Global TEC Map Fusion Through a Hybrid Deep Learning Model: RFGAN

Abstract Timely, reliable and comprehensive global observation information is essential for space weather research. However, limited technology hinders the consecutive coverage of data. For integrity continuity data, deep learning can obtain a Ionospheric total electron content (TEC) map by fusing multi‐source TEC maps. Different from previous methods, in study, hybrid model (RFGAN) based on Dual‐Discriminator Conditional Generative Adversarial Network (DDcGAN) Free‐Form Image Inpainting with Gated Convolution (Deepfill v2) proposed to fuse Massachusetts Institute Technology (MIT)—TEC, International Global Navigation Satellite System (IGS‐TEC) altimetry satellite TEC. Throughout RFGAN structure, we use an autoencoder gated convolution inpaint missing parts MIT‐TEC Meanwhile, DDcGAN fuses inpainted (MIT'‐TEC) IGS‐TEC get high accuracy. To certain extent, ocean area through RFGAN. At same time, keeps consistency RFGAN‐TEC continent area. Our be easily extended widely applied other fields science, especially addressing observational data loss fusion.