Contributed papers from the 6th Soil and Water Management Field Day 18 July 2017, Lamberton, Minnesota

6th Soil and Water Management Field Day

University of Minnesota Southwest Research & Outreach Center Lamberton, Minnesota 18 July, 2017

Edited by Jeff Strock University of Minnesota – Southwest Research & Outreach Center, Lamberton Minnesota Andry Ranaivoson University of Minnesota – Southwest Research & Outreach Center, Lamberton Minnesota

FORWARD

The Corn Belt, including Minnesota, is a highly productive agricultural region. This is in large part due to a favorable environment, which includes fertile soils, abundant rainfall, and ample solar radiation. However, the region has begun to experience more volatile and variable weather patterns and conditions, including increased frequency and intensity of severe rainfall, more frequent flooding, along with extended periods of drought, often in the same year. These climate trends have an impact on corn-based cropping systems. Consequently, producers are seeking ways to ensure continued high levels of crop productivity while minimizing negative off-site environmental impact.

The 6th Soil and Water Management Field Day is an event to bring producers, researchers, contractors, state and federal agency staff, policy makers, and conservation groups together around a common issue: soil and water management for productivity, profitability and environmental benefit. The overarching objectives of the 6th Soil and Water Management Field Day were to (1) provide a forum for researchers to share the results of on-going research with stakeholders, (2) provide an opportunity for stakeholders to participate in educational activities, and (3) provide stakeholders an opportunity to provide input into efforts addressing soil, water, and nutrient management issues.

The 6th Soil and Water Management Field Day was designed to highlight progress on soil and water management research and is an example of inter-institutional collaboration. The focus of this year’s event is primarily on research being conducted as part of five past or currently funded research projects. This year’s program features a variety of presentations including bioreactors for mitigating nitrogen and phosphorus from tile drainage, managing infield water using cover cropping, side-inlet controls for water quality protection, and supplemental irrigation for enhancing corn and soybean yield.

The project titled “Quantifying hydrologic impacts of drainage under corn production systems in the upper Midwest” is supported by the Minnesota Corn Research and Promotion Council under award number 4116-14SP. This research uses a combination of field research and modeling to quantify the water balances of corn production systems, with and without the presence of subsurface drainage, along a precipitation gradient from eastern South Dakota to south central Minnesota. Understanding the hydrologic response of drainage and crop water consumption at both the field and watershed scale will help corn growers be economically competitive while also informing development of tools and management approaches that can minimize their environmental impact under various climate conditions.

The project titled “Managing Water for Increased Resiliency of Drained Agricultural Landscapes” is supported by the National Institute of Food and Agriculture, U.S. Department of Agriculture, under award number 2015-68007-23193. This project, also known as "Transforming Drainage", is a collaborative effort aimed at addressing important drainage management questions through the assessment and development of new water storage practices and technologies.

The project titled “Nutrient removal in agricultural drainage ditches” was supported by the State of Minnesota Clean Water Land and Legacy Fund contract number 63906. The scope of this i project was to investigate the potential of treating subsurface drainage water before it enters a ditch or stream. The outcome of this research was the development of a prototype modular bioreactor capable of removing nitrogen and phosphorus from subsurface drainage water.

Two recently funded projects carry on and expand the scope of some of the fore mentioned research. The project titled “Integrated landscape management for agricultural production and water quality” was supported by the State of Minnesota Clean Water Land and Legacy Fund contract number 123945. This project carries on the modular bioreactor project and expands the scope to other in-field, edge-of-field and in-stream practices for mitigating nitrogen and phosphorus from agricultural runoff and subsurface drainage. The project titled “Investigation of a novel approach to mitigate nitrogen and phosphorus from tile drainage” is being supported by the University of Minnesota Water Resources Center and the US Geological Survey. This project carries on the modular bioreactor research.

The proceedings and presentations from the Field Day include seven papers and seven presentations, which discuss research projects conducted by scientists from the University of Minnesota and South Dakota State University.

Jeffrey S. Strock, University of Minnesota – Southwest Research and Outreach Center

ii

ACKNOWLEDGMENTS

There are many people, too numerous to mention individually, that were instrumental in helping organize, coordinate, and execute this 6th Soil and Water Management Field Day. The first and most important thank you is extended to the many participants of the 6th Soil and Water Management Field Day. A second expression of gratitude goes to the presenters who helped make this program successful by sharing the results of their research and contributing to this proceedings document. Finally, a heartfelt thank you goes the staff at the Southwest Research and Outreach Center, especially Andry Ranaivoson, Emily Evans, Gretchen Thillen, Mark Coulter, Cole Werner, and Tyler Vogel for their help to coordinate the overall logistics, tours, publicity, and this proceedings document.

DISCLAIMER

The information given in this publication is for educational purposes only. Reference to commercial products or trade names is made with the understanding that no discrimination is intended and no endorsement by the University of Minnesota is implied.

iii

CORRESPONDING AUTHORS OF PAPERS/PRESENTERS

Lauren Ahaiblame Brent Dalzell South Dakota State University University of Minnesota Agricultural and Biosystems Engineering Department of Soil, Water and Climate SAE 223, Box 2120 439 Borlaug Hall Brookings, SD 57007 1991 Upper Buford Circle St. Paul, MN 55108

Satish Gupta Alexander Hummel, Jr. University of Minnesota University of Minnesota Department of Soil, Water and Climate Department of Agronomy and Plant Genetics 439 Borlaug Hall 411 Borlaug Hall 1991 Upper Buford Circle 1991 Upper Buford Circle St. Paul, MN 55108 St. Paul, MN 55108

Andry Ranaivoson Jeff Strock University of Minnesota University of Minnesota Southwest Research and Outreach Center Southwest Research and Outreach Center 23669 130th St. 23669 130th St. Lamberton, MN 56152 Lamberton, MN 56152

Tamas Varga Lu Zhang University of Minnesota University of Minnesota Department of Soil, Water and Climate Department of Bioproducts and Biosystems 439 Upper Buford Circle Engineering St. Paul, MN 55108 1390 Eckles Avenue St. Paul, MN 55108

iv

TABLE OF CONTENTS

Forward i Acknowledgements iii Corresponding authors/presenters iv

Title Authors Isotope uses in drained agricultural Lu Zhang, Brent Dalzell, Joe Magner, and 1 landscape water budget Jeff Strock

Integrating field-based hydrology data Brent Dalzell, Lu Zhang, Jeff Strock, and 11 with watershed sale models Joe Magner

Mitigating water loss in soybean-corn Alexander Hummel, Jr., Nathan Dalman, 18 rotations with winter cover crops Ronghao Liu, and Axel Garcia y Garcia

Novel design and field performance of the Jeffrey Strock, Andry Ranaivoson, Gary 24 phosphorus-sobing and denitrifying Feyereisen, Kurt Spokas, David Mulla, bioreactors and Marta Roser

Supplemental irrigation in southwest Tamás Varga and Jeffrey Strock 35 Minnesota for corn and soybean

Assessment of side inlet designs Jeffrey Strock, Karl Bear, and Bruce 48 Wilson

Bank erosion in the Minnesota River Satish C. Gupta and Andrew C. Kessler 55 Basin

v

ISOTOPE USES IN DRAINED AGRICULTURAL LANDSCAPE WATER BUDGET

Lu Zhang, Brent Dalzell, Joe Magner, Jeff Strock

Department of Bioproducts and Biosystems Engineering, University of Minnesota Department of Soil, Water, and Climate, University of Minnesota St. Paul, MN

Keywords: Isotope, Controlled Drainage, Water Budget, Water Balance, Residence Time

EXECUTIVE SUMMARY

Stable isotopes of oxygen and hydrogen were used in addition to modeling of water budget in a drained agricultural landscape. Evaporation and condensation cause water to fractionate, therefore changing the percent composition of oxygen and hydrogen isotopes in the residual water. Using the differences in the isotopic composition, researchers will be able to gain more information on the sources of water and estimated average residence time of that part of water budget. Monthly isotope samples were collected at the following field research sites: Beresford, SD, Tracy, MN, and Waseca, MN. Local meteoric water lines were established for the comparison of magnitude of fractionation from different water sources. These include soil water, shallow well, deep well, tile and river. Stable isotopes of hydrogen and oxygen were also used to estimate hydraulic residence time (HRT) of the field hydrologic storage for these sources. As part of the hydrologic impact research, this study will aid in quantifying field water budgets for drained agricultural landscapes and provide information for computer model simulation.

1

ISOTOPE USES IN DRAINED AGRICULTURAL LANDSCAPE WATER BUDGET

INTRODUCTION

The implementation of tile drains alters the local hydrology by providing a “shortcut” for water to travel through the field. Therefore, water moves through the field faster and more ends up in the downstream receiving water in a shorter time. The major negative impacts of tile drains are less soil water storage and flashier streams. Looking beyond the direct impacts, this could also discourage evapotranspiration (ET), induce erosion, and reduce water availability for plants, thus reducing production. Controlled drainage is one of the agricultural best management practices (BMP) designed to manage root zone moisture condition. By installing a control structure that serves the same purpose as the weir, farmers will be able to manually control the water level in the field to retain the most water without damaging the plant roots (Minnesota Department of Agriculture). Controlled drainage also provides nutrient removal benefit by increasing the residence time of water in the field. However, controlled drainage may have the same impact as the tile drains for releasing water downstream. Instead of releasing water continuously, controlled drainage is more likely to release a larger amount in a limited time period. The question this study seeks to answer is how controlled drainage affects crop production and whether controlled drainage has a minimum environmental impact. To further explore the impacts of controlled drainage, modeling approach was taken to predict the crop yield as well as downstream hydrographs. Modeling crop yield will directly answer the question whether controlled drainage is more effective and modeling hydrographs will estimate the percent of tile flow in the streamflow and predict the timing and amount of drainage water being released. In the modeling process, isotope will be used to help better define groundwater recharge rate and verify hydrograph separation. Isotope will also be used to estimate hydraulic residence time of different components of the watershed storage. Oxygen and hydrogen stable isotopes have been used in water science related research to help understand hydrologic pathway and mixing mechanisms. They have also been used to study lake evaporation, waterbody residence time, and extreme event impacts on local hydrology (Fritz and Fontes, 1980; Simpson and Herczeg, 1991; Payne and Yurtsever, 1974; Burns and McDonnell, 1998). Those different uses can also be applied in agricultural landscapes. Main isotopes used in this study are deuterium (D) and oxygen-18 (18O). Deuterium is one of the two stable isotopes of hydrogen. It is heavier than protium. Hydrogen also has a third isotope, tritium, which is the heaviest and radioactive. The half-life of tritium is 12.3 years. Protium comprises about 99.985% of the hydrogen atoms in the atmosphere, whereas D only accounts for 0.015% (Mook, 2001). Oxygen has three isotopes as well: 16O (99.895%), 17O (0.038%), and 18O (0.2%). All three isotopes are stable. Vienna Standard Mean Ocean Water (VSMOW) is a universal standard set by International Atomic Energy Agency (IAEA) for comparison of heavy isotopes as they are

2 usually in trace amount. The following equations are used for calculating the relative abundance of D and 18O:

δD (‰) = (1) δ18O (‰) = (2)

Where: R equals to and , respectively; and δ represents the ratio.

Although isotope hasn’t been used much in agricultural settings, the same concept can be applied to agricultural system and help provide insight to the modeling approach. This paper outlined the methods used for the on-going isotope analysis and presented the result and discussion obtained so far.

METHOD

Sampling

Isotope samples were collected monthly from three locations: South Dakota Research Station near Beresford, SD in the Vermilion River Watershed, Brian Hick’s Farm, near Tracy, MN in the Cottonwood River Watershed, and Southern Research and Outreach Center near Waseca, MN in the Le Sueur River Watershed. The focus at the stage of the research is to analyze isotope data gathered from Tracy site and calibrate the model. At Tracy site, isotope samples were gathered from root zone by piezometer and suction cup lysimeter, subsurface drain tile, groundwater well, wetland, and river. Table 1 shows the depths of the subsurface sample locations. Precipitation samples were collected when available. Samples were stored in tightly capped bottles at room temperature prior to analysis to prevent evaporation and condensation induced fractionation.

Meteoric Water Line

Meteoric water line (MWL) is established by plotting δD against δ18O from meteoric water, meaning precipitation. The global MWL plots samples from across the globe to establish an average slope of 8. When plotting samples from one region, the local MWL tends to deviate from the global average due to the source of the water vapor. The ocean is rich in 18O as 16O evaporates first. When tropical cloud moves up, precipitation along the way removes more 18O from the cloud. The cloud will be lighter when it gets to the higher latitude. Therefore, the polar region waters are richer in 16O. If the precipitation comes from clouds in the south, it tends to be heavier; if the precipitation comes from clouds in the north, it tends to be lighter. Therefore, the points on the local MWL are scattered along the line. The slope of the line is due to further fractionation during precipitation. Local MWL for all three site were established with the available precipitation data.

3

Hydrograph Separation

Researcher started using isotopes for hydrograph separation in late 1960s (Klaus & McDonnell, 2013, Hubert et al., 1969; Crouzet et al., 1970; Dincer et al., 1970; Martinec et al., 1974; Martinec, 1975). Mass balance hydrograph separation model can be a two- component model or a three/multiple-component model. The challenge is to accurately identify end-members (Klaus & McDonnel, 2013). Selecting sources of streamflow is also important as goundwater, soil water, precipitation, and stormwater may all have a contribution to the flow samples. The general expression provided in Isotope Tracers in Catchment Hydrology (Kendall & McDonnell, 1999) was given by Pinder and Jones (1969) and Dincer et al. (1970). It was adopted by Sklash et al. (1976) and Kennedy et al. (1986).

(3) Where: D is deuterium; V is volume; E is total runoff due to precipitation event; PW is pre-event; and R is newly added water.

In this study, first attempt to use hydrograph separation was on Tracy data. Five dates were picked where all locations had data available. Because there was no surface runoff isotope analysis, isotope data from a flow-through wetland was used.

Hydraulic Residence Time Burns & McDonnell (1998) introduced a method for estimating water of interest residence time based on annual isotope signature fluctuation from water of interest and meteoric water. The model is written as the following equation:

(4) Where: is the residence time (days); is the angular frequency of variation (2π/365 days); is the input (precipitation) amplitude; is the output (surface water) amplitude.

This model assumed the waters represent a steady-state and well-mixed reservoir. The authors calculated mean residence time of stream water, soil water, and groundwater. They estimated stream water HRT to be around 100 days, soil water HRT to be 63-80 days, and groundwater HRT to be 160 days at a depth of 8.2 m and 60 m from the stream channel. The analysis was done by plotting precipitation and water of interest as time series plot and fit a sine curve through the data and calculate the amplitudes of the curves.

4

When there is a lack of precipitation data, a linear relationship between temperature and can also be used (Yurtserver, 1975) and this regression equation should be developed with local temperature and precipitation data for accuracy.

(5) Where: is temperature of the day.

At the stage of the research, HRT analysis was done on Tracy and Waseca data. Result and discussion are presented in the latter sections.

RESULT AND DISCUSSION

Meteoric water lines obtained for the three sites are shown in Figure 1. The regression equations from west to east are Equation 6, 7, and 8:

(6) (7) (8)

The slope increased from west to east. The difference in the slopes are caused by a combination of source of the cloud/precipitation and evaporation during precipitation events induced by temperature difference. The next step will be to perform statistical analysis like Kruskal-Wallis or analysis of variance (ANOVA) with post-hoc test to examine the statistical difference of mean air temperature of the three locations. The result of hydrograph separation is being processed at the time of this proceeding. Figure 2 shows the plot of the raw data. By looking at the raw data from different locations, river samples (red line) were consistently lighter than piezometer samples expect for the month of July. This could be caused by input of heavy isotopes from rain event. The challenge of hydrograph separation analysis for these sites is mainly the selection of the end-members. As river being an average value of all the sources, all could have an impact and be a contributor of the river flow. Rain samples represent the on-channel precipitation input. Lysimeter represents the soil water input to the stream. Deep well represents the groundwater input and piezometer represents shallow groundwater input. Tile is the drainage water input. Since only two isotopes (D and 18O) were being measured, the solution to the mixing equation can vary. Therefore, the analysis should be combined with modeling and field observation to eliminate negligible inputs. The preliminary result for the HRT calculation is shown in Table 2. The rest of sampling locations from Waseca will be analyzed when more data points become available. The unexpected long residence time at piezometer depth from Tracy site could possibly be caused by a few factors. Despite the low goodness of fit of 0.35 which indicates a poor fit of the sine model, the water itself in deeper soil horizon can also cause complication. Evaporation induced fractionation occurs while water moves downwards and becomes more enriched in 18O. Freezing and thawing of the soil cause water to fractionate as well. After thawing, water in the soil profile is not well mixed. In addition, deep soil water (>50 cm) is only replaced during snowmelt or by substantial rainfall (Gazis and Feng, 2004).

