Utilizing Machine Learning Framework to Evaluate the Effect of Climate Change on Maize and Soybean Yield

Abstract

In recent years, climate change patterns and extreme weather events have adversely affected agricultural production and have raised concerns about its effect on crop yields. These changes can affect the crop yield in many ways including the changes in the length of the growing season, planting and harvest time windows, precipitation amount and frequency, and the growing degree days, etc. So, it is important to analyze the effect of climate change on yield variability for a better understanding of its effect on the crops. This study aims to analyze the historical monthly county-level data for the state of Ohio to quantify maize (Zea mays) and soybean (Glycine max) yield temporal variability and to study the impact of proposed climate change scenarios on the yield. Machine learning algorithms were used to model the effect of monthly temperature and precipitation levels on yield variability, and to study the effect of climate change scenarios on yield. In addition, yield prediction models were integrated with proposed climate change scenarios for higher and lower emissions scenarios to predict the maize and soybean yield for the year 2100. To model the effect of weather parameters on temporal yield variability, the Random Forest model outperformed with RMSE of 0.61 Mt/ha and R2 of 0.73 for maize and with RMSE of 0.21 Mt/ha, R2 of 0.64 for soybean crop. Maximum temperature for July which has a negative correlation with yield was found to be the most dominant parameter for maize yield followed by the precipitation in August, July, and June which have a positive correlation with maize yield. Precipitation in August was found to be the most dominant weather parameter for soybean yield followed by maximum temperature in July. Based on the projected average increase in Ohio’s temperature and precipitation levels alone, maize yield was predicted to drop by 13.2% or 18.5% respectively due to lower emissions and higher emissions climate change scenario of Ohio. Similarly, a 6.64% and a 9.63% drop is expected in the soybean yield by the year 2100 for both lower emissions and higher emissions climate change scenarios. In economic terms based on current commodity values, this is an astounding loss of 254.7 to 369.7 million USD for maize and 535.9 – 751.1 million USD for soybean crop to the state of Ohio.