Optimizing Agricultural Use of Diverse Soil Landscapes: Small Organic Vegetable Farms in the Driftless Area

Progress report for LNC22-474

Project Type: Research and Education
Funds awarded in 2022: $249,939.00
Projected End Date: 10/31/2025
Grant Recipient: UNIV. OF MINNESOTA
Region: North Central
State: Minnesota
Project Coordinator:
Dr. Kyungsoo Yoo
Expand All

Project Information



The Driftless Area covers 24,000 square miles at the junction of four North Central States: Minnesota, Wisconsin, Iowa, and Illinois. The name indicates that the region, in comparison to the surrounding areas, lacks glacial deposits. Even though the region's glacial history is more complex, the name captures the region's strong physiographic identity. Its complex topography does not allow mechanized industrial farming, which is overwhelmingly present in the North Central states. 

Farmers contributing to this proposal were eager to list topography-affected soil problems: (i) extremely diverse soil texture, moisture, and organic matter contents that limit timely field preparation, adopting ideal rotation, and fertilizer application, (ii) flooding and saturation of low-lying fields, (iii) distinct micro-climate between the ridge tops and valleys, and (iv) rapid soil erosion across the range of slopes (see collaboration letters). As the farmers’ responses highlight, it is within the individual farms that the highly variable soil conditions occur in this unique region. However, most research on soils and soil erosion focus on watershed scales. Consequently, we lack a practical guideline for  small-scale farmers to optimize the sustainability and productivity of their lands according to their complex soils and topography. This issue is urgent as the Driftless Area is often seen as a magnet for small-scale farmers.


This research project will:

  1. Collaborate with at least six organic vegetable farms in the Driftless Area, as organic vegetable production is one of the most common agricultural land-uses in the region. Via focus group and individual farmer interviews, we will learn how soil-topographical relationships are factored into the land-use decisions the farmers make.
  2. Characterize how topography affects land uses, vegetable productivity, and soil health within individual farms owned by our collaborator farmers. We will collect soil properties data and use remote sensing and high-resolution elevation data. Remote sensing-based characterization of vegetable health and yield will be made as a function of topography.
  3. Integrate farmer narratives and biophysical data in order to (1) document novel economic and environmental effectiveness of within-farm farmer soil-landscape use and (2) co-develop alternative models for agricultural use of complex soil-landscapes in part by applying precision agriculture principles and technologies in organic agriculture.

Relevance: All soil parameters and the agricultural impacts on soil erosion and hydrology strongly reflect topography that varies within the scales of small farms in the Driftless Area. Assessing and improving these connections are important for organic farm productivity and sustainability.

Project Objectives:

Our objectives are to: (1) better understand within-farm agricultural use of soil landscapes, and (2) to support positive land-use decisions, or propose improved or alternative within-farm agricultural use of soil landscapes and management practices by combining precision agriculture and organic agriculture.

Our expected outcomes are (1) farmers have increased knowledge and awareness about the role of topography on the productivity and sustainability in their farms and (b) farmers have plans to change or do something differently based on our research disseminated through our outreach.


In the first year, our research team focused on (1) understanding and documenting topography-related issues and (2) characterizing the topography and soils of participating organic vegetable farms.

In this report, I have included the hypotheses specific to the Year 1 activities. These hypotheses were tested through a mix of activities that included farmer focus groups, farm visits, GIS analysis, and soil sampling. These activities are described in the Research and Project Activities sections of this report.


Click linked name(s) to expand/collapse or show everyone's info
  • Dr. Anna Cates



In the first year, our research team focused on (1) understanding and documenting topography-related issues and (2) characterizing the topography and soils of participating organic vegetable farms. Below are the active hypotheses we tested to address these research objectives in Year 1.

  • Within the cropland of organic vegetable farms in the Driftless Area, the spatial distribution of cropland is largely determined by topography.
  • There are two types of organic vegetable farms in terms of topography. The first type of farm selects and cultivates only relatively flat areas of ridge or valley bottoms. The second type of farm - without access to relatively flat areas - cultivates hillsides.
  • In the first type of farm, ridge soils are more spatially and temporally variable.
  • In between the two types, the farms that farm the hillsides face greater challenges of soil variability.
Materials and methods:

In the first year, we worked to test the hypotheses using four approaches. The first was focus group meetings with farmers, the second was farm visits, the third was geospatial analysis, and the fourth was soil sampling and laboratory analysis.