5

The piezometer could be sampling a mix of newer water and older water. The sampling could also be biased when water in the depth is not well mixed. The HRT analysis was done using precipitation as the input water. When using lysimeter as the input and piezometer as the output, it put the HRT at 199 days. When using lysimeter as the input and well as the output, the residence time became 101 days. As to why the piezometer depth appeared to allow more time for evaporation than the well depth requires further investigation. Preferential flow paths may be involved. Also, the piezometer and well are only installed at one location of the field, which may cause biased sampling.

REFERENCE

Burns, D. A., McDonnell, J. J. 1998. “Effects of a Beaver pond on Runoff Processes: Comparison of Two Headwater Catchments.” J. Hydro. 205: 248-64. Crouzet, E., Hubert, P., Olive, P., Siwertz, E., Marce, A. 1970. Le tritium dans les mesures d’hydrologie de surface. Determination experimentale du coefficient de ruissellement. J. Hydrol. 11 (3), 217-229. Dincer. T., Payne, R., Florkowski, T., Martinec, J. and Tongiorgi. E. 1970. Snowmelt runoff from measurements of tritium and oxygen-18. Water Resour. Res. 6: 110- 124. Gazis, C., Feng, X. 2004. A Stable Isotope Study of Soil Water: Evidence for Mixing and Preferential Flow Paths. Geoderma. 119: 97-111. Fritz, P., and Fontes, J. Ch. 1980. Handbook of Environmental Isotope Geochemistry. Hubert, P., Marin, E., Meybeck, M., Ph. Olive, E.S. 1969. Aspects hydrologique, geochimique et sedimentologique de la crue exceptionnelle de la dranse du chablais du 22 Septembre 1968. Arch. Sci (Geneve) 3, 581-604. Kendall, C., McDonnell, J.J. 1999. Isotope Tracers in Catchment Hydrology. Elsevier Science & Technology. Oxford, United Kingdom. Kennedy, V.C., Kendall, C., Zellweger, W.G., Wyerman, T.A. and Avanzino, R.J. 1986. Determination of the components of streamflow using water chemistry and environmental isotopes, Mattole River Basin, California. Jour. Hydrol. 84: 107- 140. Klaus, J., McDonnell, J.J. 2013. Hydrograph separation using stable isotopes: Review and evaluation. Journal of Hydrology. 505: 47-64. Martinec, J., Siegenthaler, U., Oeschger, H., Tongiorgi, E. 1974. New insights into the run- off mechanism by environmental isotopes. Proceedings Symposium on Isotope Techniques in Groundwater Hydrology. International Atomic Energy Agency, Vienna. pp: 129-143. Martinec, J. 1975. Subsurface flow from snowmelt traced by tritium. Water Resour. Res. 11(3), 496-498. Minnesota Department of Agriculture. Conservation Practices | Minnesota Conservation Funding Guide. Conservation Drainage. Mook, W.G. 2001. Environmental Isotopes in the Hydrological Cycle: Principles and Applications. IHP-V. Technical Documents in Hydrology. No.39. Vol. II. UNESCO, Paris.

6

Payne, B. R., and Yurtsever, Y., 1974. “Environmental Isotopes as a Hydrogeological Tool in Nicaragua.” In Isotope Techniques in Groundwater Hydrology: proceedings of a symposium, IAEA, Vienna, 11-15 March 1974, vol 1: 193-201. Pinder, G.F. and Jones, J.F. 1969. Determination of the groundwater component of peak discharge from the chemistry of total runoff. Water Resour. Res. 5:438-445. Rozemeijer, J. C., van der Velde, Y., McLaren, R. G., van Geer, F. C., Broers, H. P., & Bierkens, M. F. P. (2010). Integrated modeling of groundwater–surface water interactions in a tile-drained agricultural field: The importance of directly measured flow route contributions. Water Resources Research, 46(11). doi:10.1029/2010WR009155 Simpson, H. J., and Herczeg, A. L., 1991. “Stable Isotopes as an Indicator of Evaporation in the River Murray.” Water Resour. Res. 27: 1925-35. Sklash, M.G., Farvolden, R.M., Fritz, P. 1976. A conceptual model of watershed response to rainfall, developed through the use of oxygen-18 as a tracer. Canadian Journal of Earth Science. 13: 271-283. Tomer, M. D., Wilson, C. G., Moorman, T. B., Cole, K. J., Heer, D., & Isenhart, T. M. (2010). Source-pathway separation of multiple contaminants during a rainfall- runoff event in an artificially drained agricultural watershed. Journal of Environment Quality, 39(3), 882–895. doi:10.2134/jeq2009.0289 Yurtsever, Y., 1975. Worldwide survey of isotopes in precipitation. IAEA report, Vienna.

7

Table 1. Subsurface Sampling Depth.

Location Depth (m) Piezometer 2.4 Lysimeter 0.76 Drain tile 1.2 Well 8.2-13.1

Table 2. Hydraulic Residence Time for Tracy Site and Waseca Site.

Site Location HRT (days) r-squared Tracy Lysimeter 404 0.89 Tracy Piezometer 1,453 0.35 Tracy River 141 0.82 Tracy Tile 503 0.74 Tracy Well 808 0.89 Tracy Wetland 49 0.95 Waseca 12” Tile 136 0.66 Waseca Well 609 0.56

8

Figure 1. Global MWL and Local MWL of the Three Field Locations.

9

Figure 2. Raw Data from Hydrograph Separation Analysis.

10

INTEGRATING FIELD-BASED HYDROLOGY WITH WATERSHED SCALE MODELS

Brent Dalzell1, Lu Zhang2, Jeff Strock3, and Joe Magner2

1Department of Soil, Water, and Climate, University of Minnesota, St. Paul, MN. 2Department of Bioproducts and Biosystems Engineering. University of Minnesota, St. Paul, MN. 3Southwest Research and Outreach Center, University of Minnesota, Lamberton, MN.

Keywords: Watershed, SWAT, Tile Drainage, Evapotranspiration, Model

EXECUTIVE SUMMARY

When working to understand how agricultural landscapes respond to varying management practices and weather or climate regimes, field-scale studies are key to identify and quantify the mechanisms responsible for export of water, sediment, and nutrients. In contrast, watershed scale studies focusing on water samples collected from the stream network are valuable for quantifying basin-scale export, but it can be difficult to attribute field-scale mechanisms to data from samples collected at points downstream. Watershed scale models can be helpful to bridge this knowledge gap by simulating key processes at the field scale and then integrating those results into a watershed scale response which can be compared against observed data. Watershed scale models are most robust when model outputs from multiple processes are compared against available observed data to ensure that key processes are represented in a realistic manner. In the upper Midwest, this includes evapotranspiration, tile drainage, and soil water storage. When these processes are represented realistically, model performance for streamflow shows good agreement with observed data. The resulting calibrated and validated model can then be employed to simulate alternative scenarios with greater confidence and also to guide further field sampling campaigns.

11

INTEGRATING FIELD-BASED HYDROLOGY WITH WATERSHED SCALE MODELS

INTRODUCTION

In order to understand long-term implication of farm management decisions, field studies and long term monitoring can be complemented by modeling efforts. Simulation models allow us greater insight into potential mechanisms behind environmental observations and can help us to form more realistic expectations about the likelihood of meeting environmental goals for given scenarios of management, weather and/or climate. Further, results from simulation models can provide spatially continuous data necessary for further study such as economic analysis of landscapes under alternative cropping systems and/or alternative policy scenarios. However, watershed models with large numbers of parameters can be complicated to use and it can be possible to achieve similar model results through contrasting combinations of model parameters. Put another way, you can get the right answer for the wrong reason. The ability to achieve similar model results through multiple combinations of parameter values is referred to as the equifinality thesis (Bevin, 2006) and serves as a caution against accepting model results without a deeper look into different model components. When key model components can be supported with ancillary sources of data, however, watershed models can be considered more robust and their results can be used to provide meaningful insight into the physical processes that can drive larger watershed scale responses to changes in management, weather, or climate. For this study, a model of the Cottonwood River watershed was developed to evaluate how farm sale management may result in watershed scale changes to hydrology. While the ultimate use of this model will be to develop and test alternate management scenarios, this paper focuses on demonstrating how different modeled hydrology components can be compared against observed datasets in order to achieve a good model that is robust for future applications.

STUDY AREA

The Cottonwood River Watershed is located in south-central Minnesota and is a tributary of the Minnesota River. The watershed is about 3,339 km2 and land use is predominately corn and soybean row crop agriculture which comprises about 88% of the watershed (Fig 1). Soils are mainly formed from glacial till and many fields have subsurface tile drainage installed to improve growing conditions and reduce inter-annual variability in yield (Zucker and Brown, 1998). The Cottonwood River Watershed and other tributaries in the Minnesota River Basin are the subject of many current studies that aim to understand the causes of, and to identify potential solutions to increasing streamflow and high levels of sediment and nutrients that negatively impact water quality in the Minnesota River Basin as well as sites downstream such as Lake Pepin and the Gulf of Mexico (Belmont et al., 2011; Dalzell et al., 2017; Engstrom et al., 2009; Pennington et al., 2017; Schottler et al., 2013) .

THE SWAT MODEL

12

The model used for his study is the SWAT model (Soil and Water Assessment Tool). It relies on both physically based and empirical approaches to quantify the effects of land cover and management on water qua1ity and quantity (Arnold and Fohrer 2005; Arnold et al. 1998; Gassman et al. 2007). The model was developed primarily for use in watersheds with agricultural land use and it has extensive databases of common agricultural management practices. The model operates on a daily time-step and requires weather inputs of precipitation, temperature, relative humidity, and wind speed. Key spatial inputs are land use, soils, and topography; these spatial datasets are intersected and unique combinations of these inputs form hydrologic response units (HRUs), the basic modeling unit. Model performance is evaluated by comparing simulated results against observed data and a commonly used metric of performance is Nash Sutcliffe Efficiency (NSE; Nash and Sutcliffe, 1970).

Where is the observed monthly value (in this case, stream flow) and is the modeled value, and is the mean value of the observed data. NSE values can range from -∞ to 1 with 1 indicating perfect model performance and values greater than zero indicating that the model provides greater information than would be provided by the mean value alone. For watershed scale model calibration, NSE values greater than 0.50 are generally considered acceptable while values greater than 0.65 and 0.75 are considered good and very good, respectively (Moriasi et al. 2007).

Figure 1. Study watershed map showing the Cottonwood River Basin and its major land cover classes. Plant Growth and Evapotranspiration

The SWAT model used a daily accumulated heat unit approach to determine plant growth and over the course of the growing season, water use will reflect the plant biomass and roots as well as available soil water and weather conditions. Because evapotranspiration comprises roughly 75% of precipitation in Minnesota (this varies across

13 the state and from year-to-year; Baker et al., 1979), it is important to ensure that simulated plant growth and evapotranspiration is reasonable. In order to ensure that simulated evapotranspiration was realistic for this study, observed data from the Ameriflux network (http://ameriflux.ornl.gov/) were used to characterize plant water use over the course of the growing season. Plant growth parameters in the model were adjusted to ensure that daily trends in simulated evapotranspiration followed observed data for row crops as well as perennial grasses (Fig 2).

Figure 3. Comparison of simulated drainage efficiency in the Cottonwood River Watershed (red dots) against Figure 2. Comparison of observed and observed values from field Scale studies in simulated evapotranspiration for (a) perennial Waseca, MN (black dots and open circles; grasses and (b) corn in the upper Midwest. (from Dalzell et al., 2011). Dalzell et al., 2017)

Subsurface Tile Drainage and Soil Water

The common use of subsurface tile drainage in the Cottonwood River Watershed has the potential to change the pathways by which water will leave the farm fields with implications for changing landscape scale water yield. The SWAT model simulates tile drainage if the water table is above the depth of drainage and lateral flow of water through drains can be determined by a lag time (Neitsch et al., 2005) or Hooghoudt and Kirkham equations (Moriasi et al., 2014). For watershed-scale studies, it is helpful to compare drainage efficiency against data from field-scale studies (Fig 3). The general relationship between drainage efficiency and total annual precipitation in the Cottonwood River

14

Watershed agrees with data from field plots located in Waseca, MN (Dalzell et al., 2011) and similar comparisons can be performed with annual trends in soil water to ensure that they reflect observed trends to show periods of recharge (spring and fall) as well as consumption during the growing season (Baker et al., 1979; data not shown).

Watershed Scale Calibration and Validation

After ensuring that key aspects of hydrology can be represented in a reasonable manner, it is suitable to perform calibration and validation of the model at the watershed scale. For the Cottonwood River Watershed model, calibration was performed on the 10-year period from 1/1/2004 through 12/31/2013. Once calibration was complete, the model was validated by applying the same calibration parameters to the Redwood River Watershed which is located adjacent to the north of the Cottonwood River. Results from both models showed good agreement between observed and simulated daily stream flow (Fig 4).

Figure 4. Daily observed and modeled stream flow for the Cottonwood and Redwood River Watersheds (used for model calibration and validation, respectively).

Model applications

Once calibrated and validated, watershed scale models can be employed to evaluate alternative management scenarios and to help set realistic goals for water quality improvements under varying weather and climate conditions. Validated models can also help to provide insight into specific mechanisms underlying watershed behavior as well as identify potential knowledge gaps. These models can also help to inform field sampling campaigns by identifying conditions that can yield water samples more likely to provide helpful new information. 15

For example, isotope tracer studies are helpful for identifying different sources of water (e.g., surface runoff vs. groundwater flow) which can help our understanding of how different flow paths can influence watershed scale water quality (e.g., Magner et al., 2004). However, cost and time constraints often limit the number of field samples that can be collected and analyzed in support of this work. Model results can be used to identify which flow paths are most dominant under varying watershed conditions (Fig 5) and results such as these can help to provide insight on how field-scale processes and management may be influencing water quality and quantity at the watershed scale.

Figure 5. Simulated water yield components in the Cottonwood River Basin in April-May 2005. Results are from SWAT model outputs and can help to identify how key flow paths may dominate under varying conditions.

SUMMARY

In order to help ensure that model calibration is robust and that the model can provide meaningful results, it is important to compare multiple model outputs to available datasets where possible. In the agricultural landscapes of the upper Midwest, this includes ensuring that simulated evapotranspiration, tile drainage, and soil water values are in agreement with observed data. When properly calibrated and validated, watershed scale models can be helpful for evaluating alternative management scenarios as well as provide insight into the physical processes underlying watershed behavior. This can yield new information and guide future field sampling efforts in order for scientists and watershed managers to work toward their goals.

