A High-Efficiency Multivariable TEC-SOFTS Model for Ionospheric TEC Prediction: Validation Over China Area During Low and High Solar Activity Periods

This repository contains the code and data for the paper "A High-Efficiency Multivariable TEC-SOFTS Model for Ionospheric TEC Prediction: Validation Over China Area During Low and High Solar Activity Periods". It implements a deep learning-based model (TEC-SOFTS) for predicting ionospheric Total Electron Content (TEC) over the Chinese region, validated during both low and high solar activity periods. Data Sources CODE GIM data: Obtained from the Crustal Dynamics Data Information System (CDDIS) at https://cddis.nasa.gov/archive/gnss/products/ionex/. Space weather indices: The six key influencing factors (SSN, F10.7, Lyman-alpha, Dst, Ap, and Kp) were provided by the OMNI data set, accessible at https://omniweb.gsfc.nasa.gov/form/dx1.html. Code Structure layers/: Core neural network layers and components of the TEC-SOFTS model. models/: Definition of the TEC-SOFTS model architecture. exp/: Experiment scripts for training, validation, and testing the model. data/: Data preprocessing and loading utilities. checkpoints/ and checkpoints24/: Pre-trained model checkpoints for different experimental setups. utils/: Helper functions for data processing, evaluation, and visualization. results/: Output directory for prediction results and evaluation metrics.