Neural network approach for aerosol retrieval

A neural network technique for nonlinear problems is applied for aerosol retrieval using satellite data. First the numerical results given by light scattering simulations in an atmosphere-ocean system are stored into a look-up table (LUT), and then the LUT is learned by our neural network system. After this learning process, our neural network code is applied to retrieve aerosol properties over the ocean using ADEOS/OCTS data. It is shown that the neural network technique is possible to reduce the processing time, and our code looks promising for effective aerosol retrieval on a global scale.