16

REFERENCES

Arnold, J.G. and N. Fohrer. 2005. SWAT2000: current capabilities and research opportunities in applied watershed modelling. Hydrological Processes 19:563-572. doi:10.1002/hyp.5611. Arnold, J.G., R. Srinivasan, R.S. Muttiah, and J.R. Williams. 1998. Large area hydrologic modeling and assessment - Part 1: Model development. Journal of the American Water Resources Association 34:73-89. doi:10.1111/j.1752-1688.1998.tb05961.x. Belmont, P., K.B. Gran, S.P. Schottler, P.R. Wilcock, S.S. Day, C. Jennings, J.W. Lauer, E. Viparelli, J.K. Willenbring, D.R. Engstrom, and G. Parker. 2011b. Large shift in source of fine sediment in the Upper Mississippi River. Environmental Science and Technology 45:8804-8810. doi:10.1021/es2019109. Dalzell, B.J., J.Y. King, D.J. Mulla, J.C. Finlay, and G.R. Sands. 2011. Influence of subsurface drainage on quantity and quality of dissolved organic matter export from agricultural landscapes. Journal of Geophysical Research-Biogeosciences 116. doi:G0202310.1029/2010jg001540. Dalzell, B.J., and D.J. Mulla. 2017-in press. Perennial vegetation impacts on stream discharge and channel sources of sediment in the Minnesota River Basin. Journal of Soil and Water Conservation. 2017-in press. Engstrom, D.R., J.E. Almendinger, and J.A. Wolin. 2009. Historical changes in sediment and phosphorus loading to the upper Mississippi River: mass-balance reconstructions from the sediments of Lake Pepin. Journal of Paleolimnology 41:563-588. doi:10.1007/s10933-008-9292-5. Gassman, P.W., M.R. Reyes, C.H. Green, and J.G. Arnold. 2007. The soil and water assessment tool: Historical development, applications, and future research directions. Transaction of the ASABE 50:1211-1250. Magner, J.A., G.A. Payne, and L.J. Steffen. 2004. Drainage effects on stream nitrate-N and hydrology in south-central Minnesota (USA). Environmental Monitoring and Assessment. 91: 183-198. Moriasi, D.N., J.G. Arnold, M.W. Van Liew, R.L. Bingner, R.D. Harmel, and T.L. Veith. 2007. Model evaluation guidelines for systematic quantification of accuracy in watershed simulations. Transactions of the ASABE 50:885-900. Nash, J.E. and J.V. Sutcliffe. 1970. River flow forecasting through conceptual models: Part 1. A discussion of principles. Journal of Hydrology 10:282-290. Neitsch, S.L., J.G. Arnold, J.R. Kiniry, and J.R. Williams. 2005. Soil and Water Assessment Tool Theoretical Documentation, Version 2005. Temple Tex.: USDA- ARS Grassland, Soil and Water Research Laboratory. Pennington, D.N., B. Dalzell, E. Nelson, D. Mulla, S. Taff, P. Hawthorne, and S. Polasky. 2017. Cost-effective land use planning: Optimizing land use and land management patterns to maximize social benefits. Ecological Economics. 139:75-90. doi:10.1016/j.ecolecon.2017.04.024 Schottler, S.P., J. Ulrich, P. Belmont, R. Moore, J.W. Lauer, D.R. Engstrom, and J.E. Almendinger. 2013. Twentieth century agricultural drainage creates more erosive rivers. Hydrological Processes doi:10.1002/hyp.9738. Zucker, L.A. and Brown, L.C. 1998. Agricultural drainage – Water quality impacts and subsurface drainage studies in the Midwest. Bulletin 871. The Ohio State University, 40. 17

MITIGATING WATER LOSS IN SOYBEAN-CORN ROTATIONS WITH WINTER COVER CROPS

Alexander Hummel Jr.1, Nathan Dalman2, Ronghao Liu1,2, and Axel Garcia y Garcia1,2

1Department of Agronomy and Plant Genetics, University of Minnesota, St. Paul, Minnesota 2Southwest Research and Outreach Center, Lamberton, Minnesota

Keywords: Water Use Efficiency, Winter Camelina, Field Pennycress, Winter Rye

EXECUTIVE SUMMARY

Soybean [Glycine max (L.) Merr.] and corn (Zea mays L.) are the dominant crops in the Midwestern United States. The incorporation of cover crops in the soybean-corn rotation is becoming popular in the region. While the strategy is expected to provide benefits to the system, its impact on soil water availability to major crops is a concern. A 3-yr study on crops water use in a soybean-corn rotation with the integration of winter annual cover crops [camelina (Camelina sativa L. Crantz), field pennycress (Thlaspi arvense L.), and winter rye (Secale cereale L.)] was initiated in 2015 in southwest MN. Winter annual cover crops are interseeded into soybean at developmental stage R7 and into corn at R6. Soil moisture is monitored weekly at six fixed depths, including 10, 20, 30, 40, 60, and 100 cm. Water balance using a mass conservation approach is used to determine the water use of crops and cover crops. Our preliminary results from the 2015-2016 season showed that winter annual cover crops used less water in fall than in spring and that cover crops’ water use followed a pattern of winter rye > camelina > pennycress. Water used by pennycress and winter camelina during fall and spring was similar to water loss from the treatment without cover. Water used from winter rye was around 50% higher than the water loss in the no cover treatment.

18

INTRODUCTION

Soybean and corn are the most important crops in the Midwestern United States. In 2016, 93.4 million acres of corn and 83.4 million acres of soybeans were harvested to produce a combined production value of 93 billion dollars (USDA, 2017). The Midwestern Corn Belt states represent over 80% of maize production, emphasizing the importance of the soybean-corn rotation in the region (USDA, 2012). The temperate and humid climate of this region accentuates the need to conserve water not only to maintain soil quality but also to prevent nitrogen leaching. Heavy precipitation events can cause disruption to soil aggregates, especially when fields are left bare (Kaspar, 2011). Cover crops allow for soil coverage during times when fields are left fallow and provide protection against soil surface erosion (Kaspar, 2011). This evidences the role of cover crops as an option for sustainable cropping practices when integrated into the soybean-corn rotation. Simultaneously, cover crops could either use excess water or limit the availability of soil moisture to major crops. This is of major importance in rainfed crop production, where water is the most limiting factor that affects crop yields. Sustainable crop management practices are based on the development of cropping systems that provide minimum negative effects to the environment, are accessible and operational to producers, and have the potential to improve productivity and enhance the quality of the environment (Pretty, 2008). Cover crops or novel crops that can be used as double purpose crops (cover and grain yield) are suitable for such an approach. The integration of a third crop like cover crops into a soybean-corn rotation means intensification; therefore, this approach may succeed if the productivity of the major crops is not negatively affected. Cover crops increase soil moisture through infiltration but deplete soil water content through transpiration and uptake; however, the effect of soil water depletion by cover crops in humid regions is negligible when precipitation is adequate (Unger and Vigil, 1998). Findings from a study conducted by Gesch and Johnson (2015) in the northern Corn Belt region reports that dual- and relay-cropped camelina – soybean used more water than the full season soybean alone; however, the water use efficiency in the relay-cropping system was similar to the single cropped soybean but less for the double-cropping system. Low yield and high water use were the cause of less water use efficiency in dual cropping. Campbell et al. (1984) observed decreased soil water content in a soybean dual cropping system utilizing winter rye; soybean establishment was delayed but yields were greater in the dual crop treatments due to water conservation during a late season drought. Similar results were observed by Moschler (1967) in a study comparing corn harvest using conventional tillage or planting into rye. Corn yields averaged higher in rye cover treatments in 4 of 13 comparisons and similar yields were achieved in the remaining nine. Yields can be maintained or increased when planting into a winter cover crop, especially when water conservation is critical. For example, water use of soybean and corn peaks around anthesis, which usually occurs during a dry period in the region. The addition of cover crops into the soybean-corn rotation brings diversity and therefore the demand for more resource use. Research on the effect of cover crops on soil-water budget is limited in the region. While the benefits of cover crops is well documented, more research is needed to advance our understanding on their benefits in temperate and humid climates. We hypothesize that cover crops will affect soil water but will have positive effects on the growth and yield of primary crops and the environment. The objectives of 19 this study are to determine i) the water use of cover crops and primary crops and ii) the water use efficiency of crops.

MATERIAL AND METHODS

A 3-yr study was established in 2015 at the University of Minnesota’s Southwest Research and Outreach Center located near Lamberton, MN. The study utilizes a soybean- corn rotation with the integration of winter cover crops + a control treatment (no cover). The study is set on a randomized complete block design with four replications. The dominant soil type found in the study plots is a Webster clay loam (fine-loamy, mixed, superactive, mesic Typic Endoaquolls). Winter camelina, field pennycress, and winter (cereal) rye are being used as winter cover crops in this study. Winter rye is a winter hardy (De Bruin et al., 2005; Feyereisen et al., 2006) dominant cover crop in the corn belt of USA (Singer, 2008; Kladivko et al., 2014). Field pennycress is a winter-annual crop being developed by the University of Minnesota with potential as source of biofuel for industry (Cermak et al., 2013). Winter camelina (Camelina sativa L. Crantz) is a crop with potential for edible oil and biofuel that can be integrated into cereal-based cropping systems (Moser, 2010; Guy et al., 2014). Both major crops are present each year and are planted for optimum yield. Information on soil, plant growth and yield, management practices, and weather conditions is collected. Volumetric soil moisture content will be obtained at 10, 20, 30, 40, 60, and 100 cm depths using PR2 probes (Delta-T Devices®; www.dynamax.com). Access tubes for the PR2 probes are installed between two rows in the center of each plot. A water balance (Allen et al., 1998) was performed to calculate the water use of cover crops. The components of our simplified water balance include the water content in the root zone (to a depth of 1.0 m) and rainfall. The experimental field was relatively flat and the soil is poorly drained so runoff and drainage losses are considered negligible. The water use of soybean and corn presented in this document was estimated using weather data and crop coefficients following the approach of Allen et al. (1998). The water use efficiency of soybean, corn, and cover crops will be obtained as the ratio of yield (grain or total biomass) to water used to make that yield. Measured and derived variables are analyzed with the appropriate analysis of variance and regression and correlation analyses using SAS (SAS Institute Inc. 2012. SAS/STAT Users Guide. Cary, NC) and SigmaPlot® (SigmaPlot v12.3; Systat Software, Inc., San Jose, CA).

PRELIMINARY RESULTS

The water use of corn and soybean showed a typical seasonal pattern, with a gradual increase in the early part of the season and mid-season high values followed by decreasing values associated with the onset of senescence (Fig. 1). The total water use for corn and soybean was 570 and 510 mm, respectively, yielding water use efficiency of 2.3 and 0.87 kg m-3, respectively. Cover crops’ water use in fall and spring following corn in 2015-2016 growing season was similar with water use following soybean (Fig. 2). The winter camelina and field pennycress water use in fall and spring following corn was similar with the fallow treatment. Winter rye following corn used more water compared to the fallow, which increased by 51.5% in fall and 51.8% in spring, respectively. Following soybean, all the 20 three cover crops caused less water use in fall compared to the fallow treatment. In spring, field pennycress following soybean had a similar water use with fallow treatment. Compared to the fallow, winter camelina and winter rye increased water use by 21.4% and 59.2% in spring, respectively. Throughout the growing season (2015-16), winter camelina and field pennycress water use following corn was similar with the fallow treatment (Fig. 3). Winter rye increased water use by 51.7% and 31.5% following corn and soybean, respectively. The total water use of the winter cover crops followed a pattern of winter rye > winter camelina > field pennycress in the corn-soybean rotation.

FINAL REMARKS

Our preliminary results show that winter annual cover crops used less water in fall than in spring, suggesting that water used by winter cover crops in the soybean/corn rotation is marginal during the fall but more substantial during the spring. Winter cover crops water use followed a pattern of winter rye > camelina > pennycress. Water use from field pennycress and winter camelina during fall and spring was similar to water loss from the treatment without cover. On the other hand, water use from winter rye was around 50% higher than water loss in the no cover treatment. Further steps include analysis of the 2016- 2017 research results and data collection during the 2017-2018 season.

Acknowledgements

This study is supported by the Minnesota Soybean Research & Promotion Council. The authors thank Lindsey Englar, personnel from SWROC, and summer students for their contribution during the 2015-2016 season.

REFERENCES

Allen, R.G., L.S. Pereira, D. Raes, and M. Smith. 1998. Crop evapotranspiration - Guidelines for computing crop water requirements. Irrigation and Drainage, Paper No. 56. FAO, Italy, 300 pp. Campbell, R. B., R. E. Sojka, and D. L. Karlen. 1984. Conservation tillage for soybean in the U.S. Southeastern Coastal Plain. Soil Tillage & Research 4:531-541. Cermak, S.C., G. Biresaw, T.A. Isbell, R.L. Evangelista, S.F. Vaughn, R. Murray. 2013. New crop oils—Properties as potential lubricants. Industrial Crops and Products, 44:232-239. De Bruin, J.L., P.M. Porter, and N.R. Jordan. 2005. Use of rye cover crop following corn in rotation with soybean in the Upper Midwest. Agron. J., 97:587-598. Feyereisen, G.W., B.N. Wilson, G.R. Sands, J.S. Strock, and P.M. Porter. 2006. Potential for a rye cover crop to reduce nitrate loss in southwestern Minnesota. Agron. J., 98:1416-1426. Gesch, R.W. and J.M.F. Johnson. 2015. Water use in camelina-soybean dual crop systems. Agron. J., 107:1098-1104. Guy, S.O., D.J. Wysocki, W.F. Schillinger, T.G. Chastain, R.S. Karow, K. Garland- Campbell, I.C. Burke. 2014. Camelina: adaptation and performance of genotypes. Field Crops Research, 155:224-232. 21

Kaspar, T. C. and Singer, J. W. 2011. The use of cover crops to manage soils. Publications from USDA-ARS / UNL Faculty. Paper1382.http://digitalcommons.unl.edu/usdaarsfacpub/1382 [Accessed on: 06/30/2017]. Kladivko, E.J., T.C. Kaspar, D.B. Jaynes, R.W. Malone, J. Singer, X.K. Morin, and T. Searchinging. 2014. Cover crops in the upper Midwest United States: potential adoption and reduction of nitrate leaching in the Mississippi river basin. Journal of Soil and Water Conservation, 69(4):279-291. Moschler, W.W., G.M. Shear, D.L. Hallock, R.D. Sears, and G. D. Jones. 1967. Winter cover crops for sod-planted corn: Their selection and management. Agron. J., 59:547-551. Pretty, J. 2008. Agricultural sustainability: concepts, principles and evidence. Phil. Trans. R. Soc. B, 363447-465. Singer, J.W. 2008. Corn Belt assessment of cover crop management and preferences. Agron. J., 100:1670-1672. Unger, P.W., and M.F. Vigil. 1998. Cover crop effects on soil water relationships. Journal of Soil and Water Conservation, 53(3):200-207. USDA Census of Agriculture. 2012. Census Publications. https://www.agcensus.usda.gov/Publications/2012/ [Accessed on: 06/30/2017]. USDA, National Agricultural Statistics Service. 2017. Crop Production 2016 Summary. http://usda.mannlib.cornell.edu/usda/current/CropProdSu/CropProdSu-01-12-2017.pdf [Accessed on: 06/30/2017].

22

FIGURES

(a) (b) Fig. 1 Water use of (a) soybean and (b) corn at Lamberton, MN during the 2015 growing season.

(a) (b) Fig. 2 Water use of cover crops in corn-soybean rotation at Lamberton, MN during (a) fall 2015 and spring 2016. (WC = winter camelina; FP = field pennycress; and WR = winter rye).

Fig. 3 Water use of cover crops in corn-soybean rotation at Lamberton of Minnesota during 2015-2016 growing season. (WC = winter camelina; FP = field pennycress; and WR = winter rye).

23

NOVEL DESIGN AND FIELD PERFORMANCE OF PHOSPHORUS-SORBING AND DENITRIFYING BIOREACTORS

Jeff Strock1, Andry Ranaivoson1, Gary Feyereisen2, Kurt Spokas2, David Mulla3 and Marta Roser3

1Southwest Research and Outreach Center, University of Minnesota, Lamberton, MN. 2 USDA-ARS, Soil and Water Research Unit, St. Paul, MN. 3Department of Soil, Water and Climate, University of Minnesota, St. Paul, MN,

Keywords: Side Inlet, Runoff, Sediment, Water Quality

EXECUTIVE SUMMARY

Field experiments were conducted at the University of Minnesota Southwest Research and Outreach Center (SWROC) in Lamberton, Minnesota to experimentally assess the impact of a novel two phase bioreactor design for removing N and P from agricultural subsurface drainage water. Modular bioreactors were constructed using mixed woodchips plus corn cobs for facilitating denitrification plus either crushed concrete, steel slag or limestone fragments for phosphorus sorption. Flows from the bioreactors were directed to flow gauges and water sampling equipment. The response of the different bioreactors was assessed using a calibration and a treatment period. During the calibration period only subsurface drainage water was delivered to the bioreactors. During the treatment period subsurface drainage water spiked with potassium acetate was delivered to the bioreactors. Data were collected and analyzed to determine the performance and efficiency of the modular bioreactors under various temperature regimes. Nitrate removal was tied to the hydraulic retention time in the bioreactor coupled with the addition of acetate. Longer retention time resulted in a greater removal of nutrients, however, acetate improved nitrogen removal efficiency. Results also indicate that reduced conditions occurred within the bioreactors but only consistently when acetate was added to the subsurface drainage water.

