodhue Co.

Fillmore Co.

Houston Co.

Fillmore Co.

The difference in years of the alachlor usage and ESA concentration regression models was used to estimate the residence time.�

The difference between the predicted x-intercept of the alachlor ESA concentration year and the normalized alachlor use year was considered the estimated residence time in years. Alachlor use was normalized by dividing the mean alachlor ESA concentrations by the mean alachlor use during the trend period. In many cases, the slopes of the fitted lines were not parallel, which can result in exaggerated differences between the two intercepts. To estimate a more centralized difference between the fitted lines, the Excel® add-in Solver was used to identify the optimal combination of values that minimized the residual sum of squares (RSS) or error of the model. This approximates the difference between the center of the fitted lines rather than one specific point at which they cross the x-axis. The year where the RSS was at a minimum or least error represented the best approximation for the difference between the fitted lines.

Method 2, or the non-linear intersect method, was used for instances when alachlor ESA concentrations were asymptotic near the detection limit. The point at which the lower 95% confidence interval of the non-linear model intersected with the lab detection limit was the estimated x-intercept year. Non-linear models that were evaluated included quadratic, exponential, and asymptotic convex, with the latter providing the best-fit results. This non-linear solving method was optimal for all Galena springs because alachlor ESA concentrations were consistently decreasing and plateauing near the detection limit and exhibiting long-concentration tails, as demonstrated by Burr Oak Spring in Fig. 4c.

Method 3, or the peak-to-peak approach, was used for sites like Big Spring and Peterson Spring (Fig. 4d), which emanate from the top of the Jordan and Lone Rock aquifers, respectively. The year in which alachlor ESA concentrations had presumably peaked was subtracted from the year alachlor use had peaked, which was quantified by solving for the second derivative of the quadratic equation. Additional method details are provided in ESM 3.

10 of 32

Commonly used herbicide from the 1970’s-1990’s.
Use fell dramatically by 2000.

Ideal tracer characteristics

Well-defined start, peak and end use patterns.
Alachlor ethane sulfonic acid (ESA) has conservative/stable properties once in groundwater.

Alachlor and Methods

Estimated alachlor sales/use in Minnesota

22 years

Methods

The difference in years between the fitted regressions for normalized alachlor use and alachlor ESA concentrations was considered the estimated residence time in years.

11 of 32

Study area: SE MN Driftless Area

Driftless Area

Primarily four different aquifers

30 spring and well study sites

Age dating methods

Alachlor ESA tracer
Atmospheric tracer (USGS)
14C (UM)

The springs and wells evaluated in this study are in the Driftless Area of southeast Minnesota. The Driftless Area also includes parts of Wisconsin, Iowa, and Illinois, covering an area of 62,000 sq km (24,000 sq mi) in the upper Midwest.

For reference, inset A shows the driftless area in relation to North America.

Inset B shows the portion in the state of Minnesota.

The counties outlined in black show the counties our study focused on.

The dark green hues across the Prairie du Chien Plateau are deeply incised valleys.

Map numbering of springs (1-16) and wells (17-30) corresponds to the numbering used in the paper’s figures and tables.

Due to the thin cover of mostly sandy to silty unconsolidated sediment (generally <15 m, 50 ft), the landscape largely reflects the topography of the bedrock surface.

12 of 32

Geologic setting

Study Springs

Area 1

Area 2

Area 3

13 of 32

Geology

Galena Group

Group A- Burr Oak, Fountain E., Fountain W, Rainy

Group B- Amherst, Lanesboro SFH, Cold S., Spring Ck, Big Spring, Hwy 76

Group C- Canfield, Crystal SFH, Peterson SFH

Group A

Group B

Group C

Decorah Shale Aquitard

Area 2

Area 3

14 of 32

Groundwater residence time, 13 study springs

14

‘Shallow’ Springs

‘Deeper’

Springs

‘Shallow’ Springs

‘Deeper’

Springs

Group A

Group C

Group B

12 -41 years

Ktau=0.75, p<0.0001

RT = -3.41590 + 0.33 * Depth

RT increases 10 years every 100’ of additional depth

15 of 32

Groundwater age, alachlor ESA and trends

15

‘Shallow’ Springs

‘Deeper’

Springs

‘Shallow’ Springs

‘Deeper’

Springs

Big Spring

16 of 32

Groundwater age, nitrates, and trends

16

‘Shallow’ Springs

‘Deeper’

