A Novel Robust High‐Precision Ionospheric Delay Modeling for PPP‐RTK Using Crowdsourced Data

Abstract The performance of PPP‐RTK is critically dependent on the quality of ionospheric corrections derived from a reference network. While the proliferation of mass‐market Global Navigation Satellite System (GNSS) devices offers a promising crowdsourcing opportunity to densify or replace this infrastructure, it introduces the formidable challenge of QC for the crowdsourced Slant Ionospheric Delays (SIDs). To address this challenge, we propose a robust SID purification method combining preliminary spatiotemporal consistency checks with final permutation tests. To simulate a realistic crowdsourced environment, dynamic restart experiments were conducted using stations from the Hunan Continuously Operating Reference Station (HNCORS) network. Results show a 50.4% reduction in the Root Mean Square (RMS) error of the crowdsourced SIDs, achieving a final accuracy of 0.6 TECU. Then, those QC‐processed SIDs were assimilated into the Quasi‐4‐Dimensional Ionospheric Model (Q4DIM), and applied to PPP‐RTK positioning for further validation. The proposed method enables the crowdsourced solution to match the performance of PPP‐RTK using SIDs from a continuous network. Compared to the raw crowdsourced SID solution, the QC‐processed solution improves accuracy from 0.20 to 0.02 m and reduces convergence time from 3 min to a single epoch. This demonstrates the proposed method effectively reduces outliers in crowdsourced SIDs, with the potential to enable mass‐market users to provide reliable PPP‐RTK services.