Thrust 3: GHG Modeling

In this research thrust, the team will measure the greenhouse gases and develop a statistical and machine learning model to predict the emissions.

For example, N₂O emissions from soil are responsible for 31% of all agricultural GHG emissions, have about 265 times the impact on global warming than carbon dioxide, and stay in the atmosphere for over 100 years. These emissions primarily come from the conversion of applied nitrogen fertilizers by microbes. Quantifying N₂O emissions from agricultural soils is extremely challenging, partly because the related microbial processes, mainly about incomplete denitrification and nitrification, are controlled by many environmental and management factors such as temperature and water conditions, soil and crop properties, and Nitrogen fertilization rate, which collectively lead to large temporal and spatial variabilities of N₂O emissions. Effective field-level GHG emissions quantification relies on accurate field sampling, subfield-scale process model, and detailed model validation.

While processed-based (PB) or artificial intelligence (AI) models show a promising performance in the improvement of GHG estimation, there are a set of key technical challenges on GHG fluxes simulation that the context of the developing world imposes on it, for example limited dataset and spatial-temporal variability, high computational cost in PB models for simulating GHG modeling, etc.

The team led by Natarajan and Diaz will study the data collection and modeling of the GHGs from the field collected measurements from the Kansas agricultural field.