Detection of GNSS-TEC Noise Related to the Tonga Volcanic Eruption Using Optimization Machine Learning Techniques and Integrated Data

Total Electron Content (TEC) is the integral of the electron density along the path between receivers and satellites. TEC measured from Global Navigation Satellite Systems (GNSS) data is valuable to monitor space weather and correct ionospheric models. TEC noise detection is also an essential channel to forecast space weather and research the relationship between the atmosphere and natural phenomena like geomagnetic storms, earthquakes, volcanos, and tsunamis. In this study, we apply optimization machine learning techniques and integrated GNSS and solar activity data to determine GNSS-TEC noise at the International GNSS Service (IGS) stations in the Tonga volcanic region. We investigate 38 indices related to the geomagnetic field and solar wind plasma to select the essential parameters for forecast models. The findings show the best-suited parameters to predict vertical TEC time series: plasma temperature (or Plasma speed), proton density, Lyman alpha, R sunspot, Ap index (or Kp, Dst), and F10.7 index. Applying the Ensemble algorithm to build the TEC forecast models at the investigated IGS stations gets the accuracy from 1.01 to 3.17 TECU. The study also shows that machine learning combined with integrated data can provide a robust approach to detecting TEC noise caused by seismic activities.