  1. Focus group meeting: We met with the owners and managers of the four participating organic vegetable farms in the Driftless Area. During this meeting, the farmers described the topography-related issues and challenges specific to their farms. The list generated from the meeting served as our checklist during the farm visits. I summarized the information from the meeting in the project activities.
  2. Farm visits: We visited the four participating organic vegetable farms. The visits were informative and confirmed many of the issues that the farmers shared with us during the focus group meeting. The visits revealed the different ways in which the farmers use or cope with the topography. We also learned that there are drastic differences between the large organic vegetable farms and the small ones in the way their fields are tied to different topographic zones. I have summarized the field visits in the project activities.
  3. GIS Analyses
    • Google Earth: We obtained the fields and farm boundaries in Google Earth format from four farmers who participated in the focus group (2/1/2023).
    • Topographic analyses: From the LIDAR data, we constructed a 3-meter DEM for the four farms. From the DEM, topographic variables such as slope, aspect, topographic position index, and topographic ruggedness index were calculated. We calculated their mean values and deviations per individual field by superimposing the field and farm boundaries on the topographic data. During the first year, method development was focused on the Featherstone farm in Minnesota.
    • Soil Survey: Web soil survey data were extracted for the four farms that participated in the focus group.
    • Based on the Planetscope 3-meter satellite imagery and farm visits, we began creating hypothetical management zone boundaries within which topography and soil conditions are homogeneous.
    • We collected crop data from PlanetSCOPE (3m resolution, 8 bands with near infrared (885-845nm), red edge (697-713nm), red (650-680nm), green (547-583nm), green I (513-549nm), yellow (600-620nm), blue (465-515nm), coastal blue (431-452nm) for the two periods: 2023 May to September and 2022 May to NOV for all farms.
    • We collected bare ground satellite images from Planetscope for 2023 Apr, March, 2022 Apr, March, Nov or Dec and 2021 Apr, March, Nov or Dec and 2020 Apr, March, Nov or Dec.
    • We also collected and began organizing soil survey data per field and per farm.
  1. Soil Sampling: 
    • Based on the preliminary GIS analysis, we conducted intensive soil sampling in two topographically contrasting areas on the farm B.
    • These samples will be analyzed for organic carbon and pH

------------------------------------------------------------------Below is the text from the original proposal ---------------------------------------------------

  • Farmer/Rancher Involvement

The need and research questions for this project originated from conversations with farmers with whom we have interacted at regional farming conferences such as MOSES, as well as past research collaborations between the Grossman lab and Featherstone Farm, one of our project partners.  Our goal is to identify six total farms where we will conduct our research, and at least eight farmers who are willing to participate in focus groups. We have been met with enthusiastic support for our project, confirming four of the desired six sampling sites (farms). These farmers provided us with valuable information on which to base our project objectives via interviews conducted during this full proposal preparation stage, highlighting their specific challenges and needs that can potentially be addressed via this project. Letters of collaboration highlight the ranging needs of these farmers, and the overlap among the challenges they are facing we believe speaks to the wide-ranging experiences of many farmers in the Driftless. Farmers also provided us with preferred outreach and educational strategies in order to best disseminate project findings in appropriate ways, also articulated in this proposal. The four farmers providing letters of support are excited to host us on their farms to allow detailed soil sampling needed to provide project data, and also to participate in a focus group. If funded, additional farmers will be recruited. We believe it will not be difficult to identify these farmers based on farmers’ initial responses. All farmers will be compensated at rates commensurate with their degree of participation (see budget justification).

Upon project funding, we will host a focus group to further define similarities among farming experiences. In addition to the researchers from UMN, our collaborator, Natalie Hoidal (Extension educator who has expertise in local foods & vegetable production), will also participate with an interest in helping farmers come up with a soil management plan for their farms. The focus group will be conducted via Zoom (farmer preference to avoid travel-time among farms).  If any farmer does not want to be recorded, we will instead take detailed notes by hand. From this focus group, we expect to learn how specific farms compare broadly in the Driftless Area. 

