A novel machine learning-driven approach for predicting nitrous oxide flux in precision managed agricultural systems

N2O, also known as nitrous oxide, is a greenhouse gas that is roughly 300 times more potent than CO2 and destroys the the stratospheric ozone layer causing climate change. One of the primary causes of the rapid increase of N2O in our ecosystem is the application of nitrogen fertilizer to agricultural land. This stimulates N2O emissions and accounts for approximately 5% of the global greenhouse gases, forcing harm to the environment and atmosphere (Aronson and Allison, 2012). Previous models severely underestimated N2O flux in various crops, causing inaccurate predictions to form. In our study, we utilized data from automated flux chambers to train and evaluate different machine-learning models to predict the field-level flux of N2O which assist farmers to predict fertilizer amounts to use. The best machine learning model, Random Forest, performed considerably better than the standard empirical and biophysical models by roughly 15%, and show promise in improving predictive accuracy and guiding sustainable agricultural practices.