Water quality issues are very important in the State of Iowa. For the last decade, research has been focused on meeting priorities set in the Iowa Nutrient Reduction Strategy. Cover crops have been identified as one potential solution for nutrient management. Cover crops are commonly planted in late summer or early fall around harvest time and terminated in the spring. The goal in planting cover crops is to protect exposed cropland from wind and water erosion during the months when cash crops are not growing.
There is a need by entities in Iowa, including current IowaView partners, to have recent and reliable cover crop data. The goal of this project is to create a repeatable and shareable process for detecting cover crops using publicly available imagery and data products. The first year of our research has focused on understanding the behavior of cover crops in Iowa, determining a methodology for distinguishing cover crops on the landscape, and developing a model for repetition.
Phenological cycles vary by region. Normalized Difference Vegetation Index (NDVI) is a method used to observe and quantify phenological difference on the landscape. With this project we wanted to create a more detailed NDVI range based on phenological evidence of the state. To help with this calibration, specific data from rye cover crop research fields (cover crop planting, biomass sample dates, biomass weight, and date of cash crop planting) were used to better understand NDVI values.
Another piece of this year’s research focused on using different techniques to visualize the landscape change over time. Initially, researchers reviewed images from September to June of a growing season, then late March through early June. For the test site in south central Iowa, the most critical image dates were between mid-April and late May, the time between peak of the cover crop green up and cover crop termination before the seeding of the cash crop. It is best to have images with minimal cloud cover; however, spring in Iowa can be unpredictable which did pose a challenge to this method.
IowaView staff have looked at several different methods for visualizing the cover crop cycle over time. Two examples shown: A) calculating difference in NDVI over two spring dates and B) using color band channels (RGB) to show different image dates over time. Additionally, to process images faster, a model was created that takes 4 bands of an image and then creates a file geodatabase from which it creates a Normalized Difference Vegetation Index (NDVI) image as well as using field boundaries (provided by the user) to create a statistics table including the average pixel value within the field boundary. Field boundaries were created using data available from the USDA and CropScape. Imagery inputs were from Landsat 8 and Sentinel 2, depending on which had an acceptable, cloud-free date of interest.
We will refine the prediction process using three input dates to give a more accurate portrait of cover crop fields on the landscape. In addition, we will continue to produce cover crop existence data in additional watersheds across the state.