We will then implement the project through five stages: (1) GIS data collection from organic vegetable farms (N=6) in the Driftless Area, (2) Soil sampling and sensor installation, ground sensing, UAV (Unmanned Aerial Vehicle) and satellite remote sensing data collection, (3) assessment of current agricultural use of soil landscapes in terms of agricultural yields and soil erosion, (4) development of improved or alternative organic vegetable management practices by considering soil-landscape conditions  for small farms, (5) Outreach to broader small farm communities in the Driftless Area.

At project completion, we will present to the farmers our assessment of their crop productivity as affected by soil-landscape factors, which includes detailed GIS and soil property data. We plan to collaboratively develop recommendations for improved or alternative management practices by considering soil-landscape conditions that are in accordance with the farmers' aspirations. At each stage, we will have different research questions (Q) and methods (M).

Stage 1

Q: How do within-farm land use and management (eg., crop selection and distribution, tillage timing and frequency, rotation, irrigation) reflect topography (eg., aspect, slope gradient, curvature, and length, groundwater table depth, or distance to streams or water sources) and soil properties (eg., temperature, texture, and erosion).

M: Farm visits and informal interviews with farmers and farmer focus groups. Geospatial analyses of remote sensing images, LIDAR (Light Detection and Ranging), digital soil survey, and surficial geology.

Stage 2

Q: How do crop  choices and yields reflect topography and soil properties?

M: Farm visits with farmers and farmer focus groups. Using proximal sensing, UAV, and satellite remote sensing to estimate soil conditions and crop choices and yields.  Soil sampling and analyses and soil temperature monitoring.

Stage 3

Q: How do soil erosion and sediment redistribution reflect within-farm agricultural land uses and topography?

M: Farm visits with farmers and farmer focus groups. Assessing soil 137Cs and 210Pb radioisotopes to quantify the timing and intensity of soil erosion and deposition.


Stage 4

Q: What are the missed opportunities to maximize the economic and environmental benefits when considering soil-landscape conditions for vegetation health? What are the practices that the farmers are doing right and thus should be promoted?

M: Farm visits and informal interviews.   Analyses and comparisons across GIS data layers, proximal and remote sensing data, and soil-landscape properties.

Stage 5

Q: What can be done to remedy the missed opportunities or support current land-use and management decisions that result in high yield or reduced soil losses in steep regions?

M: Farm visits and informal interviews with farmers and farmer focus groups. Outreach to farmers in the Driftless Area in collaboration with NGOs and the University Extension. Listening sessions with the regional farmers. Developing improved management practices by applying precision agricultural technologies in organic farming systems.


  • Ground- and Lab-based Soil Analyses

Research has abundantly shown that local soil and water conditions are heavily influenced by topography such that crop yields reflect the local soil-topography relationship. These studies, however, are mainly conducted for mono-cropped fields. Researchers have rarely examined multi-cropping multi-landuse situations at field scapes, a situation that individual farmers experience within their farms in the Driftless Area.

To assess the severity of within-farm soil erosion and deposition, we will use gamma spectrometers to quantify the activities of fallout radionuclides  (210Pb or 137Cs). Those are environmental isotopes (we are not adding any chemicals or isotopes to the farms or soils). A well-type germanium crystal gamma detector is available at Yoo's lab. Yoo has an extensive publication record in using radio-nuclides for quantifying soil mixing and erosion rates. 

Among the soil properties, our collaborating farmers singled out soil texture as the one that has far-reaching impacts on their management. They further noted that temporal variability of soil conditions differs substantially within their farms, which makes it challenging to conduct timely tillage and planting. Thus, in addition to determining soil texture, we will monitor soil temperature. Soil temperature can be more readily and cheaply monitored, while it is an important component of soil micro-climate that is sensitive to topography and hydrology. Soil samples will be subject to basic soil health measures such as total carbon and readily available nitrogen. As the project proceeds, we will secure more funding to expand the soil health analyses to include biological and chemical properties.

