Nitrate Trends in Private Well Networks 2008-2018Nitrate Results and Trends in Private Well Monitoring Networks 2008-2018 5 . Figure 1. Southeast VNMN 2018 results by location. As

In accordance with the Americans with Disabilities Act, this information is available in alternative forms of communication upon

request by calling 651-201-6000. TTY users can call the Minnesota Relay Service at 711. The MDA is an equal opportunity

employer and provider.

Minnesota Department of Agriculture

Nitrate Results and Trends in

Private

Well Monitoring Networks

2008-2018

Central Sands Private Well Network and the Southeast

Volunteer Nitrate Monitoring Network

June 2019

Table of Contents

Minnesota Private Well Networks .................................................................................................. 2

Southeast Volunteer Nitrate Monitoring Network 2008-2018 .................................................. 2

Trend Analysis- Southeast VNMN ........................................................................................... 7

Central Sands Private Well Monitoring Network 2011-2018 ..................................................... 9

Trend Analysis-CSPWN .......................................................................................................... 13

Summary ................................................................................................................................... 15

References ................................................................................................................................. 16

PRIMARY AUTHORS

Kim Kaiser, Brennon Schaefer, and Bill VanRyswyk

REVIEWERS AND CONTRIBUTORS

Larry Gunderson, and Jaime Nielsen

FUNDING

Clean Water Fund

Nitrate Results and Trends in Private Well Monitoring Networks 2008-2018

2

MINNESOTA PRIVATE WELL NETWORKS

The Minnesota Department of Agriculture (MDA) has been working closely with other state and

local agencies on developing regional private well nitrate-nitrogen (nitrate) networks.

Homeowner volunteers are the cornerstone of the private well networks. Homeowners collect

their own water sample and send it by mail to be tested by a laboratory or a county

collaborator using MDA equipment at no cost to the homeowner. This method has been

developed from years of collaboration with other state and local agencies through pilot projects

testing different methods of collection and sample delivery. The MDA currently coordinates

two volunteer nitrate monitoring networks with assistance from other state agency and local

partners, the Southeast Minnesota Volunteer Nitrate Monitoring Network and the Central Sands

Private Well Network. The two networks are discussed in greater detail below.

SOUTHEAST VOLUNTEER NITRATE MONITORING NETWORK 2008-2018

Drinking water quality is a concern across southeastern Minnesota where nitrate loading to the

subsurface can be significant and hydrogeologic sensitivity is highly variable within short

distances. In 2008, the Southeast Minnesota Water Resources Board (SEMNWRB) and several

state agency partners (MPCA, MDA, MDH) began collecting samples and data from the

Southeast Volunteer Nitrate Monitoring Network (VNMN). This region was selected as a pilot

because of its vulnerable and complex geology.

This pilot began as a network of 675 private drinking water wells representing a stratified-

random distribution across the nine counties (Dodge, Fillmore, Goodhue, Houston, Mower,

Olmsted, Rice, Wabasha and Winona) of southeastern Minnesota. The wells included in the

network represent groundwater nitrate concentration data from several aquifers.

Before data collection began, well network coordinators (county staff) enrolled volunteers (well

owners) into the program by collecting detailed information about well location, well

construction, and nearby nitrate sources. Between February 2008 and August 2018, 13

sampling events occurred representing approximately 5,421 samples. During this period, the

percentage of wells exceeding the Health Risk Limit (HRL) for each sampling event ranged

between 7.5 and 14.6 percent (Table 1). The most recent sampling results from 2018 are

presented in Figure 1.


3

Table 1. Summary of the 2008 through 2018 long- term Southeast VNMN results.