Springs

‘Shallow’ Springs

‘Deeper’

Springs

17 of 32

As residence time increases, nitrate trends increase

27 spring and well sites, alachlor ESA and atmospheric methods

18 of 32

Age date results (all methods)

Post 1953 mixtures

n=32

Pre & Post 1953 mixtures

n=14

Source: Kuehner et al. 2025 and Faulkner et al. 2022

1-4 decades

(modern)

Source data: Alexander & Alexander, 2018

n=135, 8-county area

Wells

very old/pre-modern

3-8 decades

(modern + premodern)

Springs and Wells

10,000

40,000

25,000

7,343 years

Wells

19 of 32

Historical perspective: land use and precipitation

Precipitation trends (younger water influence)

+20%

+10%

+5%

-2 to -3%

20 of 32

Stream nitrate trends

Driftless Area- 21 sites in SE MN and 22 sites in northeast Iowa
Extensive Glacial Till, 18 sites in south central Minnesota

Extensive

Glacial Till

(n=18)

Trend Period

2000-2021

2012-2021 (very wet)

21 of 32

Stream nitrate trends�in tile drained glacial till settings

-1.4%

-0.23 mg/L/yr

No Trend

2000-2021

2012-2021

-6.9%

-1.3 mg/L/yr

-4.8%

-0.5 mg/L/yr

Seven Mile Creek-Nicollet County

Blue Earth near Rapidan-Blue Earth County

22 of 32

Remarkably wet in southern MN

1895

5 of the 12 wettest years in the past 128 occurred consecutively between years 2016-2020

2024

1930 1936

Monthly Palmer Index

Source data: NOAA

2000

1978

2018

60.21” Harmony (state record)

2016 Waseca 53.8”

23 of 32

Stream nitrate trends

2000-2021

2012-2021 (very wet)

Regardless of historically high precipitation, increasing trends in Area 3 (PdC plateau) largely remained unchanged

Area 1

Area 2

Area 3

Area 2

Area 3

MN

IA

MN

IA

24 of 32

Improved nitrogen management

Plant genetics- rooting systems, seeding populations, longer relative maturity, more N uptake later in the season, N use efficiency and sulfur

Application methods and technology- split applications, GPS, nitrogen stabilizers and controlled release products, incorporation and placement of fertilizers and manure

Soil Health and perennial cropping initiatives- cover crops, Forever Green Initiative

Regulations- feedlot rules, nutrient management plans, no fall commercial nitrogen

Programs/funding- Clean Water Fund, Ag Water Certification Program

Future advancements?

Winfield United

25 of 32

Groundwater age conceptual model

Area 2

Area 3

26 of 32

Nitrate conceptual model

Typically >4

Typically <1

Area 2

Area 3

Higher risk of elevated nitrate and increasing trends

27 of 32

Example well scenarios

Area 2

Area 3

d

b

c

a

b. 500’ well, cased and grouted in 1998, no nitrate, millennia aged water

a. 150’ well, pre 1974 construction, high nitrate, mixture of <20 year old water. No trend or occasional decreasing

c. 280’ well, cased and grouted in 2004, medium nitrate, mixture of >40 year old water, high risk of increasing trend

d. 550’ well, cased and grouted in 2018, no nitrate, millennia aged water

28 of 32

Residence time (modern fraction) ranged from 10 to 40 years old in many shallow springs and wells to thousands in deeper aquifers.��Younger groundwater mixtures typically have elevated nitrate, and trends are relatively stable or declining.

29 of 32

At many sites with older groundwater mixtures, nitrates were lower, but levels are slowly increasing. ��This trend will continue until an equilibrium is reached between current and historical land use practices (especially in the PdC plateau).

30 of 32

Although it may take decades to measure, the cumulative effect of today’s best practices will help reduce nitrate loading to groundwater over the long term.

Value of a collaborative, interdisciplinary approach.

Importance of supporting and maintaining long-term water quality datasets.

31 of 32

Big Spring�Beaver Creek Valley State Park

“Water bubbling out of the ground is a natural wonder. Where does it come from? How long has it been in the ground?”

“The water coming from Big Spring has not been in the ground for a very long time- most likely less than a year”……

The water coming from Big Spring is likely a mixture of water that dates back at least three decades…..

32 of 32

Thank You

Kevin Kuehner

Hydrologist, CCA

kevin.kuehner@state.mn.us