My CSRESF Research Project

Over the past weekend I competed in the Central Sound Regional Engineering and Science fair, where many students in central sound region competed by presenting their research projects, that they’ve been working on all school year, to judges. The project I presented was called: Using Machine Learning and Crop Simulation Models to Find a Sustainable Alternative to Common Industrial Farming Practice.

In this project I aimed to address the unsustainability of modern-day agriculture. Currently, 70% of the freshwater worldwide and 42% of the freshwater in the United States are consumed in agriculture according to the Organization for Economic Cooperation and Development (OECD). 38.6% of ice-free land worldwide is consumed in agriculture according to the National Geographic. Lastly, currently 140 gallons of fossil fuels are required to grow and harvest just one acre of corn according to Cornell University. Ultimately, if farming modern-day farming continuous to consume these nonrenewable resources at this rate, they will be completely depleted in the near future before they can be replenished, which will cause agricultural food production to drop, and this will send the world into a worldwide human crisis.

To help make farming more sustainable I found a more sustainable cropping system, to monocultures for growing corn in the Midwest. Monocultures are a cropping system where only one type of crop is grown on a particular field over a period of time. This cropping system is widely used in the Midwest, USA region especially for growing corn. The disadvantages of it however were that requires a huge input of fertilizer and causes significant land and soil degradation. There has been prior research done into finding sustainable alternative cropping sequences or sustainable management practices for growing corn. However, due to their being a lot of research done into this field, farmers have been presented with an excessive number of options on potential management practices or cropping sequences they can implement to make farming more sustainable. Therefore, in my research I aimed to simplify all this research done by comparing and analyzing multiple cropping sequences and management techniques proposed in order to develop one clear alternative cropping system for growing corn. My research question was: Considering sustiability, productivity, and environmental friendliness, what is the best cropping system for growing corn in the Midwest, USA?

In the research, I analyzed and compared 10 cropping sequences, 5 nitrogen rates (0 Kg/Ha, 65 Kg/Ha, 135 Kg/Ha, 225 Kg/Ha, 300 Kg/Ha), and 2 row spacings 37.5cm and 75cm). Each one of these components put together made up one cropping system (1 Cropping sequence + 1 row spacing + 1 Nitrogen rate = 1 cropping system) and since there was 100 distinct combinations to put these 3 components together, I analyzed 100 distinct cropping systems. For each cropping system I collected data using the Agricultural Production Systems sIMulator (APSIM) which is an internationally recognized platform for modeling and simulating agricultural systems. For each cropping system I had to parametrize the model accordingly and specify the model to collect data on certain variables. I specified the model to collect variables tracking data on the change in soil quality over time, crop yield, and greenhouse gas emissions. For each cropping system I collected 35 years of data. Once I collected all the data, I then preprocessed and then analyzed it using the GroupBy function in Pandas and the Principal Component Analysis (PCA) unsupervised machine learning algorithm.

For my results I concluded that Maize-Winter Wheat cover cropping sequence was the best cropping sequence since it allowed the soil to retain high soil moisture, high water retaining efficiency, and high nitrogen and carbon content as well as emitting very little N2O. For nitrogen rate I noticed a pattern that as nitrogen rate increased water retaining efficiency in the soil and soil moisture tended to decrease while nitrogen and carbon content in the soil as well as crop yield increased. Therefore, I initially concluded that the nitrogen rate of 135 kilograms per hectare was the best one since it allowed for a moderate balance of both extremes. However, since crop yield was a high priority for farmers, I concluded that a nitrogen rate of 225 Kg/Ha is the best nitrogen rate. For row spacing I found no significant difference between 37.5cm or 75cm.

Overall, I concluded that a cropping system that implements a cropping sequence of Maize-Wheat Cover Cropping, a nitrogen rate of 225 Kg/Ha, and a row spacing of 37.5 or 75 centimeters is the best cropping system for growing corn in the Midwest, USA. This cropping system is more sustainable than the monoculture cropping system because as per my data, it allowed the soil to retain a higher soil moisture, water retaining efficiency, and produced higher crop yields. As shown on the map down below climate change is causing a lot of the lands in the Midwest, USA to become dried out so it is especially important that soils retain their moisture in order to minimize impacts. The next steps for this project would be to verify these results on the field and then spread the message to farmers.