Atmospheric Methane Retrieval Based on Back Propagation Neural Network and Simulated AVIRIS-NG Data

Methane (CH ₄ ) is one of the main greenhouse gases, whose retrieval is easily affected by atmospheric water (H ₂ O) and surface albedo. In this paper, based on a radiative transfer model, the Airborne Visible/Infrared Imaging Spectrometer-Next Generation (AVIRIS-NG) radiance with different H ₂ O and surface albedo are simulated as training data. Back Propagation (BP) feed-forward neural network algorithm in machine learning is used to train the CH ₄ retrieval model, which is applied to quantify the atmospheric CH ₄ concentration. This method can effectively decrease the impact of atmospheric H ₂ O and surface albedo on CH ₄ retrieval. Moreover, this machine learning-based approach separates the processes of model training and prediction. This enables rapid characterization of CH ₄ emission point sources in images as the Matched Filter (MF) method, while also obtaining the column-averaged concentration of CH ₄ , similar to the Optimal Estimation (OE) method. The research results indicate that the Mean Absolute Percentage Error (MAPE) of the optimal BP model is as low as 0.33%. If necessary, further increases in training data can improve the resolution and applicability of the model.