Year Total Wells

50th Percentile (Median)

90th Percentile % of Wells ≥10 % of Wells <10

Nitrate-N (mg/L)

2/1/2008 519 0.3 12.0 14.6% 85.4%

8/1/2008 510 0.3 11.0 11.4% 88.6%

2/1/2009 494 0.2 11.0 11.1% 88.9%

8/1/2009 471 0.3 11.5 11.0% 89.0%

8/1/2010 422 0.7 9.5 9.2% 90.8%

8/1/2011 428 0.6 10.0 10.3% 89.7%

8/1/2012 411 0.4 8.8 7.5% 92.5%

8/1/2013 315 0.1 8.8 8.3% 91.7%

8/1/2014 361 0.2 9.4 8.9% 91.1%

8/1/2015 373 0.2 9.4 8.8% 91.2%

8/2/2016 387 0.3 10.8 10.9% 89.1%

8/1/2017 341 <0.25 10.1 10.0% 90.0%

8/1/2018 389 0.3 9.5 9.0% 91.0%

The 2018 nitrate results by aquifer are presented in Table 2. The highest median and upper

percentile concentrations were observed in wells open to the Quaternary water table aquifer

followed by the Prairie du Chien-Jordan and the Prairie du Chien aquifers. The Quaternary

aquifer consists of unconsolidated sediments and generally represents the shallowest aquifers

in this study with a mean depth of 56 feet. The Prairie du Chien-Jordan and the Prairie du Chien

aquifers are limestone and sandstone bedrock aquifers that are generally deeper with a mean

well depth of 315 and 300 feet, respectively.


4

Table 2. Southeast VNMN nitrate concentrations by aquifer in 2018.

Aquifer Number of Wells

Mean Depth

(ft)

Values Percentiles

Min Max Mean Median 75th 90th 95th 99th

Nitrate-N (mg/L)

Not Available 83 NA <0.03 29.4 3.9 0.74 7.0 11.8 16.8 26.5

Quaternary Water Table 11 56 <0.03 23.6 4.8 2.70 4.8 15.2 22.9 23.6

Quaternary Undifferentiated 8 84 <0.03 11.7 1.6 <0.03 0.6 8.5 11.7 11.7

Quaternary Buried 8 142 <0.03 0.90 0.2 <0.03 0.3 0.8 0.9 0.9

Cedar Valley Wapsipinicon 16 139 <0.03 7.24 0.8 <0.03 0.1 3.3 6.1 7.2

Spillville 11 149 <0.03 9.62 1.5 <0.03 1.9 6.6 9.4 9.6

Maquoketa-Dubuque 9 174 <0.03 4.59 1.6 0.34 4.1 4.5 4.6 4.6

Galena 22 148 <0.03 24.8 3.0 <0.03 4.8 8.9 16.6 24.8

St. Peter 29 237 <0.03 22.2 2.2 0.6 1.9 5.9 10.5 22.2

Prairie du Chien 78 300 <0.03 49 4.4 1.3 7.7 11.3 14.2 40.3

Prairie du Chien-Jordan 17 315 <0.03 20.6 6.1 6.6 9.1 14.8 19.0 20.6

Jordan 49 365 <0.03 24.9 3.3 0.2 6.4 8.3 13.0 24.9

Saint Lawrence-Tunnel City 5 320 0.066 10.4 3.6 2.7 6.2 10.4 10.4 10.4

Tunnel City-Lone Rock 25 373 <0.03 3.64 0.7 <0.03 0.5 3.5 3.6 3.6

Tunnel City-Wonewoc 6 410 <0.03 3.15 0.7 <0.03 1.0 2.9 3.2 3.2

Wonewoc-Eau Claire 12 338 <0.03 7.13 1.0 0.03 1.1 3.5 6.6 7.1

Total 389 269 <0.03 49 3.16 0.3 4.7 9.5 13.0 24.3


5

Figure 1. Southeast VNMN 2018 results by location.

As Table 1 indicates, the number of homeowners (wells) participating in the project is variable

with time, ranging from 519 in 2008 to 389 in 2018. To investigate the possible impact of this

change, a comparison of depth and age (Tables 3 and 4) was performed for the Southeast

VNMN, for wells that had known information. The results suggest that, with respect to well

characteristics, there did not appear to be large differences in the sampled wells over the years

as the depths and ages did not change significantly (p value= 0.933 and 0.817 respectively).