24

NOVEL DESIGN AND FIELD PERFORMANCE OF PHOSPHORUS-SORBING AND DENITRIFYING BIOREACTORS

INTRODUCTION

Contemporary end-of-tile bioreactors consist of a water level control structure used to route water into the bioreactor. The typical bioreactor consists of a narrow (<1 m wide) trench, 1 to 1.5 m deep and 10’s of meters long. Bioreactors may be lined to prevent seepage or unlined. The bioreactors are then filled with sources of carbon which may include saw dust, wood chips or corn cobs. These bioreactors are solely designed to reduce NO3− loading to surface water by denitrification. Treatment of agricultural tile drain water through current designs of bioreactors is mainly through horizontal flow through the bioreactor media. It is difficult to know how much of the reactive area of the bioreactor is involved when water flows horizontally as preferential flow patterns may by-pass zones within the bioreactor. The longevity and maintenance of bioreactors is not fully known because there are very few long-term bioreactors sites in existence. This is the first known bioreactor designed specifically to remove N and P from agricultural drainage water. For two-phase bioreactors designed to remove N and P there are three factors that will affect longevity – the availability of exchange sites for P sorption, the supply of carbon to denitrifying organisms, and the saturated hydraulic conductivity of the bioreactor. We designed a novel bioreactor capable of removing both N and P and also which would be accessible to easy maintenance. The prototype design consists of a reinforced tank, porous lava rock, a sheet of Brotex, a honeycomb shaped geotextile cellular containment material (EnviroGrid; Geo Products, LLC, Houston, TX), a layer of wood chips and a layer of corn cobs. These layers encompass the hydraulic filtering media and the dentrification media. Three materials were selected for the P removal media including sieved steel slag, sieved crushed recycled concrete or limestone. The concept for the bioreactor system is a modular system that can be installed in the field for drainage water treatment, removed when necessary for maintenance and replaced. Once removed from the field the N and P materials could be recycled. The system is designed for installation adjacent to individual tile outlets along a drainage ditch in order to remediate dissolved nitrogen and phosphorus. The number of modules installed at a particular location would be in part determined by the size of the outlet pipe and the desired treatment efficiency (hydraulic residence time and/or nutrient reduction).

METHODS

Construction of the Modular Bioreactor

The top of the reinforced tank was cut off and then the tank/bioreactor constructed with three layers of materials. The bottom, non-reactive layer (outlet) consisted of 0.67 ft. of lava rock plus a sheet of Brotex cut to fit the tank. The denitrification layer consisted of two 0.67 ft. thick layers of carbon media. The first layer of denitrification media consisted of mixed species woodchips encased in the honeycomb shaped geotextile cellular containment material (van Driel et al. 2006; Jaynes et al. 2008; Karanasios et al., 2010; Robertson and Merkeley, 2009). The second layer of denitrification media consisted of 25 coarsely ground corn cobs, also encased in the honeycomb shaped geotextile cellular containment material. The purpose of encasing the denitrifying media in the honeycomb shaped geotextile material was to create multiple “treatment columns” within each bioreactor in order to minimize preferential flow through the bioreactor. The top layer of the modular bioreactor consisted of the P sorbing material – steel slag, crushed concrete or limestone layered 0.5 ft. thick over the denitrification media. The lava rock/Brotex layer was to aid in controlling hydraulic residence time (HRT) and to increase the surface area where additional biofilms could colonize.

Location of Study

Following proof-of-concept lab testing and modification, the modular bioreactor system was ready for field deployment and experimentation. The field location chosen for experimentation was at the Southwest Research and Outreach Center, near Lamberton, MN. The site was chosen because of the availability of infrastructure, labor and resources to conduct this research. The SWROC is located in the Cottonwood River watershed of the Minnesota River Basin, in southwest Minnesota. The site is located on a lower elevation of glacial till lowland plains. The climate is interior continental with cold winters and moderately hot summers with occasional cool periods. Total annual precipitation of 670 mm is adequate for row crop production, because 74% of the annual rainfall comes during the growing season from April to September. Subsurface drainage from approximately 125 ha, discharges into a channel adjacent to the bioreactors. Subsurface drainage discharge is seasonal, with higher flows from April through June when spring snowmelt combines with spring rainfall. The contributing watershed area comprises 74% cropland (row crops), 20% pasture, and 6% farmstead. The soils of the watershed are of the Canisteo−Ves association. Canisteo soils [Fine-loamy, mixed, superactive, calcareous, mesic Typic Endoaquolls] are poorly drained and are found on the broad lowland glacial till plain. The Canisteo soils and other poorly drained soils in this association require artificial drainage to make them suitable for crop production. Ves [Fine-loamy, mixed, superactive, mesic Calcic Hapludolls] soils are well drained and occupy convex knolls above the lowland till plain. Erosion is a concern in management of this soil. These soils are used mainly for row−crop production.

Research Design

Flow measuring devices and water sampling equipment were installed at the experimental site in order to characterize flow rate and volume and to obtain water samples for chemical characterization (Figure 1) (Woli et al., 2010). Subsurface drainage was measured in a flow control structure using a pressure transducer connected to a datalogger. The flow control structure was also used to divert water to the modular bioreactors. The outlet elevation of the flow control structure was managed at the highest level possible which created a column of water on the upstream side of the flow control structure. A single 5 cm dia. PVC pipe was installed in the bottom, up-stream side of the flow control structure which then diverted subsurface drainage water through an in-line PVC debris Y-filter to the modular bioreactors.

26

In biological treatment processes which promote denitrification, an external carbon source is frequently added as an electron donor in order to stimulate the process of denitrification. Some carbon sources which have been used to promote denitrification include: acetate (C2H3O2), glucose (corn syrup, C6H12O6), Lactate (C3H6O3), sucrose (molasses, C12H22O11), ethanol (C2H6O) and soybean oil (C18H32O2). Commercially available carbon substrates have some characteristics that are similar, including some degree of degradability and solubility. release of soluble carbon or are immediately available at a higher concentration. In this project, potassium acetate (CH3CO2K) was used as the external carbon source (Table 1 and Figure 2). The concentration was based on reducing nitrate concentration at 20mg/L for a subsurface drain flow rate of 9.0 gpm in the 2”-PVC water mainline; the stochiometric ratio of C/N was set at 0.82 (Lew et al., 2012). The target flow rate of subsurface drainage water delivered to each bioreactor was 4 L per minute (1 gpm). The single PVC pipe from the flow control structure was split into three separate distribution pipes each fitted with a paddlewheel flowmeter which was connected to a datalogger. Each of the three distribution pipes routed water to a block of three modular bioreactors. Flow to a block of three bioreactors was split again into three separate distribution pipes and routed to individual bioreactors in which flow was controlled with a small, PVC ball valve. Flow into the top of the bioreactor was delivered through a section of plastic gutter with holes drilled into the bottom in order to attempt to distribute water as evenly as possible across the surface of the bioreactor (Figure 3). The outlet of the bioreactor consisted of an adjustable standpipe which was used to raise or lower the elevation of water in the bioreactor. The stand pipe was also used to divert bioreactor effluent to a flow gauge and to collect water samples for chemical characterization. A tipping bucket flow gauge with approximately 4 L capacity per tip was used to measure bioreactor discharge volume and rate.

RESULTS

The weather conditions during 2016 would be considered abnormal. Annual precipitation at the SWROC was 41% above average, 960 mm compared to the 30-year average 679 mm. Mean growing season precipitation, April through September, is 435 mm whereas during 2016 a total of 646 mm was observed. Subsurface drain flow normally begins in mid-March when snowmelt begins and soils begin to thaw and ends in mid- to late-July when crop evapotranspiration exceeds precipitation. During 2016, drain flow was relatively constant beginning in April through November. Drainage system and bioreactor outflow and water analysis operations were terminated in mid-November when air temperatures dropped to at or below freezing for an extended period of time.

Bioreactor Hydrology

Statistical analysis did not reveal any significant differences (p=0.359) for cumulative discharge volume among bioreactors during 2016 (Table 2). The period from No Acetate Period 1 through Acetate Period 2 occurred during late spring and early summer. Late spring to early summer cumulative discharge volume from the bioreactors varied among periods and among the P treatments (Table 4.3). The greatest discharge volume was 27 observed during No Acetate Period 1 and then gradually declined during Acetate Period 1 and 2. Cumulative bioreactor discharge volume increased from Acetate Period 2 to Acetate Period 3. One plausible explanation for the reduction in cumulative discharge volume between the No Acetate Period 1 and Acetate Period 1 was due to the decrease in supply of subsurface drain flow. This was expected as crop demand increased during the growing season, which effectively reduced drain flow. In addition, some flow reduction could be attributed to a buildup of algae in the PVC supply lines. Algae was observed in the inline sediment traps and flow meters of the pipe delivering water to the bioreactors and was abated on a daily basis to ensure optimal flow. We hypothesized that a combination of dissolved N, P, K and C in the drainage water coupled with light transmission through the white PVC supply line and the relatively long distance between the point of injection of Acetate and the modular bioreactors contributed to the algae growth.

Hydraulic Residence Time

The mean daily discharge rate is an indicator of hydraulic performance of the bioreactors. Mean porosity and hydraulic residence time (HRT) of the modular bioreactors is shown in Table 3. The target HRT for the modular bioreactors was one hour. Statistical analysis did not reveal any significant differences (p=0.203) for HRT among P treatments during 2016. Mean HRT was always greater than the one-hour target HRT. Results indicated that the shortest HRT occurred during No Acetate Period 4. The longest HRT’s occurred during Acetate Period 2 and No Acetate Period 2. The data indicate that the HRT for the Crushed Concrete and Limestone P treatments were similar across all periods. In contrast, the HRT for the Steel Slag treatment was more variable and frequently had the longest HRT (e.g. Acetate Period 1, Acetate Period 2, No Acetate Period 2 and Acetate Period 4). The Steel Slag HRT ranged from 1.1 to 1.8 times greater than the other two treatments. The observed changes in HRT for the Steel Slag could be due in part to a conglomeration of the individual aggregates into a conglomerate with lower hydraulic conductivity.

Nitrogen

Statistical analysis did not reveal any significant differences (p=0.186) for NO3-N concentration among P treatments during 2016. The subsurface drainage source water NO3-N concentration was greater than 20 mg/L during No Acetate Period 1 and Acetate Period 1 that coincided with late spring/early summer. During the initial period, No Acetate Period 1, the percent NO3-N concentration reduction from the bioreactors was about 3% (Table 4). During the second no acetate period, No Acetate Period 2, the percent NO3-N concentration reduction ranged between 7 and 14%. The largest percent NO3-N concentration reduction from the bioreactors occurred during Acetate Period 2 and the smallest NO3-N reduction occurred during Acetate Period 4. A combination of warm air temperatures coupled with relatively long HRT contributed to near complete denitrification of the subsurface drainage water during Acetate Period 2. During Acetate Period 4, there was 0% reduction in NO3-N concentration (Table 4). In contrast, cold air temperature was the major contributing factor to the lack of denitrification during Acetate Period 4. 28

Table 4 shows mean bioreactor NO3-N load during 2016 (Table 5). The largest mean bioreactor NO3-N loads observed occurred during No Acetate Period 1 and the smallest during Acetate Period 2. The data also exhibit a gradual decline in mean bioreactor NO3- N load from No Acetate Period 1 to Acetate Period 2. Mean bioreactor NO3-N load increased from Acetate Period 2 to Acetate Period 3. Mean bioreactor NO3-N load was generally uniform regardless of P treatment from Acetate Period 3 to Acetate Period 4 with the exception of the Steel Slag treatment during No Acetate Period 2. In the case of these modular bioreactors, mean bioreactor NO3-N load for the different periods was attributed to a combination of differences in mean bioreactor NO3-N concentration and HRT.

Phosphorus

Statistical analysis did not reveal any significant differences (p=0.259) for total phosphorus (TP) concentration among P treatments during 2016 (Table 6). The largest concentration of TP was observed during Acetate Period 2, which occurred during late spring and early summer. The smallest concentration of TP was observed during Acetate Period 2, which occurred during late autumn/early winter. Despite no significant difference among P treatments, the data showed reductions in TP for some treatments during all periods, except Acetate Period 4 compared to subsurface drainage source water. From the data, it did not appear that Acetate had any impact on TP concentration. During Acetate Period 4, all the treatments behaved as a source of TP compared to the subsurface drainage source water. This suggests that there was dissolution of phosphorus when the TP concentration in the subsurface drainage source water was lowest. Statistical analysis revealed significant differences (p=0.0544) for TP load among P treatments. Table 6 shows mean bioreactor TP loads during 2016. The largest mean bioreactor TP loads observed occurred during No Acetate Period 1 and the smallest during Acetate Period 2. There is a noticeable drop in TP load after the addition of acetate beginning with Acetate Period 1 (Table 7). The data also exhibit a gradual decline in mean bioreactor TP load from No Acetate Period 1 to Acetate Period 2. Mean bioreactor TP load increased from Acetate Period 2 to Acetate Period 3. Mean Total phosphorus load data did exhibit some variability between treatments during different periods. Mean bioreactor TP load for the different periods was attributed to a combination of differences in mean bioreactor TP concentration, HRT, water temperature, pH and oxidation-reduction potential.

CONCLUSION

Year 2016 was very wet with an annual precipitation higher by 41% compared to that of 30-year average. This unusual occurrence has allowed the experiment to be carried out till early December. On average, the bioreactors reduced nitrate concentration by 28, 28, and 31.5% at crushed concrete, limestone, and steel slag treatments, respectively; total phosphorus concentration reductions were larger than those of nitrate with 50.5%% at crushed concrete, 65.3% at limestone, and 62.2% steel slag treatments. Overall, nitrate load reduction occurred in the following ranks with steel slag (34.1%) > limestone (30.3%) > crushed concrete (23.3%). The largest annual load reduction for total phosphorus load was

29 observed at crushed concrete (37.4%) and followed by limestone (30.8%); steel slag was associated with the least reduction of TP load (27.6%). Hydraulic residence tile has played a major role in the reduction of concentration and load for nitrate. During Acetate Period 4 (early December), most of the treatments became source of phosphorus under a low temperature regime.

REFERENCES

Jaynes, D.B., T.C. Kaspar, T.B. Mooreman, and T.B. Parkin. 2008. In situ bioreactors and deep drain-pipe installation to reduce nitrate losses in artificially drained fields. J. Environ. Qual. 37:429–436.

Karanasios, K.A., I.A. Vasiliadou, S. Pavlou, and D.V. Vayenas. 2010. Hydrogenotrophic denitrification of potable water: a review. J. Hazardous Materials. 180: 20-37.

Lew B., P. Stief, M. Beliavski, A. Ashkenazi, O. Svitilica, Abid Khan, S. Tarre, D. de Beer, and M. Green. 2012. Characaterization of denitrifying granular sludge with and without the addition of external carbon source. Bioresource Technology, 124: 413- 420.

Robertson, W.D. and L.C. Merkley. 2009. In-stream bioreactor for agricultural nitrate treatment. J. Environ. Qual. 38:230-237. van Driel, P.W., W.D. Robertson and L.C. Merkley. 2006. Denitrification of agricultural drainage using wood-based reactors. Trans. ASAE 48:121-128.

Woli, K.P., M.B. David, R.A. Cooke, G.F. McIsaac, and C.A. Mitchell 2010. Nitrogen balance in and export from agricultural fields associated with controlled drainage systems and denitrifying bioreactors. Ecological Engineering, 36:1558-1566.

