This grant will design, build, test and analyze a unique and innovative computational tool for accurately predicting the occurrence of equatorial spread-F. The tool will be based on an artificial neural system (ANS), or neural network. These systems are unique in their ability to learn to characterize complex relationships using historical data. The system will learn to interpret pre-sunset data from a DMSP satellite probe and various ground-based remote-sensing systems, and categorize these data to predict the likelihood of post-sunset equatorial spread-F. Finally, once the ANS-based predictive tool has been developed, tested and validated, the nonlinear mapping the system has learned will be examined and contrasted with the analytical framework of plasma physics.