Therefore, it is reasonable to say that changes in water quality are not being strongly affected

by the small changes in well characteristics of the network.


6

Table 3. Summary of well depth of submitted samples by year in Southeast VNMN.

Date Sampled

Number of Wells

Number of Wells with

Depth

Number of Wells without

Depth

% of Wells without Depth

Min Depth

(ft)

Max Depth

(ft)

Mean Depth

(ft)

Median Depth

(ft)

2/1/2008 519 451 68 13.1% 18 725 254 220

8/1/2008 510 455 55 10.8% 18 725 256 225

2/1/2009 494 442 52 10.5% 18 725 257 223

8/1/2009 471 423 48 10.2% 18 725 257 225

8/1/2010 422 394 28 6.6% 18 725 261 230

8/1/2011 428 393 35 8.2% 20 725 267 240

8/1/2012 411 368 43 10.5% 18 725 264 230

8/1/2013 315 291 24 7.6% 18 725 266 230

8/1/2014 361 328 33 9.1% 18 725 256 220

8/1/2015 373 338 35 9.4% 18 725 257 220

8/2/2016 387 347 40 10.3% 20 725 268 242

8/1/2017 341 293 48 14.1% 18 725 265 240

8/1/2018 389 341 48 12.3% 20 725 269 245

Table 4. Summary of well age of submitted samples by year in the Southeast VNMN.

Date Sampled

Number of Wells


Dates


Dates

% of Wells without

Dates

Min Year

Max Year

Mean Year

Median Year

2/1/2008 519 415 104 20.0% 1870 2007 1969 1973

8/1/2008 510 418 92 18.0% 1870 2008 1969 1973

2/1/2009 494 406 88 17.8% 1870 2008 1970 1974

8/1/2009 471 388 83 17.6% 1870 2008 1970 1973

8/1/2010 422 363 59 14.0% 1870 2010 1970 1973

8/1/2011 428 361 67 15.7% 1870 2010 1970 1975

8/1/2012 411 337 74 18.0% 1870 2011 1970 1974

8/1/2013 315 273 42 13.3% 1870 2013 1971 1975

8/1/2014 361 304 57 15.8% 1870 2013 1970 1974

8/1/2015 373 315 58 15.5% 1870 2012 1970 1973

8/2/2016 387 320 67 17.3% 1880 2016 1971 1975

8/1/2017 341 268 73 21.4% 1900 2016 1973 1976

8/1/2018 389 308 81 20.8% 1870 2016 1973 1977


7

TREND ANALYSIS- SOUTHEAST VNMN

Groundwater nitrate concentration measurements are often not normally distributed data sets.

Consequently, using parametric trend analysis is not appropriate. For this analysis, a

nonparametric Mann-Kendal test for the time trend analysis (Helsel and Hirsh, 2002) was

determined to be the most appropriate.

There were two sampling events early on in the network that occurred in February; otherwise,

sampling occurred in August of each year. The February and August sampling events in 2008

and 2009 were evaluated to see if there were any differences between the events in these

years. A boxplot was produced (Figure 2) and a nonparametric Kruskal-Wallis comparison test

was performed (Table 5) to assist with this evaluation. It was found that there were no

significant differences between the events in 2008 and 2009, so only the August event data

from each year was utilized. This also permitted traditional trend testing, where time steps

between the data being analyzed were consistent.

Figure 2. Boxplot of nitrate concentrations from the Southeast VNMN during the February and

August sampling events in 2008 and 2009.


8

Table 5. Comparison test results for the Southeast VNMN data during the February and August

sampling in 2008 and 2009.