30

TABLES

Table 1. Experiment phases of no acetate and acetate addition to modular bioreactors

Phase Time period Calibration (No Acetate) Period 1 5/10/16 to 6/20/16 Period 2 10/25/16 to 11/18/16 Treatment (Acetate) Period 1 7/8/16 to 7/23/16 Period 2 7/30/16 to 8/8/16 Period 3 9/20/16 to 10/13/16 Period 4 12/1/16 to 12/5/16

Table 2. Cumulative discharge volume (m3), in chronological order, from bioreactors grouped by P-treatment No Acetate Acetate Acetate Acetate No Acetate Acetate P-Treatment Period 1 Period 1 Period 2 Period 3 Period 2 Period 4 Crushed Concrete 178 73 20 50 52 22 (n=3) Limestone (n=3) 179 47 12 54 49 21 Steel Slag (n=3) 200 52 9 30 54 14

Table 3. Mean porosity and hydraulic residence time (hour), in chronological order, for bioreactors grouped by P-treatment Porosit No No y Acetate Acetate Acetate Acetate P-Treatment Acetate Acetate volume Period 1 Period 2 Period 3 Period 4 Period 1 Period 2 (gal) Crushed Concrete 87 2.0 2.4 6.6 3.8 4.2 1.5 (n=3) Limestone (n=3) 85 1.9 2.5 6.5 3.5 4.6 1.5 Steel Slag (n=3) 85 1.6 3.1 7.9 3.4 8.3 1.7

Table 4. Mean NO3-N concentration (mg/L), in chronological order, from bioreactors grouped by P-treatment No No Acetate Acetate Acetate Acetate Acetate P-Treatment Acetate Period 1 Period 1 Period 2 Period 3 Period 4 Period 2 Crushed Concrete 22.9 9.6 0.4 13.5 17.9 18.0 (n=3) Limestone (n=3) 22.9 9.0 0.6 14.4 17.7 18.0 Steel Slag (n=3) 22.9 8.0 0.3 12.8 16.6 18.1 †Source water (n=1) 23.6 22.9 12.1 19.7 19.2 17.4

31

†Source water is subsurface drainage water

Table 5. Mean NO3-N load (kg), in chronological order, from bioreactors grouped by P- treatment No Acetate Acetate Acetate Acetate No Acetate Acetate P-Treatment Period 1 Period 1 Period 2 Period 3 Period 2 Period 4 Crushed Concrete 3.7 0.7 0.004 0.7 0.8 0.4 (n=3) Limestone (n=3) 4.2 0.6 0.001 0.8 0.7 0.4 Steel Slag (n=3) 4.6 0.5 0.005 0.8 0.4 0.3 †Source water (n=1) 5.6 1.3 0.2 1.1 0.7 0.3 †Source water is subsurface drainage water

Table 6. Mean total phosphorus concentration (ug/L) from bioreactors grouped by P-treatment No Acetate Acetate Acetate Acetate No Acetate Acetate P-Treatment Period 1 Period 1 Period 2 Period 3 Period 2 Period 4 Crushed Concrete 91 46 139 133 64 206 (n=3) Limestone (n=3) 96 52 107 90 52 79 Steel Slag (n=3) 109 100 128 90 29 62 †Source water (n=1) 121 260 764 110 64 52 †Source water is subsurface drainage water

Table 7. Mean total phosphorus load (g) from bioreactors grouped by P-treatment No No Acetate Acetate Acetate Acetate Acetate P-Treatment Acetate Period 1 Period 1 Period 2 Period 3 Period 4 Period 2 Crushed Concrete 14.8 2.3 1.0 4.6 2.7 1.5 (n=3) Limestone (n=3) 17.8 2.2 1.2 5.6 2.4 3.5 Steel Slag (n=3) 22.9 3.6 1.0 4.9 1.3 0.8 †Source water (n=1) 25.0 9.7 3.0 4.7 2.5 1.0 †Source water is subsurface drainage water

32

FIGURES

Figure 1. System setup from control structure, acetate pump station, and cube bioreactors.

Figure 2. Acetate pump station with 300-gal tank

33

Figure 3. Cube bioreactors, ISCO automated sampler, tipping bucket, and dataloggers

34

SUPPLEMENTAL IRRIGATION IN SOUTHWEST MINNESOTA FOR CORN AND SOYBEAN

Tamás Varga1, University of Minnesota Jeffrey Strock2, University of Minnesota

1Department of Soil, Water and Climate St. Paul, Minnesota 2Department of Soil, Water and Climate and Southwest Research and Outreach Center, Lamberton, MN

Keywords: Supplemental Irrigation, Drainage, Water & Nutrient Recycling

EXECUTIVE SUMMARY

Supplemental irrigation is an agronomic tool that has the potential to become a viable option for risk mitigation on humid continental, temperate climates with high crop production potential, such as in southwest Minnesota. This irrigation method supplements water for the crop from drainage water storage only at critical times when rainfall distribution is not meeting the crops demand. With this approach, important and expensive nutrients (nitrogen and phosphorus) can be recovered and recycled which may have been lost otherwise and become pollutants for the environment downstream. The 2016 growing season had adequate precipitation over the entire growing season and thus serves as a benchmark, or best case example.

35

SUPPLEMENTAL IRRIGATION IN SOUTHWEST MINNESOTA FOR CORN AND SOYBEAN

INTRODUCTION

Dryland corn and soybean farming carries a lot of risks. This production system is well established in southwest Minnesota and occurs on highly productive soils under humid conditions. Direct production risk mainly consists of unpredictable weather, especially considering precipitation amount and distribution. Climate change is affecting agriculture production substantially. Although the average precipitation and temperature are increasing, their extreme occurrences are also increasing, and furthermore their distribution within a year also worsen for crop production. Hence water management is facing emerging challenges due to these changes and requires continuous improvement from farmers, watershed managers, and from the whole society. Water carries important and expensive resources to the edge of the fields – like nitrogen and phosphorus – which become pollutants downstream. More water in the spring may result in delayed planting and slower soil warming, as well as higher drain flows. On the other hand, summer precipitation is decreasing, especially at times when crops needs are the highest. As Baker et al (2012) reveals, the anomaly of water related agriculture production risk, water logging and drought-induced losses many times happens within the same year and same field. At both occasions, spring time delays and summer stresses cause yield losses and make accurate crop and watershed management more difficult, impacting agronomics, environmental concerns and the farm economics as well on the socio- economical scale downstream. Managing agriculture drainage water differently may bring solutions to many of these issues. The development of water management strategies in agricultural production requires multi-disciplinary approaches and site-specific applications in order to increase its efficiency and lower the environmental impact and weather associated risks (Hatfield, et al., 2001). The Transforming Drainage project (https://transformingdrainage.org/) is investigating alternatives to see how these problems could be handled better and benefit stakeholders. Our main objectives are to evaluate the effects of drainage water retention and recycling as water management option with a combination of nitrogen management for crops (corn and soybean, Figure 1).

METHODS

The experiment is located at the University of Minnesota Southwest Research and Outreach Center (44°14'16.60"N, 95°18'17.59"W). The soil type of the experimental field is Normania loam (Fine-loamy, mixed, superactive, mesic Aquic Hapludolls). Primary tillage is in-line ripping in the fall and field cultivation for seedbed preparation in spring. The crop rotation is corn-soybean with both crops present every year. The corn hybrid is DKC 52-84 RIB at a seeding rate of 35,000 seeds/acres and the soybean variety is Asgrow 2035, at 150,000 seeds/acres planting rate. All other farming practices consist of standard practices in the area, such as weed control, row width (30 inches), and fertility plans. There are 24 individually drained (45’ x 50’) plots with a plastic barrier around them to a depth of 6 feet. Subsurface drains are installed 4 feet deep and simulate a drain spacing of 90 feet. The drainage water quantity and quality are manually measured at the drainage outlets.

36

The agronomic experimental design is Randomized Complete Block Design in split- array. Main effects are crop (corn and soybean) and four water management treatments (rain-fed, limited, full- and excess irrigation) in a factorial layout resulting in eight main plot treatments in each of three replications. Rainfed treatment does not receive any additional water, the goal of the full irrigation treatment is that it receives 100% evapotranspiration (ET) replacement and soil water content is kept between 60-80% field capacity. The limited irrigation treatment receives the same amount of water as full irrigation, but it is applied between V14 to the end of R2 (blister) phenological stages for corn as that is the most water sensitive period for yield losses. Soybean limited irrigation occurs between the R3 - R6 (beginning pod – full seed) phenological stages. The excess irrigation treatment receives water 25% above the full irrigation level. The irrigation water is delivered by drip irrigation system because of its high application efficiency (92%+) and uniformity even on small scales like a research plot. Drip lines are on 60” spacing. Water application speed is about 0.11 inches/hour, which is matched to the soil type’s infiltration rate. The water source is from the station’s drainage pond, about 500 yards away from the research plots. Sub-plots have six nitrogen application rates for corn (0-80-120-160-200- 240 lbs N/acres) all applied a few days prior to planting and incorporated. Soybean plots are treated as scavengers of any residual nitrogen with no nitrogen fertilizer applied. Measurements include ambient meteorological measurements (wind speed and direction, solar radiation) for site characterization. Plot level measurements are: crop yield, yield components (kernel weight, kernel number, plant population), total above ground biomass, non-destructive leaf area index, volumetric soil water content (0-36” depth in six increments), drainage flow and water quality. Starting from 2017, crop measurements will be expanded with canopy temperature, leaf level transpiration and stomata conductance measurements, as well as within canopy air temperature and relative humidity, as micro meteorological measurements along with an ambient mixed layer relative humidity and air temperature data collection.

RESULTS AND DISCUSSION

AGRONOMICS

Water management – irrigation treatments

During the growing season of 2016, there were no significant differences among the rainfed and irrigated treatments for either crop (Table 1 and Figure 2). However, irrigation lowered the yield variability, especially for soybeans (Figure 2). Both crops were irrigated after their most water sensitive phenological stages because neither the soil water content readings nor the weather conditions indicated the need for irrigation (Figures 3 and 4). Its important implication is that the yield heterogeneity caused by soil variability might be lowered with irrigation. The only parameter where visible differences were observed is the leaf area index (LAI) measurements in the soybean crop (Figure 5). The rainfed treatment had less leaf area for photosynthesis than the irrigated treatment, with no other differences among plots and treatments. This difference although, did not translate to yield difference.

37

Nitrogen management

Corn nitrogen treatments show a highly significant response to the increasing amount of nitrogen fertilizer (Figure 6). Because of the no differences between water management treatments, the nitrogen treatment results were combined and used as further replications for the nitrogen rate response analysis. Soil nitrate samples from the depths of 0-6 inches and 6-12 inches also did not show differences among water management treatments (Figure 7).

Weather conditions

The above average precipitation, which rain events were also very timely (Figure 3), did most likely contribute to the non-significant irrigation treatments and the lack of water and nitrogen management interaction. Crop water requirement is mostly driven by atmospheric conditions, also called crop evapotranspiration (ETc) (Allan and Richard, 1998). This value is adjusted for the given crop by its characteristics from the (reference) evapotranspiration (ET0). The ET0 was retrieved for the research location from the UW Extension Ag Weather website for Wisconsin and Minnesota ET predictions and is showed in Figure 3. Average adjusted crop water demand (ETc) numbers were obtained from Nebguide G1850 (Kranz et al. 2008). In this way, they represent a range of water requirement for crop production in contrast of the cumulative precipitation, but without soil water storage capacity. Corn and soybean reached record yields during 2016, which means they had record high ETc values as well, but precipitation and soil water storage together were able to meet this demand. The yield data from this experiment also indicated that irrigation would not further increase yields for either corn or soybeans (Table 1 and Figure 2). This suggests that some other factor could have been the yield limiting factor. With 2016 being a record yielding year, it represents a benchmark with which to compare previous years’ yield performances and upcoming years’ yield estimations. As an example of the drainage water recycling system at work, a large rain event resulted in 4.37 inches of water coverage on July 16th, 2016. The drying soil was recharged (Figure 4) and even tile flow started, producing almost peak flow measurements visible in Figure 8. An average of 2.13 inches of soil water content improvement was documented after the event. It did not include two days of ET losses, drainage, and run-off; however, the last two were captured in the reservoir and would have covered alone the irrigation water, which was applied a couple weeks later.

ENVIRONMENTAL IMPLICATIONS

Drain flow and nitrate concentrations

Water applications started on July 29th, to supplement water in a drier period of the 2016 growing season (Figure 3). Irrigation water application did not start or lead to increased drain flow; however, they only received a little more water than the rainfed treatments. Drainage water nitrate concentration samples are only partially processed at this date (5/26/2016-7/25/2016), and since these data are not covering the period after the irrigation, there are no conclusions that can be derived concerning irrigation treatments. However, these pre-irrigation samples show the variability among the randomly assigned main blocks 38 and the difference between crops. The concentration difference was about 4.8 mg/L higher in the drain water flowing from corn plots (31.63mg/L) relative to the soybean plots (26.88mg/L) (Table 2).

ECONOMICS

Farm scale and socio-economic outlook

The effects of supplemental irrigation on economics – including farm and water shed, maybe even socio-economic – cannot be evaluated after one year at one location, especially with a record yielding year in Minnesota’s history. Looking at only the farm scale economics and only for this single, unique year, might suggest, that the extra cost of the supplemental irrigation option did not carry a positive effect on the farm budget, by its extra direct and indirect expenses such as labor and energy and capital and depreciation, respectively. On the other hand, it stored drainage water (in a reservoir and in the soil profile) without decreasing yields or increasing drain flow, so it could be an effective watershed management tool as well. The research site where this experiment takes place is looked at as mostly homogenous small field but still has heterogeneity in its soil composition and topography. On a farm or watershed scale, the spatial and temporal variability is much larger, and as most of the soybean yield results show (Figure 2), supplemental irrigation has the capability to help improve production efficiency by lowering yield heterogeneity. Therefore, supplemental irrigation is an attractive risk mitigation option in many ways but comes with a cost. It needs proper evaluation spatially and temporarily, as well as on a large scale and beyond the boundaries of farm fields and production. Further evaluation is needed to discover the real economic potential on different scales and management situations.

CONCLUSIONS

The first year of the experiment (2016) provided a nearly optimal growing condition for corn and soybeans in Southwest Minnesota; thus, a record high average yield was set statewide. This result likely contributed to the lack of agronomic response of irrigation and lack of interaction between irrigation and nitrogen fertilizer treatments. The experiment, however, verified that this season set a benchmark for both crop yields in the area, which wouldn’t be possible to be identified otherwise. Furthermore, supplemental irrigation is more of a risk mitigation technique – especially from recycled drainage water – in the hands of farmers and watershed managers, rather than a necessary agronomic tool under current conditions of this humid region.

REFERENCES

Allen, Richard G. Crop Evapotranspiration: Guidelines for Computing Crop Water Requirements. FAO Irrigation and Drainage Paper; 56. Rome: Food and Agriculture Organization of the United Nations, 1998. Baker, John M, Timothy J Griffis, and Tyson E Ochsner. 2012. “Coupling Landscape Water Storage and Supplemental Irrigation to Increase Productivity and Improve 39

Environmental Stewardship in the U . S . Midwest” 48: 1–12. doi:10.1029/2011WR011780. Hatfield, Jerry L, Thomas J Sauer, and John H Prueger. 2001. “Managing Soils to Achieve Greater Water Use Efficiency : A Review Managing Soils to Achieve Greater Water Use Efficiency : A Review.” Agronomy Journal 93: 271–80. Kranz, William L, Suat Irmak, Simon J. van Donk, C. Dean Yonts, and Derrel L. Martin. 2008. “Irrigation Management for Corn.” NEBGUIDE, University of Nebraska- Lincoln Extension, 1–4. doi:10.2489/jswc.69.3.67A. UW Extension Ag Weather website for Wisconsin and Minnesota ET predictions http://agweather.cals.wisc.edu/sun_water/et_wimn - accessed on 09/25/2016. Wright, Jerry. 2002. Irrigation scheduling checkbook method http://www.extension.umn.edu/agriculture/irrigation/irrigation- management/irrigation-scheduling-checkbook-method/index.html Accessed on 7/12/2017

40

TABLES AND FIGURES

Table 1: Crop yield results and total irrigation water applied to the crops in 2016

Soybean Corn Irrigation Irrigation Yield Irrigation Yield Treatments [in] [Bu/ac] [in] [Bu/ac] Excess 1.47 66.9 1.50 208.9 Full 1.17 66.2 1.21 197.3 Limited 1.13 65.8 0.75 200.6 Rain-fed 0.00 63.1 0.00 206.6

Table 2: Average Nitrate concentration in tile drain water by crops

Nitrate Confidence Crop concentration StDev StErr Limit 95% [mg/Lit] Corn 31.63 2.95 0.26 0.52 Soybean 26.88 3.06 0.28 0.55

41

Figure 1: Scheme of drainage water recycling water for supplemental irrigation Source of image: https://transformingdrainage.org/practices/drainage-water-recycling/

42

Figure 2: Crop yield results by Water management treatments [Bushels/ac] for Corn (top figure) and soybean (bottom figure), respectively.

43

Figure 3: Cumulative Precipitation, evapotranspiration with the Historical average and average crop water demand

Figure 4: Average Available Soil Water for the 2016 growing season (Graph source: https://swroc.cfans.umn.edu/weather/soil-water-graphs/historic-soil-water-graphs)

44

Figure 5: Leaf area index measurements of the soybean crop by water management treatments

Figure 6: Distribution of corn yield data by nitrogen rates

45

Figure 7: Soil nitrate concentration changes during the growing season of 2016 in selected Nitrogen treatments by water management at the given soil depths [ppm]

46

Irrigation events

Figure 8: Average Drain Tile Water flows by Water management treatments

47

ASSESSMENT OF SIDE INLET DESIGNS

Jeffrey Strock1, Karl Bear2 and Bruce Wilson3

1Southwest Research and Outreach Center, University of Minnesota, Lamberton, MN. 2Department of Soil, Water and Climate, University of Minnesota, St. Paul, MN, Former Graduate Research Assistant 3Department of Bioproducts and Biosystems Engineering, University of Minnesota, St. Paul, MN.