Field soil sampling will be designed in consultation with the farmers. When consulting, three primary considerations from our perspective will be shared with the farmers: (a) topographic positions, (b) current production modes and management history, and (c) feasibility of replications and generalization.

Our plan – which will be tailored for individual farms in consultation with the farm owners – is to cover the four combinations of two topographic positions and two agricultural management types at each farm. At each combination, we plan to collect soil samples and install temperature sensors at five randomized replicate soils. This scheme will result in sampling 120 soils: 5 replicates X 4 combinations X 6 farms. At each soil, we will sample the top 0-10 cm For fallout radionuclides, which are most useful when analyzed as a function of soil depth, we will work with two replicates per combination at three select farms (24 soils =  2 replicates X 4 combinations X 3 farms). For those select soils, we will sample the first top 30 cm at 4 depth intervals (0-5 cm, 5-10 cm, 10- 20 cm, 20- 30 cm). This will result in a total of 96 fallout nuclide measurements (24 soils X 4 depths).


  • Geospatial Analyses

We are also interested in characterizing how general land-use types reflect topography and soils within the Driftless Area and how our targeted organic vegetable farms fit into the general trends. For example, two of our collaborating farmers noted that they focus on the ridge tops and valleys while recognizing that these two landscape components provide drastically different farming conditions.

For topography analyses, we will use LiDAR (Light Detection and Ranging) data that allows high resolution (< 1 m) topographic analysis. LiDAR is freely available for the Driftless counties in Minnesota and Wisconsin, and it is likely to become available within a year in Iowa. The current land-use and land-use changes will be examined with the National Land Cover Databases which provides nationwide data on land cover and land cover change at a 30 m resolution with a 16-class legend. The database is updated regularly. We will first seek to understand how these landuse categories reflect topography and then move into analyzing the effects of soils. The gSSURGO, which is derived from Soil Survey Geographic (SSURGO) Database, offers a 10 m cell size raster.

Although the geospatial data of farm boundaries is not publicly available, land parcel data – as an approximation for farm boundaries - is available online at the county level in Minnesota and at the state level in Wisconsin. We will examine how within-parcel land-cover types, topography, and soil are related within select parcels. 


  • Developing Zone-specific Management Practices based on Precision Agriculture Technology

Precision agriculture was originally developed for large-scale mechanized agriculture. However, its concept and principles are applicable in any scale of farming systems. For small-scale organic farms, a practical strategy could be delineating a farm into a few relatively uniform management zones based on crop productivity and key soil-landscape variables. Remote sensing technologies will be used to monitor vegetable growth conditions and identify stresses.

Precision agriculture will allow us to adjust management practices according to the yield potential and soil-landscape conditions in each zone. We will be particularly interested in exploring the types of adjustments that can be implemented with the farmers’ current equipment without needing large machines.  For example, we can adjust the vegetable types, planting densities, tillage types and depths, organic fertilizer types and application rates, and the possible incorporation of cover crops, etc.

We will calculate vegetation indices based on UAV and PlanetScope satellite remote sensing images with a 3-meter resolution (available to UMN researchers without charges) in order to represent vegetable plant growth and productivity as a function of within-farm topography. Machine learning models will be developed to use soil-landscape variables and vegetable type, weather and management information to predict the selected vegetation indices, and key soil-landscape factors will be identified.

Such key soil-landscape factors will be used together with multi-year PlanetScope remote sensing images to delineate a farm into a few management zones with distinct vegetable productivity and soil-landscape conditions. These results will be shared and presented to the farmers. We will consult the farmers in exploring zone-specific management practices that involve adjusting the vegetable types, planting densities, tillage types and depths, organic fertilizer types and application rates, and the possible incorporation of cover crops, etc. We expect that our data on soil-landscape conditions and yield potential in each zone will help us to improve vegetable yield while reducing runoff, soil erosion and nutrient losses.

