Remote Sensing of Soil Health Management

Project outputs

Daily Erosion Project - See sub-watershed-scale estimates of precipitation and soil loss for one day or a period of your choice.

Maps of crop residue - Find the average percentage residue cover across cultivated land at planting time. 

Maps of cover crop emergence - See what percentage of row cropped land has cover crops growing in the late fall. 

The three products above are designed to support natural resource conservation research and outreach. Uses include:

  • Communication and education: Find local stories in the data to share with the public and decision-makers.
  • Water resource assessment: Assess local or regional trends in erosion to target critical pollution source areas. Combine land management information with monitoring data to help identify drivers of water quality patterns. Communicate with planners and decision-makers about the geography of resource concerns.
  • Conservation planning: Identify local patterns of conservation practice adoption to locate opportunities to expand adoption. Identify areas to target for promoting conservation adoption.
  • Modeling: Use the data to improve the land cover information in water quality modeling efforts.

About the project

The Board of Water and Soil Resources was allocated Clean Water Fund dollars to generate daily estimates of wind and water erosion in Minnesota and to use satellite imagery to map crop residue levels and cover crop use. To complete the work, BWSR contracted with a University of Minnesota research team led by Dr. David Mulla. MOSH is helping share project results by hosting this web page and other project products. Learn more about the project on the BWSR project page and in the Interim report (pdf).

*D. Mulla, B. Gelder, L. Olmanson, B. Dalzell, D. Wheeler, J. Nelson, J. Galzki. 2020. Assessing Soil Residue Cover, Cover Crops and Erosion using Remote Sensing and Modeling: A final report to the Minnesota Board of Water and Soil Resources.

Project personnel

  • David Mulla (University of Minnesota)
  • Brian Gelder (Iowa State University)
  • Leif Olmanson (University of Minnesota)
  • Grace Wilson (University of Minnesota)
  • Dan Wheeler (University of Minnesota)
  • Joel Nelson (University of Minnesota)
  • Jake Galzki (University of Minnesota)
  • Ann Marcelle Lewandowski (University of Minnesota)
  • Matt Drewitz (Project Manager, Board of Water and Soil Resources)

Funding

The project is administered by the Minnesota Board of Water and Soil Resources with funding provided by the Clean Water Fund through the Land and Legacy Amendment.
 

Crop Residue at Planting

These maps indicate how much residue is on the soil at planting time. From this, users can infer the type of tillage being used.

How were the maps made?

The Minnesota remote sensing project team used Sentinel-2 and Landsat-8 satellite imagery to estimate the percentage of soil covered by residue in each 20 or 30 m pixel, respectively, that falls within cultivated agricultural fields. This is accomplished by developing residue models using the relationship between field measure residue and the satellite imagery. The model is applied to all cultivated agricultural pixels identified using the CDL. Then the data is summarized in two ways: 1) by averaging all the cultivated pixels for the watershed or county, and 2) by determining the proportion of cultivated pixels in the watershed or county that were over 30% residue coverage. Land that is not identified as cultivated such as forested, developed, water, wetlands or forage are excluded when calculating estimates.

Landsat 8 and Sentinel 2 satellite images were used in tandem to assess soil residue cover for 67 Minnesota agricultural counties. Landsat 8 provides multi-spectral imagery with a 30m spatial resolution and a 16 day revisit frequency, and Sentinel 2 provides multi-spectral imagery with a 20m spatial resolution and a revisit of 5 days. Estimates are based on clear available imagery that have a date as close as possible to the window after planting and before crops have emerged.

Only the 67 agricultural counties in Minnesota are mapped. Additional gaps in the data occur where cloud-free satellite imagery is not available in the right window of time.

Data accuracy. The accuracy of the residue estimates are determined using field measured validation data. Field measurements are collected in a uniform method involving taking pictures looking straight down at the field surface, from about 5 feet above the surface. The images are then analyzed to determine percentage of soil coverage and compared to satellite imagery results.
To some extent, errors cancel out at larger geographic scales so estimates for counties and larger watersheds are better than for small areas. For the sake of individual privacy and to avoid consequences of any field-level errors, data is not released at the field scale.

See the interim report  for more detail about the mapping and field work:
D. Mulla, B. Gelder, L. Olmanson, B. Dalzell, D. Wheeler, J. Nelson, J. Galzki. 2020. Assessing Soil Residue Cover, Cover Crops and Erosion using Remote Sensing and Modeling: A final report to the Minnesota Board of Water and Soil Resources. http://bwsr.state.mn.us/

Static maps

These maps show the average proportion of soil covered by residue for each major watershed (HUC-8), sub-watershed (HUC-12), and county. Only cultivated acreage is considered in the estimate. Tillage type can be inferred from the amount of residue.

Geodatabase layers

Download maps from the Minnesota Data Commons [link will be added]

Cover Crop Emergence Maps

How were the maps made?

The Minnesota remote sensing project team [link to project page above] used satellite imagery to identify cover crop growth on row cropped land, and then summarized the percentage of ag land with cover crops for each watershed and county.

Landsat 8 and Sentinel 2 satellite images were used in tandem to assess greenness for 67 Minnesota agricultural counties. Landsat 8 provides multi-spectral imagery with a 30m spatial resolution and a 16 day revisit frequency, and Sentinel 2 provides multi-spectral imagery with a 20m spatial resolution and a revisit time of 5 days. Estimates are based on clear imagery that allow for enough time after harvest and cover crop planting for the cover crops to emerge but before snowfall. The Cropland Data Layer is used to identify row crop areas while removing areas that might be green but not cover crops, including  alfalfa, hay, and forage, along with non-agricultural areas such as forest, grassland, and developed land. The remaining green vegetation is assumed to be a cover crop.

Only the 67 agricultural counties in Minnesota are mapped. Additional gaps in the data occur where cloud-free satellite imagery was not available during the post-harvest window.

A sampling of fields were checked in-person to determine the accuracy of the process.

The University of Minnesota research team led by Dr. David Mulla continues to improve their method for interpreting satellite imagery, so results may improve slightly in the future.

See the interim report (pdf) for more detail about the maps and the field checking:
D. Mulla, B. Gelder, L. Olmanson, B. Dalzell, D. Wheeler, J. Nelson, J. Galzki. 2020. Assessing Soil Residue Cover, Cover Crops and Erosion using Remote Sensing and Modeling: A final report to the Minnesota Board of Water and Soil Resources. http://bwsr.state.mn.us/

Static maps

Cover crop emergence fall 2019 (png file)

Cover crop emergence fall 2020 (ppt file)

Geodatabase layers

Download maps from the Minnesota Data Commons [link will be added by the end of 2021]