REUSABLE ML MODEL TO ESTIMATE PRODUCED EMISSIONS OF COAL-FIRED POWER PLANTS

One of the human activities we should be concerned about is the emission from factories. To get a low-cost real-time status update about power plant/factory emission, we built a machine learning model that is trained on satellite observations (Sentinel-5), ground observed data (EPA eGRID), and meteorological observations (MERRA). A unique approach to preprocessing multiple data sources, coupled with multi neural network model steps (RNN, LSTM) produces an automated way of classifying the amount of emission produced by a single power plant of the users’ choosing or predicting the emissions produced of all power plants found in the US. Predicting ground emissions from remote sensed data has its many challenges to accurately have a consistent predictive model, but is nonetheless a positive use case to understand and quantify further emissions that could be produced in the future.