Research results and discussion:

We focused our GIS analysis on Featherstone Farm in Year 1 to develop, standardize, and test the methods before applying them to all participating farms. Figure 1 shows the areal distribution of field area by elevation. The graph shows a binomial distribution, indicating that the farm's cultivated fields are located on the valley floor or the ridge.

field area by elevation
Figure 1. Areal distribution of cultivated fields by elevation

In Figure 2, we analyzed the areal distribution of individual fields by slope gradient. Most fields, except for the currently unused field, have negligible slope gradients. However, the fields differ considerably in terms of within-field variability in slope gradients. For example, Olson's field, Peterson's farm-currently unused, and Brand's field have much greater variations in slope gradients than other fields.

Areal distribution of fields by slope gradient
Figure 2. Areal distribution of fields by slope gradient

Figure 3 shows the aspect of the fields. Our next step is to convert the aspect of the fields (by 3m X 3m) into solar radiation (energy per area per time). With relatively minimal slope gradients of the fields, the different aspects of the fields may not result in significant differences in solar radiation in this particular farm.

Fields by aspect
Figure 3. Fields by aspect

The central question of this project is how topography and soil are related to providing favorable or challenging growing conditions for farmers. For soils, we analyzed the bare ground images to assess the soil brightness index and relate it to soil moisture and organic carbon content. Preliminary analyses show that Brand's field and Peterson's field have the highest brightness index (Figure 4), indicating high moisture and organic matter content. However, these two also have the highest variability (Figure 5).

Bare ground brightness index
Figure 4. Bare ground brightness index
Bareground brightness idex coefficent of variance
Figure 5. Bareground brightness idex coefficent of variance

Within this farm, these topographic analyses are beginning to quantitatively reveal the within-field variability of topographic characteristics and solar radiation. They also allow us to identify zones where these characteristics differ significantly from the rest of the fields and require different management.

We also worked with NDVI analyses to characterize the within-field variability. Understanding the NDVI of organic vegetable farms at the field scale proved to be extremely challenging, mainly because of the high spatial and temporal variability of the crops grown in the fields.

Figure 6 shows the seasonal trends of NDVI by different fields, which can potentially be related to crop biomass, but with considerable data collection at the ground level. What could be potentially useful is to characterize the differences in NDVI between fields and the within-field coefficients of variance of NDVI (Figure 7).

NDVI by fields

Figure 6. NDVI and its seasonal variation by fields.

Coefficients of variance in NDVI by fields
Figure 7. Coefficients of variance in NDVI by fields

In the light of the above hypotheses, these GIS results show that topography is a critical factor in the management of this farm and that individual fields - depending on their topographic location - exhibit substantially different soil and vegetation characteristics and their spatial and temporal variability.

We also met with the farmers of the four participating organic vegetable farms and visited their farms. The information collected about their fields and farming challenges is described in the project activities. These interviews and field visits confirm that the methods, observations and results developed from Farm B would be applicable to all organic vegetable farms we study in this project, but with important farm- and topography-specific differences.

Based on this progress, we have the following research objectives for the second year:

  •  Integrate topographic, soil, and vegetation GIS data to divide fields into unique management zones.
  •  Apply these methods to all participating farms.
  •  Complete laboratory analyses of soils collected from the Farm D and conduct soil sampling on all participating farms.
  •  Compare GIS data (e.g., NDVI and bare soil brightness) with ground-based data (crop survey, soil temperature, and soil organic matter analyses).
Participation Summary
4 Farmers participating in research


Educational approach:

We do not have a report about educational activities in year 1.

Project Activities

Focus group meeting
Visit to Farm A
Visit to Farm B
Visit to Farm C
Visit to Farm D
Field sampling

Educational & Outreach Activities

Participation Summary:

4 Farmers participated
Education/outreach description:

Through focus group meetings and farm visits, our Year 1 activities focused on identifying farmers' needs and understanding the details of their topographic challenges. An important part of our education and outreach is through these meetings and face-to-face discussions. In addition to these activities, we have also conducted a GIS analysis of individual farms and fields to help address the concerns of the farmers. I have incorporated these activities into research and project activities in this annual report.

We are in the process of writing a journal article on the effects of topography on organic vegetable farms in the Driftless Area. Our findings will also be the basis for a newsletter article in years 2 and 3. In Year 2, we will begin to prepare for the field days and workshops at the meetings.  We have contacted and are talking with Marbleseed about co-sponsoring the Field Day.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture or SARE.