Keywords: Side Inlet, Runoff, Sediment, Water Quality

EXECUTIVE SUMMARY

Ditches convey surface runoff water and subsurface tile drainage from artificially drained agricultural lands. A potential BMP for drainage ditches is the design or retrofitting of side inlets. Side inlets serve as surface runoff outlets for agricultural lands to drainage ditches. They also prevent gully erosion by providing a safe passage of surface runoff to the drainage ditch and remove water that would otherwise stand in a field and possibly result in crop damage or loss. The primary objective was to collect field data on the hydraulic characteristics and sedimentation processes for the different types of side inlets. The types of side inlets considered in this study were the widely used existing straight pipes and alternative designs consisting of flush pipes, Hickenbottom risers, rock inlets, and rock weirs. The experiments were designed around infrastructure developed during fall 2002 and 2012 at the University of Minnesota Southwest Research and Outreach Center (SWROC) near Lamberton, MN. The response of the different side inlet types was assessed using relatively small simulated runoff events of approximately 38 kL and a relatively large runoff event of 60 kL. Water quantity (discharge rate and total flow volume) and water quality samples were collected and analyzed for Total Suspended Solids (TSS). Cumulative soil mass at the outlet of each side inlet type was calculated as the summation of the products of the TSS concentration of each sample by the total volume of water for that sample. Sediment deposition was tied to the retention time in the basin. Side inlets with a longer retention time had a greater removal of sediment. All five controls showed the ability to trap sediment and decrease the sediment size in the basins before water is discharged into the ditch. For the ponded depths of approximately 10 inches for our tests, all five side inlet controls removed water before the ponded water would start stressing crops. The rock inlet was most efficient in removing water quickly without having it pond in the basin. The coil inlet trapped the most sediment in the basin, but had the longest standing water in the inlet by 14 hrs. In terms of water quality benefits, the coil inlet and rock weir performed the best but these designs may not be ideal in locations within close proximity to crops. However, the coil inlet and rock weir are ideal in locations where runoff ponding from the surrounding watershed will not damage crops if ponding exceeds more than 1 day. The rock inlet, flush pipe, and Hickenbottom riser have less risk for crop failure in locations within close proximity to crops but they do not trap sediment as efficiently as the coil inlet and rock weir.

INTRODUCTION 48

Ditches convey surface runoff water and subsurface tile drainage from artificially drained agricultural lands and are important to the agricultural economy of Minnesota and other Midwestern states. However, traditional methods of surface and subsurface drainage often result in degraded water quality. There are approximately 17,000 miles of public drainage ditches in Minnesota that are under the jurisdiction of MN Statute 103E (MN BWSR, 2006). There is increased interest in developing Best Management Practices (BMPs) for drainage ditches. Ideal BMPs mitigate the negative impact of artificial drainage while limiting their negative consequences to crops and farmland management. A potential BMP for drainage ditches is the design or retrofitting of side inlets. Side inlets serve as surface runoff outlets for agricultural lands to drainage ditches. They also prevent gully erosion by providing a safe passage of surface runoff to the drainage ditch and remove water that would otherwise stand in a field and possibly result in crop damage or loss. They are generally located in small depressions adjacent to ditch berms and collect surface water runoff from relatively small watersheds. Failures of these inlets are known to increase downstream sediment transport, reduced ditch conveyance capacity, increased nutrient loading and potentially cause the loss of productive farmland. However, little research has been done to quantify the impact of functional inlets on the peak flow rates, sediment loading to receiving waters, nutrient delivery and streambank erosion. Side inlets can be designed or retrofitted to temporarily store surface runoff, decreasing downstream peak flow rates and reducing the sediment contributions from croplands. There are many design variations of side-inlet controls. They include slotted standpipes, rock inlets, rock weirs and high-density drainage coils. Research is needed on the effectiveness of these alternative designs as potential BMPs for drainage ditches. The effectiveness of alternative designs for side inlets was the primary focus of this experiment. The primary objective was to collect field data on the hydraulic characteristics and sedimentation processes for the different types of side inlets. The types of side inlets considered in this study were the widely used existing straight pipes and alternative designs consisting of flush pipes, Hickenbottom risers, rock inlets, and rock weirs. This paper will describe the results from 2013 field experiments. The objectives of this experiment were 1) to evaluate the impact of different types of side inlets on their hydraulic characteristics and sedimentation processes and 2) to obtain hydraulic and sedimentation parameters needed to model them.

METHODS

This section describes the main experimental components of this research project. The project was designed around infrastructure developed during fall 2002 and 2012 at the University of Minnesota Southwest Research and Outreach Center (SWROC) near Lamberton, MN. A drainage channel research facility incorporating a paired study approach in the design was constructed in 2002 (Strock et al., 2005). A 200-m reach of existing drainage channel was converted into a system of four parallel channels: two inner experimental channels and two outer, overflow/diversion channels. In 2012, one of the outer overflow/diversion channels was redesigned to accommodate five side-inlet control types for experimentation. The side inlets were the flush pipe, Hickenbottom riser, rock inlet, coil inlet, and the rock weir. Flows from the flush pipe, Hickenbottom riser, rock inlet, and coil inlet were directed to a 2 inch diameter outlet pipe in order to increase retention times in the basins. The response of the different side inlets was

49 assessed using relatively small simulated runoff events of approximately 38 kL and a relatively large runoff event of 60 kL. Data were collected and analyzed for two replicates of the small event for each of the side inlets. Data for the large event were only collected for the flush pipe and Hickenbottom riser side inlets. The five experimental side inlet controls enabled researchers to evaluate various performance characteristics of the side inlet controls. The side inlet controls, while smaller than working types, were designed with characteristics like those along agricultural drainage channels in the region. Each side inlet control system was equipped with a flow monitoring device on the inlet and outlet ends which enabled the partitioning of inflows and manipulation of flow velocities through the system.

Side Inlet Description

Each side inlet was installed in its own corresponding basin, forcing water to pool around each individual inlet before it drained into the ditch. The edges of the basins were approximately 15 to 20 ft from the ditch. Because of the design of the rock weir, the edge of the basin was 10 ft from the ditch. Each inlet, with the exception of the rock weir, was tiled underground through the basin with the outlet being exposed 1 to 2 ft above the ditch’s water level.

Side Inlet Control Descriptions

For the experiment, the flush pipe was an 8 inch diameter tile drain. This pipe was located flush to the surface of the basin. An open grate was placed over the opening to block large residue and debris from entering the tile without restricting water flow. The slotted tile riser was an 8 inch diameter tile with a Hickenbottom HBI-8810 riser installed to increase retention time of water in the basin. The rock inlet was 15 ft long and had a 4 inch diameter perforated tile at the bottom. A pit of 15 ft by 15 by 10 ft was backfilled with coarse aggregate rock. The coil inlet was a 4 inch perforated tile buried below the middle of the basin. Starting at a depth of 2 ft below the soil surface, the tile coiled to the depth of 4 ft with the spiral having a 10 ft diameter. The rock weir was 10 ft in length and rose 1 ft above the basin. Surrounding the weir was a berm that held water from entering the ditch unless it overtopped the weir. The rock weir was constructed with rocks set on geotextile fabric. Although the outlets for the flush pipe, Hickenbottom riser, rock inlet, and coil inlet were 8 inch diameter tiles, the flow was restricted by attaching to a 2 inch diameter outlet pipe in order to increase retention times in the basins. Reducers were carefully taped in order prevent leakage from pipes under pressure from the ponded water in the basin.

Runoff Simulation

Small and large runoff events were simulated at the five side inlets. Small runoff events of approximately 38 kL were performed twice on each inlet for a total of ten events. Data were also collected from a large runoff event of approximately 60 kL for two of the inlet designs. Because of time and basin restriction, this event was only performed once for the flush pipe and once on Hickenbottom riser. To simulate a runoff event, water was pumped from water bladders sitting above the basin down into the basin approximately 20 to 30 ft from each inlet. Sediment was manually and continuously added to the water from the bladder through a funnel and PVC pipe system in order to ensure consistent mixing over the course of the event.

50

Runoff Event Sampling

During runoff events, instantaneous measurements of flow rate, stage, turbidity, and one liter water samples were collected from the outlet for sediment delivery. Water samples were collected with an automatic sampler (Teledyne ISCO). Inflow rate was sampled every minute during the runoff event using a flow meter and datalogger installed in the PVC pipe system. From the start of the runoff event to the time water stopped flowing at the outlet, the following measurements and samples were taken simultaneously in five minute intervals: outlet flow rate, stage, turbidity of the inflow and outflow, and water samples of the inflow and outflow for total suspended solids (TSS) and particle size analysis. Flow rate at the outlet was sampled with a 10 L bucket and stopwatch. Stage of the retained water in the basin was recorded next to the inlet of each basin using a meter stick. Instantaneous turbidity measurements were recorded every five minutes from the inflow and at the outlet using a Hach 2100Q portable turbidimeter. ISCO samplers were used to simultaneously sample the same place turbidity measurements were taken at both the inflow point and at the outlet, each taking one liter samples. Simultaneous measurements were used to create a turbidity/TSS relationship.

Lab Analysis

Samples were transported to the lab, put in a 4° C cooler and analyzed for TSS and particle size distribution at a later date. Sediment samples were allowed to settle in the ISCO bottles, followed by pipetting 75% of water out, and pouring the remaining contents into aluminum weighing boats. Samples were dried at 90° C and weighed to calculate TSS. Multiplying the TSS concentration of each sample by the total volume of water for that sample and adding the TSS concentration of each sample to the previous total calculated cumulative soil mass at the outlet. After the TSS analysis, samples were poured into a nested sieve set, mechanically shaken, and weighed to determine the particle size distribution.

RESULTS AND DISCUSSION

Small Runoff Event

Runoff Simulation. Water was pumped into the basins at an average rate of 6.31 L/s until the approximate total of 35 kL was reached. For each runoff event, 180 kg of soil was funneled into the water before it was pumped into the basin. Sediment concentration of the incoming runoff averaged 5.19 g/L.

Hydrology. Cumulative discharge at the outlet was highest in the rock inlet followed by the Hickenbottom riser, flush pipe, rock weir, and coil inlet (Fig. 1). The rock inlet had the highest discharge because water did not pond in the basin. The rock pit volume and discharge rate was large enough to prevent significant ponding. Water did not have time to infiltrate into the basin before being drained through the outlet. Besides the coil inlet, the other four side inlets stopped discharging water before 3 hr (Fig. 1). For the coil inlet, water ponded in the basin until it started discharging at the outlet 100 minutes into the event and continued to discharge water for 16 hr (approximately 14 hours of ponded water). Because the coiled tile was buried below the soil surface, water had to saturate the soil before it was removed by the tile, causing the delayed 51 discharge. For the rock weir, water ponded until it reached the crest of the weir. Since the weir held water back, the water had time to infiltrate before overtopping. Water below the weir slowly infiltrated into the basin once the runoff event ended. The coil inlet had the smallest maximum discharge rate of the five side inlet controls at .12 L/s with the Hickenbottom riser having the second slowest maximum discharge rate with 3.94 L/s.

Figure 1. Cumulative water discharge at Figure 2. Cumulative soil mass at outlet for the outlet for each side inlet control. each side inlet basin. Coil inlet soil mass increases to 0.16 kg after 160 min. Water Quality. Cumulative soil mass at the outlet was highest in the rock inlet followed by the Hickenbottom riser, flush pipe, rock weir, and coil inlet (Fig. 2). The rock inlet had the highest cumulative soil mass and therefore lowest trap efficiency. Minimal ponding occurred with this side inlet, giving sediment less time to settle out. The cumulative soil mass at the outlet of the coil inlet was 0.16 kg and is not accurately shown in Fig. 2. The coil inlet’s cumulative soil mass at the outlet was 91% less than the rock weir, which had the second lowest cumulative soil mass. Blind inlets constructed similar to the coil inlets have reduced sediment loads when draining closed depressions (Smith and Livingston, 2013). A summary of the small runoff events is given in Table 1. First, consider the cumulative effluent sediment mass. It was approximately constant for the flush pipe, Hickenbottom riser and rock inlet. Since the influent sediment mass was approximately constant among the tests, trap efficiencies of these side inlets were nearly equal. The effluent mass was slightly less for the rock weir, corresponding to a larger trap efficiency. Since the outlet mass was quite low for the coil inlet, the trap efficiency for this inlet was high. The maximum stage for flush pipe, Hickenbottom riser and rock inlet tests were relatively small. The ponded depth for the rock inlet was very small.

Table 1. Summary of small runoff events Flushpipe Hickenbottom Rock Inlet Coil Inlet Rock Weir Incoming Cumulative inflow (kL) 37.37 35.31 35.15 37.33 34.56 Outgoing Max stage (cm) 13.00 14.50 2.80 23.90 18.25 Max discharge rate (L/sec) 4.45 3.94 5.23 0.12 5.77 Cumulative discharge (kL) 26.52 28.04 29.44 4.98 23.17 Cumulative soil mass (kg) 2.23 2.31 2.48 0.16 1.77

52

Large Runoff Event

Runoff Simulation. Water was pumped into the flush pipe and Hickenbottom riser basins at an average rate of 12.618 L/s until the total inflow volume of 60 kL was reached. For each runoff event, 390 kg of soil was funneled into the water corresponding to average influent concentration of 6.5 g/L.

Hydrology. The Hickenbottom riser had a higher stage than the flush pipe with a maximum stage of 26.7 cm (Table 2). The cumulative discharge at the outlet was higher with the Hickenbottom riser (Table 2).

Water Quality. Cumulative soil mass at the outlet was higher in the flush pipe compared to the Hickenbottom riser (Table 2). While the flush pipe had a higher cumulative soil mass at the outlet of the large event, the Hickenbottom riser was higher than the flush pipe in the small event. It is possible that with a longer ponding time and more water, the Hickenbottom riser traps more sediment than the flush pipe.

Table 2. Summary of large runoff events. Small events and large events are labeled (S) and (L), respectively.

Flushpipe (S) Hickenbottom (S) Flushpipe (L) Hickenbottom (L) Incoming Total inflow (kL) 37.37 35.31 59.32 62.59 Outgoing Max stage (cm) 13.00 14.50 24.10 26.70 Max discharge rate (L/sec) 4.45 3.94 4.14 4.25 Cumulative discharge (kL) 26.52 28.04 40.22 42.30 Cumulative soil mass (kg) 2.23 2.31 2.56 2.42

SUMMARY

Side inlet controls such as the flush pipe, Hickenbottom riser, rock inlet, rock weir, and coil inlet prevent gully erosion, control the flow rate of runoff, and trap sediment as runoff flows through the controls. All five controls showed the ability to trap sediment and decrease the sediment size in the basins before water is discharged into the ditch. For the ponded depths of approximately 10 inches for our tests, all five side inlet controls removed water before the ponded water would start stressing crops. The rock inlet was most efficient in removing water quickly without having it pond in the basin. The coil inlet trapped the most sediment in the basin, but had the longest standing water in the inlet by 14 hrs. In terms of water quality benefits, the coil inlet and rock weir performed the best but these designs may not be ideal in locations within close proximity to crops. However, the coil inlet and rock weir are ideal in locations where runoff ponding from the surrounding watershed will not damage crops if ponding exceeds more than 1 day. The rock inlet, flush pipe, and Hickenbottom riser have less risk for crop failure in locations within close proximity to crops but they do not trap sediment as efficiently as the coil inlet and rock weir.

53

Additionally, new or revamped side inlet controls should be proposed and tested to achieve a suitable outflow rate along with an increase in sediment trapping efficiency. For example, the coil inlet in this study may be better designed by increasing the amount of coil, changing the coil pattern, and filling the coiled area with coarse aggregate rock, while leaving the soil as surface layer. Smith and Livingston (2013) used a coil inlet design with coarse aggregate rock backfill and achieved an outflow rate of 1.2 L/s compared to 0.12 L/s attained in this study. Side inlet controls trap sediment, while controlling flow rate and preventing gully erosion. However, an improvement in sediment trapping efficiency while achieving a suitable water retention time is still attainable by altering the current designs.

REFERENCES

MN BWSR. 2006. Public drainage ditch buffer study. Prepared by the Minnesota Board of Water and Soil Resources, February 2006. Available at: http://www.bwsr.state.mn.us/publications/bufferstudyweb.pdf. Accessed: April 5, 2010.

Smith, D.R. and S.J. Livingston. 2013. Managing farmed closed depressional areas using blind inlets to minimize phosphorus and nitrogen losses. Soil Use and Management. 29 (Suppl. 1): 94-102.

