Predicting unavailable parameters from existing velocity fields of turbulent flows using a GAN-based model

This study developed a mapping generative adversarial network (M-GAN) to predict unavailable parameters: streamwise velocity, temperature, and pressure from available velocity components. Two-dimensional Rayleigh--B\'enard flow and turbulent channel flow are used to evaluate M-GAN performance. The results indicate that the proposed model has good capability to map the available parameters to unavailable parameters. Furthermore, M-GAN also has good generalization to predict the parameters from channel flows at different Reynolds numbers.