Sample Events Tested Kruskal-Wallis Test

p-Value Result

February 2008 & August 2008 0.875 No Difference

February 2009 & August 2009 0.658 No Difference

No statistically significant trend (H0: No Trend vs. H1: Two-Sided Trend or H0: No Trend vs H1

Downward Trend) was found for any of the percentile data (Table 6) for the period from 2008

through 2018. Figure 3 shows the percentiles for the nitrate concentration data over time

plotted with binomial fit lines, which are used in nonparametric methods to illustrate the

central tendency of data over time (Helsel and Hirsch, 2002).

Table 6. Trend analysis for the Southeast VNMN data.

Percentile

Mann-Kendal Test Mann-Kendal Downward Trend

p-Value (95th confidence level) p-Value (95th confidence level)

50th 0.159 No Trend 0.079 No Trend




9

Figure 3. Binomial lines fit to nitrate percentile concentrations from the Southeast VNMN.

CENTRAL SANDS PRIVATE WELL MONITORING NETWORK 2011-2018

Due to the success of the Southeast VNMN, as well as the availability funding from the Clean

Water Legacy Amendment, the MDA launched a similar project in the Central Sands area of

Minnesota, which includes 14 counties, in 2011. The MDA determined that, because high levels

of nitrate had been measured in MDA’s Central Sands monitoring wells, it was important to

expand nitrate monitoring to private drinking water wells to determine if the concentrations

were similar to concentrations found in MDA’s monitoring wells in this area. In the spring of

2011, the MDA began the Central Sands Private Well Monitoring Network (CSPWN). The goals of

this project were to evaluate nitrate concentrations in private wells across the Central Sands

region and assess nitrate concentration trends over time using a representative subset of this

data.

The CSPWN was designed as an unaligned grid across the 14 county area. Homeowners from

within the grids in agricultural areas were randomly invited to participate in the network. By

July 1, 2011, the MDA had analyzed 1,555 samples for nitrate (MDA 2012). Over 88 percent of

the wells sampled had concentrations below 3 mg/L nitrate-N, 6.8 percent of the wells ranged


10

from 3-10 mg/L of nitrate-N and 4.6 percent of the wells were greater than the nitrate-N HRL

(Table 7). These results were similar to findings from a 2010 U. S. Geological Survey (USGS)

report on nitrate concentrations in private wells in the glacial aquifer systems across the upper

United States (Warner and Arnold 2010). The USGS report found that less than 5 percent of

sampled private wells had concentrations greater than or equal to 10 mg/L nitrate-N. The 2011

CSPWN results indicated nitrate concentrations varied widely over relatively short distances.

The USGS report on glacial aquifer systems indicated similar findings related to spatial

variability in nitrate concentration.

Table 7. Summary of CSPWN 2011 nitrate data.

Number of Samples

Min (mg/L)

Median (mg/L)

75 th

Percentile (mg/L)

90 th

Percentile (mg/L)

Maximum (mg/L)

% ≤ 3 (mg/L)

% 3<10 (mg/L)

% ≥10 (mg/L)

1,555 <0.03 0.01 0.66 4.15 31.9 88.6% 6.8% 4.6%

Over 500 homeowners volunteered to participate in long-term annual sampling of their private

wells. Participation in sampling decreased over time, similar to the Southeast VNMN; however,

results from 2011 through 2018 have indicated minimal variation in nitrate concentration over

time. The 2018 results indicate 3 percent of the wells had nitrate concentrations greater than

or equal to 10 mg/L nitrate-N (Table 8). Overall, 95.5-97.8 percent of wells in the CSPWN have

been below the 10 mg/L HRL for nitrate-N. The highest nitrate result was detected in a well

with an unknown aquifer source, however, elevated concentrations can be found throughout

the water table to the buried quaternary aquifers (Table 9). The most recent sampling results

from 2018 are presented in Figure 4. For further information on this sampling project, see

http://www.mda.state.mn.us/en/protecting/cleanwaterfund/gwdwprotection/characterizingni

trates.aspx

http://www.mda.state.mn.us/en/protecting/cleanwaterfund/gwdwprotection/characterizingnitrates.aspx

http://www.mda.state.mn.us/en/protecting/cleanwaterfund/gwdwprotection/characterizingnitrates.aspx


11

Table 8. CSPWN nitrate results summary 2011 through 2018.