Strock, J.S., G.R. Sands, D.J. Krebs, and C. Surprenant. 2005. Design and testing of a paired drainage channel research facility. Appl. Eng. Agric. 21:63-69.

54

BANK EROSION IN THE MINNESOTA RIVER BASIN

Satish C. Gupta1 and Andrew C. Kessler2

1Raymond Allmaras Professor of Emerging Issues in Soil and Water; Department of Soil, Water, and Climate; University of Minnesota, St. Paul, MN 2 Scientist; Department of Soil, Water, and Climate; University of Minnesota; Currently Senior Scientist at Houston Engineering, Maple Grove, MN

INTRODUCTION

Minnesota River is one of the most turbid rivers in the State of Minnesota. According to USGS (Payne, 1994), sediment loads in the Minnesota River at Mankato varied from 0.2 to 3.3 million tonnes per year from 1968 to 1992. From 2000-2005, Minnesota Pollution Control Agency estimated an annual sediment loads of 0.3, 0.8, 0.7 and 0.7 million tonnes per year in the Minnesota River at Judson, St. Peter, Jordan, and Fort Snelling, respectively (https://www.pca.state.mn.us/sites/default/files/mnriver-0308- matteson.pdf). At the HUC008 watershed scale, Musser et al. (2009) showed that the average sediment yield from 2000-2008 has varied from 27 lbs/ac (30 kg/ha) in Lac Qui Parle watershed to 656 lbs/ac (735 kg/ha) in the Le Sueur River watershed (Fig. 1). This paper briefly outlines the underlying reasons for high sediment loads in the Minnesota River and its tributaries, the associated variability in sediment yield among its watersheds, and the difficulties of achieving sediment load reductions in the Minnesota River Basin.

Figure 1: Distribution of sediment yield averaged over nine years (2000-2008) among various watersheds contributing to the Minnesota River. The graph was taken from the State of the Minnesota River Report by Musser et al. (2009).

55

Description of the Minnesota River Basin

Some 13,000 years BP, parts of Minnesota and North Dakota in United States and much of Manitoba, parts of Saskatchewan and Ontario in Canada were covered with Glacial Lake Agassiz. The lake was formed from melt waters of the continental ice sheet that developed during the Wisconsin glaciation some 30,000 to 100,000 years BP (https://en.wikipedia.org/wiki/Lake_Agassiz). The southern end of Glacial Lake Agassiz broke and released torrents of water that excavated millions of tonnes of soil leading to the creation of the Minnesota River valley that is 1.9-5.0 km (1.2-3.1 miles) wide and 30-38 m (100-125 feet) deep (Thorleifson, 1996). After the surface of the Minnesota River valley was lowered, small rivers and streams in the Minnesota River Basin were left stranded. These rivers and streams have been eroding at the base to reach the level of the Minnesota River (Gran et al., 2009) and that has led to the development of small and tall river banks along streams in the Minnesota River Basin. Some of the river banks are as tall as 60 m (Kessler et al., 2012). Kelley and Nater (2000) identified the Minnesota River and its tributaries as the major source of sediments to Lake Pepin, a natural impoundment on the Mississippi River bordering Minnesota and Wisconsin. Gupta and Singh (1996) showed that as much as 65% of the sediments in the Minnesota River at Mankato came from river banks in 1990-1992. Subsequently, LiDAR estimates suggest that rivers in the Greater Blue Earth River Basin (GBERB) contribute the majority of sediments to the Minnesota River at Mankato (Thoma et al., 2005; Belmont et al., 2011; Kessler et al., 2012). Kessler et al. (2012) estimated that river bank sloughing along the Blue Earth River, the Le Sueur River and their tributaries in Blue Earth County, MN contributed 48% to 79% of the fine sediments from 2005 to 2009. However, quantitative and qualitative historical records show that high sediment loads in the Minnesota River or its tributaries is not a recent phenomenon. One of the first sets of aerial pictures (Fig. 2) taken in 1937-38 (during the relatively dry period) clearly show that both the Blue Earth River at Mankato and the Minnesota River at Fort Snelling were turbid/muddy. In a 1907 USGS paper, Dole and Wesbrook reported that ordinarily the turbidity of the Minnesota River at Mankato was between 10- 40 ppm but during spring flush, turbidity often increased to 600-800 ppm. These observations are consistent with historical observations recorded by travelers before the mass immigration of Europeans to the territory starting in 1850. For example, Featherstonhaug (1847), a well-known geologist of his time describe the Blue Earth River at Mankato in 1835 as “…a bold stream, about eighty yards wide loaded with mud of blueish colour, evidently the cause of the St. Peter’s being turbid.” St. Peter was the previous name of the Minnesota River. In 1850, population of the Minnesota Territory was 6,077 and this muddy blue color could not have been caused by immigrants who mostly lived near the Twin Cities. Featherstonhaug further wrote that this river “…form about half the volume of St. Peter’s, and is very rapid stream.” This is very similar to USGS (Payne, 1996) finding that showed the Blue Earth River contributed 46% of the total flow and 55% of the sediment load in the Minnesota River at Mankato.

56

Fig. 2a Fig. 2b

Figure 2a: An aerial picture showing the muddy Minnesota River joining the Mississippi River at Fort Snelling, 30 June 1937. Figure 2b: An aerial picture showing the muddy Blue Earth River joining the Minnesota River at Mankato 1938 (MN DNR-MHAPO-1938).

Other historical records of the Minnesota River being muddy/turbid include J. Wesley Bond’s (1856) description of 1850 exploration tour up the Minnesota River from St. Paul to Mankato as “….celebrated voyage, as day after day we ascended the swollen and turbid waters of the St. Peter’s.” Evan Jones (1962) reported a traveler from the St. Croix describing the Minnesota River in 1856 as “…. a dirty little creek.” In letters to her family out East, Mary Ann Clark Longley Riggs, a pioneer missionary at Lac Qui Parle wrote on 18 September 1837 that the quantity of water she crossed during her journey from Fort Snelling to Lac Qui Parle was “…very small, though several muddy, sluggish brooks bear the name of rivers” (Riggs, 1996). Summarizing Major Stephen Long’s expedition to the source of St. Peter, W.H. Keating (1824) wrote that the St. Peter River in the Dakota language is called “Watapan Menesota”, which means “the river of turbid water.” The author further stated that “The name given to St. Peter is derived from its turbid appearance, which distinguishes it from the Mississippi, whose waters are very clear at the confluence.” What we currently observe downstream on the Mississippi River from its confluence with the Minnesota (Fig. 3) is also similar to what Abbie Leavitts, a school teacher at Prescott, WI wrote in her diary in 1857 “….Nature has donned her loviest garb, but the broad Mississippi with its dark turbid waters and bold rugged shores can never …enrapture my soul like the woods and fields and groves and hills of my loved New England home” (Ahlgren and Beeler, 1996). The above historical descriptions of turbid/muddy waters along with present day LiDAR measurements clearly show that the Minnesota River and its tributaries have been historically muddy and the main reason for this muddiness is the river and ravine bank sloughing.

57

Town of Prescott, WI

The St. Croix River

The Mississippi River after the Minnesota River has joined it upstream

Figure 3: The St. Croix River joining the Mississippi River at Prescott. The photograph was taken on 2 June 2004 by David Morrison of MPCA.

Mechanisms of Bank Sloughing

There are several mechanisms that cause river and ravine bank sloughing in the Minnesota River Basin. These include freezing and thawing, seepage, fluvial erosion at the base of the bank, and rapid shifts in river meanders due to flooding. Blue Earth County with Mankato as the county seat has the most river miles of any county in Minnesota. All the rivers in this county generally flow north which sometimes results in ice blockage in early spring causing large shifts in river position.

Fig. 4a Fig. 4b

Seepage Zone

Impeding Frozen seeps layer

Figure 4a: Frozen seepage water along the face of a bank in Blue Earth County, MN. Figure 4b: Recently sloughed bank along the Maple River. The sloughing occurred above the seepage zone between 6 March and 5 May 2017.

Some parts of the Greater Blue Earth River Basin which includes the Le Sueur River, the Blue Earth River and the Watonwan River watersheds were at one time covered with Glacial Lake Minnesota. In this area, the surface soils are primarily lacustrine with high organic matter content from subsequent prairies that grew in this region. Subsurface soils and river bank materials are mostly consolidated tills brought to the region by glaciers. Consolidated tills slow the downward percolation of precipitation water leading to perch water table conditions and thus horizontal inter layer flow that appears as seepage on the face of the river bank (Fig. 4a). Since water cannot seep out at the soil-air interface until soil is saturated, it leads to weakening of the soil resulting in hill slope failure above the seepage layer (Fig. 4b). If the perched water table conditions persist for a longer period, 58 the pore water pressure can build up behind the bank face leading to a rotational failure. Seepage and rotational failure generally go hand-in hand and their effects can be easily recognized through a crescent type of excavation on the bank face (Kessler, 2017; unpublished data). These effects are magnified during wet periods like the spring/summer 2014 when there were several mass failures in the towns of Blakely and Henderson, and even behind the St. Mary/Fairview Hospitals in Minneapolis. In all these cases, there were no rivers touching the hillslopes (Fig. 5). Fig. 5a Fig. 5b

Figure 5a: A sloughed hillside in the back of the St. Mary/Fairview Hospitals in Minneapolis, MN in 2014. Picture was taken from Pioneer Press (https://mobile.twitter.com/PiPressPhotos/status/480173316448088064/photo/1) Figure 5b: A hill side failure outside Henderson. Picture was taken from Le Sueur News- Herald (http://www.southernminn.com/st_peter_herald/news/article_a82c0513-c808- 5b26-8dd5-26d77699e85c.html)

Another major mechanism contributing to bank sloughing is the thawing of frozen banks in winter and during early spring. Winter thawing mostly occurs on south and west facing slopes when sun is able to thaw out a few inches of the bank surface. Under those conditions, surface few inches of the bank just slides down by gravity to the base of bank which is later carried downstream by the flowing water in spring and summer. In early spring when air temperatures are above 0 °C (32 °F), bank surface thaws out and again slide down by gravity to the base of the bank. The magnitude of bank failure due to spring thawing is generally much larger than the isolated winter thawing because of deeper thawing in the bank surface (Fig. 6).

Fig. 6a Fig. 6b

Figure 6: Thawing caused failure along the Le Sueur River as well as by the road side.

59

Fluvial erosion is another mechanism that is continuously occurring throughout the year. However its effects are much more pronounced during early spring when the toe of the slope is also thawing. As the toe of the bank is thawing, large chunks of soil fall into the river as the river water is rubbing against the bank. Relative to mass failure by seepage and thawing, the effects of fluvial erosion at a given bank are relatively small unless the river reaches a flood stage and thus changes its course (Fig. 7a). Under those conditions, there is a substantial movement of soil because of the carving of a new path or a channel (Fig. 7a). A recent example of this erosion occurred in 2016 fall flood along the Le Sueur River near River Park Drive in Mankato (Fig. 7b). This flood was caused by over nine inches of rainfall in St. Claire in a two day period (20-22 September 2016) after relatively wet fall (http://www.dnr.state.mn.us/climate/journal/160921_22_heavy_rain.html). The flood waters not only took out parts of the backyard from several houses in this area (Fig. 7b) but also excavated a large track of floodplain deposit near Co. Rd. 16 in Mankato.

Fig. 7a Fig. 7b

Figure 7: (a) Overlaying of the 2009 position of the Blue Earth River on 1938 photograph. At one place the river has moved 120 meters. (b) Flooding and bank erosion in the residential area along the Le Sueur River from over 9 inches of rain 20-22 September 2016 at St. Claire. Photograph 7b is courtesy of Rick Moore.

Factors Affecting Bank Sloughing/Erosion

Although the landscape and cropping systems are similar, sediment yield from the southern and eastern watersheds such as the Blue Earth River and the Le Sueur River watersheds, are 6 to 30 times more (549-656 lbs/acre vs. 27-103 lbs/acre) than the northern and western watersheds such as the Lac Qui Parle River, the Yellow Medicine River, and the Redwood River watersheds (Fig. 1). This is primarily due to more rainfall and presence of taller banks in the Blue Earth and the Le Sueur River watersheds compared to northern and western watersheds. For example, 30-year (1981-2010) average precipitation in the Le Sueur River and the Blue Earth River watersheds is 32 inches (812 mm) compared to 26 inches (660 mm) in Lac Qui Parle watershed (Fig. 8). Also, the effects of climate change on precipitation are more in the southern and eastern watersheds of the Minnesota River Basin than the northern and western watersheds. For example, Melillo et al. (2014) showed a 10-15% increase in precipitation in Minnesota due to recent climate change. Since average precipitation is more in southern and eastern watersheds than northern and western watersheds, the 15% increase translates to larger quantities of water in southern and eastern watersheds. A comparison of 30-year averages (1981-2010 vs. 1951-1980) shows that the

60

Blue Earth River and the Le Sueur River watersheds received 2.72 in/year (69 mm/year) additional precipitation compared to 1.41 in/year (36 mm/year) in Lac Qui Parle watershed (Fig. 8). This additional water could be an important factor in causing additional bank sloughing through seepage and fluvial erosion processes in the wetter watersheds such as the Blue Earth River and the Le Sueur River watersheds.

Figure 8: Changes in 30-year average precipitation in various HUC008 watersheds in Minnesota

Although long-term systematic data on precipitation intensities is not available for many watersheds in the Minnesota River Basin, it is well recognized that there has also been an increase in rain intensities in the area. Melillo et al. (2014) showed that there has been a 37% increase in the amount of precipitation falling in very heavy events (the heaviest 1%) from 1958 to 2012 in Midwestern United States including Minnesota. This type of increase in precipitation intensity will also lead to more flashiness in rivers. Of course, the recent example is greater than nine inches of rain on 20-22 September 2016 at St. Claire and Waseca, MN. This storm caused flooding in both these areas. According to Mark Seeley (2011, Personal communication), Waseca gets over 8 inches (200 mm) more precipitation every year now than in 1921-1950. This is over 30% increase in precipitation. Tile drainage is extensive in the Minnesota River Basin. Arguments have also been made that tile drainage has increased river flows. Physics of water flow through soils and river flow data does not indicate that. Presence of a drain pipe in the soil does not lead to increased river flows. This can be demonstrated if one visualizes pipes of different diameter connected in a series (similar to pipes connecting the house faucet to the water tower). Rate of water discharge through the connected pipes is controlled by the pipe with the smallest diameter (in the above example it is controlled by the house faucet). The same principle applies to the soil-tile system i.e. it is the smallest soil pores that determine the rate at which water drains out of the soil and drips into drain tile rather than the size of the buried drainage pipe. Relationships of river discharge vs. precipitation also show that recent increases in river flow are associated with increases in precipitation and not due to land use land cover changes such as tile drainage or adoption of corn-soybean cropping system (Gupta et al., 2015, 2016a,b,c,d,e).

Early settlers also drained many wetlands both in the rural and urban areas to make this area economically viable. For example, Lake Lamprey was drained in downtown St. Paul 61 to make room for the airport. Similarly, Lake Jackson near Amboy was drained to develop the area for agriculture. However, the magnitude of changes in surface water storage due to drainage is rather small due to relatively low relief of the Minnesota River Basin i.e. 33% of the area is <2% slope and 74% of the area is <6% slope. Similarly in the Greater Blue Earth River Basin, 54% of the land is <2% slope and 93% of the land is <6% slope. Using airborne LiDAR, Kessler and Gupta (2015) estimated the historic depressional storage of the Greater Blue Earth River Basin at 152 mm but a majority of this storage (53%) was in large depressions (>40 ha) that comprised <1% of the observed depressions. Considering that many of the potholes and wetlands in the area are shallow (Kessler and Gupta, 2015) and dry out during the growing season (Shjeflo, 1968), Gupta et al. (2015) concluded that the net effect of wetland drainage on evapotranspiration (ET) was likely small. This is mainly because the seasonal wetlands have been replaced with corn and soybean that continuously extract water from the root zone and convert it into ET. Comparatively, the shallow depressions before drainage likely lost the accumulated free water as evaporation rather quickly and there was not much loss of subsurface water to ET after depressions dried out in mid- or late-summer. Recently, arguments have also been made that initially most sediment in Lake Pepin came from near-channel sediment sources and then shifted to agricultural field sediments around 1850s and then revert back to near-channel erosion starting around 1950s (Schottler et al., 2010; Belmont and Foufoula-Georgiou, 2017). This type of conclusions must be reconciled against four scientific facts: (1) most river bank sloughing in the Minnesota River Basin is caused by seepage, thawing, pore water pressure build up and fluvial erosion. These are natural processes and it is highly unlikely that any of these processes stopped at any given time. This means river banks likely remained a major source of sediments in the Minnesota River Basin at all times. (2) Seepage and pore water pressure are driven by precipitation and if upland water erosion was dominating at any given time due to increased precipitation then seepage and pore water pressure processes must be significant too. (3) Much of the GBERB is relatively flat and one needs a slope in the landscape to get the upland eroded sediments transported to rivers. In comparison, banks are right next to the flowing river and bank sediments will be carried downstream by the flowing water. (4) There have been significant river channel modifications including construction of levees; straightening, dredging and deepening of river channels; and construction of wing dams that confined river flow in a smaller channel thus helping it scoured its base and deepen it (Gupta et al., 2011). All of these river modifications prevented spreading of water and sediments over much larger flood plain areas and thus forced more sediment to move downstream to Lake Pepin. As an example, Mankato used to get flooded and now there has not been any flooding since the levee was constructed in mid 1970s. The sediments that used to be deposited in Mankato and other flood plain areas are now going downstream to Lake Pepin. The Mankato City Center hotel next to the Minnesota River is built on a marsh land. They had to use over 1000 piling to lay its foundation.

