The Inventory Optimization Code for Exploring Deep Learning in Air Pollutant Emission Estimation Scale

The Inventory Optimization code works for emission inventory estimation. It utilized a neural network based comprehensive chemical transport model (NN-CTM) to learn the complex correlation between emission and air pollution. We can further update the emission inventory using this code based on backpropagating the gradient of the loss function measuring the deviation between NN-CTM and observations from surface monitors.