Cover crop impacts on soil organic matter dynamics and its quantification using UAV and proximal sensing

Abstract

Soil health is a critical aspect of sustainable agriculture, with soil organic matter (SOM) serving as a key indicator. In arable fields, growing cover crops has been advocated as a prime practice to promote SOM accumulation. However, the effectiveness of cover crops to promote SOM accumulation can vary widely. Furthermore, accurate quantification of SOM at field scale is severely constrained by the labour intensity and destructive nature of traditional methods, which limits the ability to quantify and monitor cover crop impacts on SOM. We tested whether cover crop mixtures promote SOM accumulation more than cover crop monocultures in a 6-year field experiment with arable crop rotation on sandy soil. We found that the cover crops radish and oat-radish mixture significantly increased SOM levels compared to the fallow treatment. Next, on soil sampled in year 4, we explored the use of proximal (VIS-NIR, MIR) and remote sensing using Unmanned Aerial Vehicles (UAVs) to upscale SOM from wet lab-based point samples to the whole field and map its SOM status. Thereto, we used Random Forest (RF), Support Vector Regression (SVR), and Partial Least Squares (PLS) models and found that the best fitting model depended on the type of spectral sensor. With proximal sensing (MIR) the best SOM prediction was achieved using SVR (R²= 0.84, RMSE= 1.55 g/kg SOM). For UAV imagery with hyperspectral camera the best model was RF (R² = 0.69, RMSE= 2.19 g/kg SOM) and enabled digital mapping of SOM distribution across the field. The accuracy of MIR enabled identifying radish cover crop treatments as having on average higher SOM levels compared to the fallow. However, infield spatial SOM variation can override cover crop effects on SOM levels. Therefore, UAV time series are required to remotely quantify cover crop impacts on SOM changes. Overall, our results show potential for combining proximal and UAV-based sensing SOM as a tool for more efficient and accurate spatiotemporal monitoring of SOM at field scale, which can aid in promoting sustainable agricultural practices that enhance soil health.