Another source of sediment that has not received much attention in the Minnesota River Basin is the ravines. Relative to number of rivers, there are many ravines in the Minnesota River Basin. Similar to river banks, ravine banks are sloughing as well and their base is also eroding to reach the level of the Minnesota River or the other rivers in the basin. Very

62 little effort has been made in large scale characterization of sediment losses from ravines in the Minnesota River Basin. This is partially due to vegetative cover within ravines which significantly reduces the accuracy of LiDAR based measurement of the underlying ravine banks. Figure 9 gives a visual magnitude of the ravine erosion in Salsbury Hills near the City of Belle Plaine, MN. Figure 9a shows the extent of coarse sediments that were excavated during culvert and channel cleaning prior to 28 September 2012. Then a comparison of the culvert pictures (Figs. 9b and 9c) show filling of the channel back to its original level (arrow on the culvert) in over a year. Considering coarse sediments in Fig. 9a are many truck loads and fine particles in glacial till material are generally >50% would suggest that there is considerable bank and base erosion from ravines that is contributing sediments to the Minnesota River, its tributaries and eventually to Lake Pepin.

Fig. 9a

28 September 2012 7 November 2013

Fig. 9b Fig. 9c

Figure 9: Pictures showing the extent of base and bank erosion from a ravine in Salsbury Hill near Belle Plaine, MN. Figure 9a shows the extent of coarse sediments that were excavated from the channel during cleaning. Figure 9b shows the base of the culvert after cleaning and then after filling (Fig. 9c) just in over a year.

Matheson and Baskfield (2008) showed that 2003-2006 average sediment yield from various watersheds of the Minnesota River Basin was related to area with >12% slope (https://www.pca.state.mn.us/sites/default/files/mnriver-0308-matteson.pdf). These areas are right next to the river channels, generally forested, and represent the ravines. In spite of heavy non-agricultural vegetation, these ravines are sloughing due to seepage, thawing, and of course due to surface runoff processes. These processes will continue until the base of these ravines is at the same level as the river in which they are emptying into.

63

Strategies for Reduction of Sediment Loads and Associated Pitfalls

The above discussion clearly shows that the Minnesota River and its tributaries have been historically muddy and turbid. This is primarily due to glacial till nature of the parent material which is densely packed and relatively high in fine particles. Furthermore, high sediment loads in rivers of the Minnesota River Basin are primarily due to bank sloughing that is caused by natural forces of seepage, thawing, and flowing water. Since land use conditions including cropping systems are similar across much of the Minnesota River basin, a comparison of sediment yield between the Lac Qui Parle and the Blue Earth or the Le Sueur River watersheds (Fig. 1) clearly shows that high loads in the Blue Earth River and the Le Sueur River watersheds are primarily because these watersheds receive more precipitation and they have taller banks. Since cropping systems are similar between western and southern watersheds in the Minnesota River basin, it clearly shows that differences in sediment load or sediment yield (Fig. 1) are not due to differences in land management and cropping system practices. This suggests three possibilities for reducing sediment loads from bank sloughing: (1) stabilization of river banks using rip-rap (Fig. 10), (2) use of high ET vegetation in the landscape, and (3) storage of excess precipitation in the watersheds.

Figure 10: Rip-rap holding the bank in place along the Le Sueur River.

Considering seepage and thawing are the two main mechanisms of bank sloughing, use of rip-rap (heavy rocks that holds the soil in place but allows the water to seep out) is one possible stabilization practice (Fig. 10). This practice has been used by the highway department in the Minnesota River Basin to stabilize roads along river banks or bridge foundation. However, it will be an expensive practice to implement along various rivers in the Minnesota River Basin. Furthermore, some banks are 200 feet tall and this will be difficult to implement on those banks. Another drawback of this practice will be that we will be converting these rivers into canals which might have some downstream ecological impact. One can also implement rip-rap only at the toe of the hillslope to cut down the fluvial erosion. However, observations suggest that much of the sediment load in these rivers is not due to fluvial erosion at the toe but due to mass failure from near the top of the bank (Fig. 4b). This means reduction in sediment load from the use of rip-rap at the toe will be marginal as banks will continue to slough from a variety of mechanisms other than fluvial erosion.

64

Increased ET through other cropping systems like adoption of miscanthus (Miscanthus x giganteus) is a possibility but there is a difficulty of establishing it in Minnesota due to cold climate (Johnson et al., 2013). Furthermore, its adoption in the basin means taking some area out of agriculture production. Hickman et al. (2010) in Illinois showed that evapotranspiration of miscanthus was 954 mm compared to 764 mm for switch grass (Panicum virgatum) and 612 mm for corn. This additional atmospheric loss of water by these plants could be helpful in reducing deep percolation and to river flows but research is need to (1) quantify the extent of additional soil water storage that will be available through the use of high ET grasses, and (2) whether or not this storage availability is in sync with the types and timing of various storms in the area. Since this practice does not impact thawing processes, its impact on bank sloughing and river sediment loads may be marginal. Also, lack of agriculture production on these lands means finding markets for these grasses to make their adaption appealing. One possible use of the perennial grasses may be in biofuel production. Storage of excess precipitation in upland can be another practice to increase evaporation, potentially reduce some river flow and in turn reduce fluvial erosion at the toe slope. However, its impact on sediment loads in river will also be marginal. Also, there is not enough natural storage (due to flatness of the landscape) to hold water from recent types of storms. For example, both the Waseca and St. Claire areas were flooded during the 20-22 September 2016 precipitation of over 9 inches. There is not enough storage to hold these kinds of storm that cause a major movement of soil from river banks in the Minnesota River Basin. Furthermore, holding water in upland (natural or engineered storage) will affect the water table, thus making tile drainage less effective. Holding water in upland can also affect the seepage processes thus enhancing bank sloughing depending upon where the holding ponds are constructed and whether or not they are lined with impermeable material. This practice will also be expensive.

SUMMARY AND CONCLUSIONS

Much of the sediments in the Minnesota River and its tributaries are a result of natural processes that have been going on for centuries. They are mainly the result of Lake Agassiz creating the Minnesota valley which led to the development of river banks as a result of base erosion in smaller streams trying to reach the Minnesota River level. This process will continue because the knick points in these rivers and ravines are still moving upstream leading to development of more and taller banks. Considering land use and cropping systems are similar across various watersheds in the Minnesota River Basin but sediment yield is less in western (drier) watersheds than the southern and eastern (wetter) watersheds (Fig. 1) suggests that these differences are a result of a combination of varying precipitation regimes and the landscape features (river bank heights) among various HUC 008 watersheds and not due to differences in land use or soil and crop management practices. This in turn would suggest that changes in land use or management practices (adaption of similar ET or lower ET crops, tillage practices) will have minimal impact on sediment loads in the rivers. Potentially, use of rip-rap along streams can stabilize banks but this will be an expensive practice and may have downstream ecological impacts. Similarly, adoption of some high ET grasses in the landscape is another potential possibility but further work is needed in

65 terms of their establishment as well as the extent of additional soil water storage that may be available through the use of these grasses and whether or not that storage is in sync with present day rain storms. Also, efforts can be made to store water on land but there is limited storage under current landscape conditions. Engineered storage is another option but again its effects will be marginal, will be costly and unlikely to fully mitigate the effects of larger storms (e.g., > 9 inch storm events) that have been experienced in southern Minnesota in recent years. Also, upland storage (natural or engineered) may exasperate the water table problems with side effects on tile drainage and even seepage along river banks.

ACKNOWLEDGEMENT

This paper is dedicated to late Doug Miller of the NRCS who graciously spent time with the senior author and his students to explore and explain the landscape features of the Minnesota River Basin. The authors also acknowledge Dr. David Thoma’s pioneer work on the use of LiDAR to characterize bank erosion in the Minnesota River Basin. The senior author thanks his other students Nate Bartholomew, Dr. Andry Ranaivoson, Dr. Holly Dolliver, Heather Johnson, Ashley Grundtner, Melinda Brown, Kari Wolf and Nathaniel Baeumler for their many contributions to understanding the landscape, the causes of increased stream flow, base flow and associated sediment and nutrient loads in the Minnesota River and its tributaries. Research on bank erosion and surface storage using LiDAR was partially supported with grants from the Minnesota Corn and Soybean Research and Promotion Councils.

REFERENCES

Ahlgren, D.E. and M.C, Beeler. 1996. A History of Prescott, Wisconsin: A river city and farming community on the St. Croix and Mississippi. Prescott Historical Society, Prescott, WI. pp605 Belmont, P., K. B. Gran, S. P. Schottler, P. R. Wilcock, S. S. Day, C. Jennings, J. W. Lauer, E. Viparelli, J. K. Willlenbring, D. R. Engstrom, and G. Parker. 2011. Large shift in source of fine sediment in the Upper Mississippi River. Environ. Sci. Technol., 45: 8804-8810. DOI: 10.1021/es2019109 Belmont, P. and E. Foufoula-Georgiou. 2017. Commentary: Solving water quality problems in agricultural landscapes: New approaches for these nonlinear, multiprocess, multiscale systems. Water Resou. Res. 53: doi:10.1002/2017WR020839. Bond, J.W. 1857. Minnesota and Its Resources. Keen and Lee, Chicago, IL. pp 412. Dole, R.B., and F.F. Wesbrook. 1907. The quality of surface water in Minnesota. USGS Water-supply and Irrigation Paper no. 193. Govt. Printing Office, Washington, DC. pp171. Featherstonhaugh, G. W. 1847. A Canoe Voyage up the Minnay Sotor, vol. I, Richard Bentley Publ., London, U. K. Gran, K., P. Belmont, S. Day, C. Jennings, A. Johnson, L. Perg, and P. Wilcock. 2009. Geomorphic evolution of the Le Sueur River, Minnesota, USA and implications for current sediment loading. The Geological Society of America Special Paper 451:1- 12.

66

Gupta, S.C. and U.B. Singh. 1996. A review of non-point source pollution models: Implications for the Minnesota River Basin. A report submitted to the Minnesota Dep. of Agri., St. Paul, MN. pp 77. [Available at: https://drive.google.com/file/d/0Bwnz27fnQiZLcnBVOEJYbmJCaGM/view?usp =sharing, last accessed 27 June 2016] Gupta, S. C., A.C. Kessler, and H. A. S. Dolliver. 2011. Natural vs. anthropogenic factors affecting sediment production and transport from the Minnesota River Basin to Lake Pepin, report, 68 pp., Minn. Corn and Soybean Growers Assoc. St. Paul, Minn. [Available at: https://drive.google.com/file/d/0Bwnz27fnQiZLenJWYmFpSDUzM2M/view?usp 5sharing, last accessed 24 March 2016.] Gupta, S.C., A.C. Kessler, M.K. Brown, and F. Zvomuya. 2015. Climate and agricultural land use change impacts on streamflow in the upper midwestern United States, Water Resour. Res., 51(7), 5301–5317, doi:10.1002/2015WR017323. Gupta, S.C., A.C. Kessler, M.K. Brown, and W.M. Schuh. 2016a. Reply to comment by Dingbao Wang on “Climate and agricultural land use change impacts on streamflow in the upper Midwestern United States.” Water Resour. Res. 52 (5): 4195-4198, doi:10.1002/2015WR018469. Gupta, S.C., A.C. Kessler, M.K. Brown, and W.M. Schuh. 2016b. Reply to comments by Keith E. Schilling on “Climate and agricultural land use change impacts on streamflow in the upper Midwestern United States.” Water Resour. Res. 52 (7): 5697-5700, doi:10.1002/2016WR018656. Gupta, S.C., A.C. Kessler, M.K. Brown, and W.M. Schuh. 2016c. Reply to comments by Schottler et al. on “Climate and agricultural land use change impacts on streamflow in the upper Midwestern United States.” Water Resour. Res. 52 (8): 6699-6705, doi:10.1002/2015WR018497. Gupta, S.C., A.C. Kessler, M.K. Brown, and W.M. Schuh. 2016d. Reply to comments by Belmont et al. on “Climate and agricultural land use change impacts on streamflow in the upper Midwestern United States.” Water Resour. Res. 52 (9) 7529-7535, doi:10.1002/2015WR018476. Gupta, S.C., A.C. Kessler, M.K. Brown, and W.M. Schuh. Reply to comments by Foufoula-Georgiou et al. on “Climate and agricultural land use change impacts on streamflow in the upper Midwestern United States.” Water Resour. Res. 52(9): 7540-7544, doi:10.1002/2015WR018494. Hickman, G.C., A.Vanloocke, F.G. Dohleman, and C.J. Bernacchi. 2010. A comparison of canopy evapotranspiration for maize and two perennial grasses identified as potential bioenergy crops. Global Change Biology Bioenergy 2: 157-168. Johnson, G.A., D.L. Wyse, and C.C. Sheaffer. 2013. Yield of perennial herbaceous and woody biomass crops over time across three locations. Biomass and Bioenergy 58: 267-274. Jones, Evan. 1962. The Minnesota: The forgotten River. Holt, Rinehart and Winston, New York, NY. 306pp. Keating, W.H. 1824. Narrative of an Expedition to the Source of St. Peter’s River, Lake Winnepeek, Lake of the Woods. H.C. Carey and I. Lea, Chestnut St., Philadelphia, PA.

67

Kelly, D.W., and E.A. Nater. 2000. Historical sediment flux from three watersheds into Lake Pepin, Minnesota, USA. J. Environ. Qual. 2000 29: 561-568. Kessler, A.C., S. C. Gupta, H.A.S. Dolliver, and D. P. Thoma. 2012. Lidar quantification of bank erosion in Blue Earth County, Minnesota. J. Envion. Qual. 41: 197-207. Kessler, A. C., and S. C. Gupta. 2015. Drainage impacts on surficial water retention capacity of a prairie pothole watershed, J. Am. Water Resour. Assoc., doi:10.1111/jawr.12288. Melillo, J. M., T.C. Richmond, and G.W. Yohe. 2014, Climate change impacts in the United States: The third national climate assessment, report, U.S. Global Change Res. Program, 841 pp., US Government Printing Press, Washington, D. C., doi:10.7930/J0Z31WJ2. Musser, K., S. Kudelka, R. Moore and Project Partners. 2009. Minnesota River Basin trends, Water Resources Center, Minnesota State University, Mankato, MN. (Available at: http://mrbdc.mnsu.edu/minnesota-river-basin-trends-report, last accessed 24 March 2016) Payne, G.A. 1994. Sources and transport of sediment, nutrients, and oxygen-demanding substances in the Minnesota River Basin, 1989-92. USGS Water Resources Investigations Report 93-4232. Riggs, M.L. 1996. A Small Bit of Bread and Butter, Letters from the Dakota Territory, 1832-1869.Ash Grove Press, South Deerfield, MA. Schottler, S.P., D.R. Engstrom, and D. Blumentritt. 2010. Fingerprinting sources of sediment in large agricultural river systems. Final report. St. Croix Watershed Research Station: Science Museum of Minnesota. Thoma, D.P., S. C. Gupta, M. E. Bauer, and C. E. Kirchoff. 2005. Airborne laser scanning for riverbank erosion assessment. Remote Sensing of Environ. 95: 493-501. Thorleifson, L.H. 1996. “Review of Lake Agassiz History”, Sedimentology, Geomorphology, and History of Central Lake Agassiz Basin, Geological Association of Canada Field trip Guidebook for GAC/MAC Joint Annual Meeting, pp 55-84.

68