Critical infrastructure in the civilian, commercial, and military sectors can be harmed by space weather caused by highly variable solar and solar wind activities. Understanding the underlying physical processes of space weather, and improving our specification and forecasting, are required at the national level to protect vital assets on the ground and in space. One major space weather threat is ionospheric disturbances, which can lead to scintillations of the Global Navigation and Satellite Systems (GNSS) signals and disrupt radio operations. My group applies advanced statistical tools and machine learning methods to overcome the challenges of sparsely and irregularly sampled ionospheric data and to improve the specification and forecasting of local and global ionospheric disturbances and their variability. Our ongoing projects include:
1. Develope an innovative matrix completion method VISTA (Video Imputation with SoftImpute, Temporal smoothing and Auxiliary data) to reconstruct time series of ionospheric TEC maps with a large amount of missing values and preserve observed meso-scale structures.
2. Apply machine learning methods, such as LSTM and the modified U-Net model, to forecast ionospheric TEC maps hours and days ahead.
3. Use machine learning methods to forecast the ionospheric equatorial plasma bubbles, which the most severe ionospheric disturbance.