Year Total Wells 50th Percentile

(Median) 90th

Percentile % of Wells ≥ 10 % of Wells < 10

Nitrate-N (mg/L)

2011 534 <0.03 3.3 3.9% 96.1%

2012 506 0.20 3.6 3.2% 96.8%

2013 487 0.20 3.6 2.7% 97.3%

2014 432 <0.03 3.2 3.5% 96.5%

2015 402 <0.03 3.5 4.5% 95.5%

2016 397 <0.03 3.0 3.5% 96.5%

2017 367 <0.03 3.3 2.2% 97.8%

2018 338 <0.03 3.0 3.0% 97.0%

Table 9. Nitrate results summary by aquifer in the CSPWN in 2018.

Aquifer Total Wells

Mean Depth

(ft)

Values Percentiles

Min Max Mean Median 75th 90th 95th 99th

Nitrate-N (mg/L)

NA 139 NA <0.03 35.70 1.35 <0.03 0.84 4.31 6.50 23.95

Quaternary Water Table 26 66 <0.03 26.70 1.64 <0.03 0.06 1.35 14.94 26.70

Quaternary Buried Unconfined 7 77 <0.03 6.79 1.69 <0.03 3.42 6.29 6.79 6.79

Quaternary Buried Undifferentiated 1 133 <0.03 <0.03 <0.03 <0.03 NA NA NA NA

Quaternary Undifferentiated 5 146 <0.03 0.11 0.02 <0.03 0.03 0.11 0.11 0.11

Quaternary Buried Artesian Aquifer 158 105 <0.03 33.70 0.88 <0.03 0.05 2.37 3.67 14.77

Hillman Tonalite 1 128 0.078 0.08 0.08 0.08 NA NA NA NA

Precambrian Rock Undifferentiated 1 353 <0.03 <0.03 <0.03 <0.03 NA NA NA NA

Total 338 102 <0.03 35.70 1.13 <0.03 0.16 3.05 6.10 23.00


12

Figure 4. CSPWN 2018 Results.

As Table 8 indicates, the number of homeowners (wells) participating in the project is variable

with time, ranging from 534 at its highest in 2011 to 338 at its lowest in 2018. To investigate the

possible impact of this change, a summary comparison of well depth and age (Tables 10 and 11)

was also performed for the CSPWN, for wells that had known information. The results suggest

that, with respect to well characteristics, there did not appear to be large differences in the

sampled wells over time as the depths and ages did not change significantly (p value= 0. 982


13

and 0.957 respectively). Therefore, it is reasonable to assume that changes in water quality are

not being strongly affected by the changing network population overtime.

Table 10. Summary of well depths for submitted samples by year in the CSPWN.

Date Sampled

Number of Total Wells


Depth


Depth

% of Wells without Depth

Min Depth

(ft)

Max Depth

(ft)

Mean Depth

(ft)

Median Depth

(ft)

2011 534 306 228 43% 15 400 106 90

2012 506 294 212 42% 15 400 105 90

2013 487 283 204 42% 15 400 105 87

2014 432 258 174 40% 15 400 104 86

2015 402 243 159 40% 15 400 106 90

2016 397 237 160 40% 15 400 103 86

2017 367 210 157 43% 15 356 103 85

2018 338 199 139 41% 15 400 102 84

Table 11. Summary of well construction dates for submitted samples by year in the CSPWN.

