Special thanks to EWG staff members Audrey Lothspeich, GIS Analyst, and Sarah Porter, Senior GIS Analyst, and Sarah Carlson of Practical Farmers of Iowa, who all made critically important contributions to this project.
METHODS AND ANALYSIS
To capture yearly cover crop emergence, EWG used the fall and spring to account for one cover crop year. Typically, fall cover crops go dormant during cold temperatures and snow cover and reemerge the following spring, and some cover crops are planted only in the spring. This is why we categorize both spring and fall cover crop detections as one “cover crop year.”
In November 2017 and April 2018, EWG collected agricultural cover type data for more than 700 fields in the three states via a visual survey. (Figure 1.). Analysts from EWG drove around central Iowa in the fall of 2017 and from Fort Wayne, Indiana, to the north central border of Iowa and Minnesota in the spring of 2018. Throughout both trips, analysts collected field locations and various cover type characteristics associated with the site.
EWG visual survey, collecting 740 field locations during Fall 2017 and Spring of 2018.
These data were paramount for spatially differentiating between various cover types using the Normalized Difference Vegetation Index, or NDVI, to create field mean values throughout the fall and spring. The satellite sensor data used to create the NDVI surfaces were taken from Landsat8 (30 meter resolution) and Sentinel2 (10 meter resolution). To make the two datasets more comparable, the Sentinel2 data (RED and NIR bands) were calibrated to Landsat8 data using overlapping imagery from the same day. All Sentinel2 detections were also rescaled to match Landsat8’s and the USDA Cropland Data Layer’s 30 meter resolution. In total, EWG assessed more than 110 Landsat8 scenes and 373 Sentinel2 for the fall and spring.
The satellite-derived NDVI thresholds for individual fall and spring imagery were used to develop a temporal signature for the cover crops and other various cover types, like tillage and no-till. The signature was then used to detect some cover crops within the three-state region.
Monthly NDVI averages from field mean samples of known cover types in Iowa, Indiana and Illinois
After satellite pixels were identified as cover crops from individual scenes, all overlapping scenes were patched together andfiltered to extract only contiguous areas of 10 acres or more. The 2008 Farm Service Agency’s geospatial Common Land Unit data, a high-resolution digitization of field boundaries, were then used to reduce for edge of field pixels and fill areas of spotty emergence. Those same features were then screened to remove elongated and linear shapes indicative of fence rows and grass waterways with metrics resulting from the feature’s area and perimeter. The resulting features were finally separated to show only areas where cover crops were planted atop corn or soybeans using the CDL. The results were then summarized by county and HUC12 watershed.
Ground control collected in the fall had a higher precision than samples collected in the spring (80 percent vs. 60 percent). In an effort to test the alignment of the NDVI signature, EWG crossed our visual survey results with the final outputs and found 15 false positives out of the 740 total samples. Thirteen of the 15 false positives were documented as winter wheat, a cash crop that is often used as a cover crop and that maintains an NDVI signature similar to cover crops. This makes it almost impossible to discriminate between the two cover types. Data from the 2018 CDL show that double crop and standalone winter wheat make up 2 percent of all Illinois total cropland, 1.9 percent of Indiana total cropland and an insignificant amount in Iowa.