Document Type

Dissertation - Open Access

Award Date


Degree Name

Doctor of Philosophy (PhD)

Department / School

Agronomy, Horticulture, and Plant Science

First Advisor

David E. Clay


Cover crop, Greenhouse gasses, Machine-learning, Meta-analysis, Soil Organic Carbon


People worldwide are challenged by multiple threats including climate change, growing populations, and soil degradation. Addressing these challenges requires understanding of the local environment, farming systems and modern technologies. These technologies include new ways to process information that include artificial intelligence, machine learning and meta-analysis. Models produced using these technologies may be useful for predicting the consequences of implementing conservation practices that reduce GHG emissions as well as for determining the carbon footprint of cropping systems that include environmentally friendly conservation technologies such as growing cover crop. Therefore, our objectives of this study were to: 1) provide an overview of conservation agriculture technology as strategy to minimize soil degradation, climate change challenges, and food insecurity issues in developing countries like Nepal, 2) conduct global meta-analysis to quantify the impact of cover crops as one of conservation agriculture technique, on soil organic carbon (SOC) and crop yield in a corn (Zea mays L.) cropping system and 3) assess different machine learning based algorithms to predict the daily N2O-N and CO2-C emission from a decomposing rye (scientific name of rye) cover crop. For the first objective, historical data analysis indicated that air temperatures in Nepal have been increasing since 1901 at a rate of y 0.016 oC yr-1, whereas precipitation has been decreasing at a rate of -0.137 mm yr-1. Increasing air temperature, when combined with decreasing precipitation, are interacting to reduce crop growth and yield, diminishing Nepal’s food security. We proposed conservation agriculture practices such as planting cover crop as farmer and environment friendly approach to mitigate and adopt the climate change impact and enhance food security. In second objective, I used meta- analysis approach to measure the effect of cover crop on SOC values in corn at a global scale. During the meta-analysis, data from 62 globally published peer reviewed literature showed that cover crops in the corn production system increased SOC by an average of 7.8%. The SOC increased at rates of 0.46 and 0.80 Mg/ha/year at the 0-15 and 0-30 cm soil depths respectively, due to cover crop planting. To meet the third objective, several different machine learning prediction models were tested, which included multiple linear regression (MLR), partial least square regression (PLSR), support vector machine (SVM), random forest (RF), and artificial neural network (ANN), on daily N2ON and CO2-C emission data which were measured from a decomposing cover crop in 2019 and 2020 at Aurora, SD, USA. Each models’ performance was accessed using coefficient of determination (R2) (higher values close to one were deemed ‘best’), root mean square error (RMSE) and mean absolute error (MAE), where lowest values were ‘best’. Out of all models, the RF model accounted for 73% and 85% of the variability explained in N2O-N and CO2-C emissions, respectively. Across the three objectives, we found that new analysis approaches such as machine learning and meta-analysis can be used to determine the carbon footprint and prediction of GHG emission from conservation agriculture practices such as planting cover crops.

Library of Congress Subject Headings

Agricultural conservation -- Technological innovations.
Cropping systems.
Cover crops.
Soils -- Carbon content.
Greenhouse gases.
Machine learning.

Number of Pages



South Dakota State University



Rights Statement

In Copyright