Date Sampled

Number of Total Wells


Dates

Number of wells without

Dates

% of wells without

Dates

Min Year

Max Year

Mean Year

Median Year

2011 534 305 229 43% 1966 2011 1997 1999

2012 506 293 213 42% 1966 2011 1997 1999

2013 487 282 205 42% 1966 2013 1997 1999

2014 432 257 175 41% 1966 2011 1997 1999

2015 402 242 160 40% 1966 2015 1997 1999

2016 397 236 161 41% 1966 2013 1997 1999

2017 367 210 157 43% 1966 2013 1997 1999

2018 338 198 140 41% 1972 2013 1998 1999

TREND ANALYSIS-CSPWN

The non-parametric time series analysis for the CSPWN indicates that there is not enough

information to reject the null hypothesis (H0: No Trend vs. H1: Two-Sided Trend) that there is

no statistically significant trend in the nitrate percentiles (50th, 75th, and 90th) through this test

(Table 12). However, there is enough evidence for a downward trend in the 90th percentile (H0:

No Trend vs. H1 Downward Trend) through the downward trend test. Figure 5 presents the

percentiles for the nitrate concentrations plotted overtime with binomial fit lines, which

indicate the central tendency of the data (Helsel and Hirsch, 2002).


14

Table 12. Trend analysis for the CSPWN 2011 through 2018.

Percentile Mann-Kendal Trend (two-sided) p-Value (95th confidence level)

Mann-Kendal Downward Trend p-Value (95th confidence level)



90th 0.097 No Trend 0.048 Downward Trend

Figure 5. CSPWN binomial fit lines.


15

SUMMARY

Although there can be variability in the sampled population and nitrate concentration in

individual wells from year to year, the regional data presented in this report is useful for

evaluating long term trends. Analysis of the data from these private well monitoring networks

indicates a concern for nitrate in groundwater from vulnerable aquifers in central and

southeast Minnesota.

Nitrate concentration data from the Southeast Volunteer Nitrate Monitoring Network (2008-

2018) and the Central Sands Private Well Network (2011-2018) were analyzed for statistical

trend. In order to ensure all years were treated equally, the February sampling periods from

years 2008 and 2009 were removed prior to the nitrate trend analysis in the Southeast

Volunteer Nitrate Monitoring Network. The data from the August event sampling from each

year (eleven years of monitoring) showed that there was no statistically significant trend for the

50th percentile (median), 75th or 90th percentile data in the Southeast Volunteer Nitrate

Monitoring Network. Data from the Central Sands Private Well Network showed a slight

statistically significant downward trend in the 90th percentile data based on eight years of

monitoring in that network.

Over time network participation did decline in both networks, so additional analysis of well

depths and ages was completed to evaluate the potential impact that these changes may have

had on the sampled populations over time. Based on an analysis of wells with known

characteristics (well depth and age) there were no significant differences in either network.

Reasons volunteers leave the networks may include selling their home, apathy based on not

seeing a change in nitrate results over time, replacing an old well with a new well, and

homeowners are too busy. Future work will include emphasis on keeping volunteers engaged

and participating in the network on an annual basis to keep the network consistent.


16

REFERENCES

Helsel and Hirsch. 2002. USGS Statistical Methods for Water Resources. TWRI 04-A3.

Minnesota Department of Agriculture. 2012. Central Sands Private Well Network 2011 Current

Nitrate Conditions Summary.

http://www.mda.state.mn.us/protecting/cleanwaterfund/gwdwprotection/~/media/Fil

es/protecting/cwf/cspwnsummary.ashx (accessed November 10, 2016).

Warner, K.L., and T.L. Arnold. 2010. Relations that Affect the Probability and Prediction of

Nitrate Concentrations in Private Wells in the Glacial Aquifer System in the United

States: A USGS National Water Quality Assessment Program Scientific Investigations

Report.

http://www.mda.state.mn.us/protecting/cleanwaterfund/gwdwprotection/~/media/Files/protecting/cwf/cspwnsummary.ashx

http://www.mda.state.mn.us/protecting/cleanwaterfund/gwdwprotection/~/media/Files/protecting/cwf/cspwnsummary.ashx

Documents

Nitrate Trends in Private Well Networks 2008-2018Nitrate Results and Trends in Private Well Monitoring Networks 2008-2018 5 . Figure 1. Southeast VNMN 2018 results by location. As