Development of deep neural network model for prediction of natural processes

The article proposes the principles of constructing a deep convolutional neural network model for solving the problem of high-precision forecasting of the development of natural processes, in particular, fires. The developed architecture of the neural network simultaneously integrates spectral and spatial information and consists of several modules: input, convolutional feature extraction, output. The model includes data processing components for accurate and reliable detection of natural processes based on Earth remote sensing materials within the deep neural network models repository.