Agricultural productivity and water quality tradeoffs of winter cover crops at a landscape scale through the lens of remote sensing

Abstract

A large body of scientific research studying the impacts of cover crops on crop-water-nutrient dynamics has been based on short-term and/or controlled experimental trials without much consideration of the spatial and temporal heterogeneity that typically exists in the working landscape. As such, those findings are difficult to reproduce on working farms. This study is focused on quantifying the long-term impacts of cover crops on both cash crop (corn and soybean) yields and water quality across the Maumee River watershed (MRW) using publicly accessible historic satellite data, from 2009 to 2019, that leverage spatial and temporal variabilities in the agricultural landscape. Corn and soybean yield maps at 30-m resolution were prepared using regression models that integrated satellite images, weather, and county-scale crop yield data. These yield maps were overlaid with 30-m resolution cover crop maps, which were created for the same period using a machine learning classification model using satellite data and ground-truth observations, to quantify yield differences between corn and soybean fields with and without cover crops. The annual variability in winter cover crop area was compared with publicly accessible data on spring nitrate (NO3–N), total phosphorus (TP), and soluble reactive phosphorus (SRP) loads and concentrations, that were collected at multiple water quality monitoring stations within the MRW. During the study period, cover crops, typically showed minimal to negative yield impacts on cash crops, with yield reductions of up to 1.35 ton/ha for corn and 0.24 ton/ha for soybean. However, crop yield variabilities were found to decline with a longer cover crop history. The winter cover crop area within the MRW was found to have water quality impact, with a strong negative correlation with the spring NO3–N concentration (correlation coefficient (R) = −0.70) and loads (R = −0.65) at the MRW outlet. There was a mixed relationship between the cover crop area and P load and concentration, with a positive correlation (R = 0.17) with spring TP concentration and a negative one with spring TP load (R = −0.29), spring SRP concentration (R = −0.30), and spring SRP load (R = −0.40). Similar relationships were also observed across sub-watersheds within the MRW, suggesting that cover crops can be more effective in reducing NO3–N losses compared to P. These findings highlight the importance of using satellite-based remote sensing to improve the understanding of cover crops and their impacts on crop productivity and water quality at a broader scale.