A-CHAIM: Near-Real-Time Data Assimilation of the High Latitude Ionosphere with a Particle Filter
• +3
• Benjamin Reid,
• David R. Themens,
• Anthony Mark McCaffrey,
• P. T. Jayachandran,
• Thomas Ulich
Benjamin Reid
University of New Brunswick

Corresponding Author:breid.phys@gmail.com

Author Profile
David R. Themens
University of Birmingham
Author Profile
Anthony Mark McCaffrey
University of New Brunswick
Author Profile
P. T. Jayachandran
University of New Brunswick
Author Profile
The Assimilative Canadian High Arctic Ionospheric Model (A-CHAIM) is an operational ionospheric data assimilation model that provides a 3D representation of the high latitude ionosphere in Near-Real-Time (NRT). A-CHAIM uses low-latency observations slant Total Electron Content (sTEC) from ground-based Global Navigation Satellite System (GNSS) receivers, ionosondes, and vertical TEC from the JASON-3 altimeter satellite to produce an updated electron density model above $45^o$ geomagnetic latitude. A-CHAIM is the first operational use of a particle filter data assimilation for space environment modeling, to account for the nonlinear nature of sTEC observations. The large number (>10^4) of simultaneous observations creates significant problems with particle weight degeneracy, which is addressed by combining measurements to form new composite observables. The performance of A-CHAIM is assessed by comparing the model outputs to unassimilated ionosonde observations, as well as to in-situ electron density observations from the SWARM and DMSP satellites. During moderately disturbed conditions from September 21st, 2021 through September 29th, 2021, A-CHAIM demonstrates a 40% to 50% reduction in error relative to the background model in the F2-layer critical frequency (foF2) at midlatitude and auroral reference stations, and little change at higher latitudes. The height of the F2-layer (hmF2) shows a small 5% to 15% improvement at all latitudes. In the topside, A-CHAIM demonstrates a 15% to 20% reduction in error for the Swarm satellites, and a 23% to 28% reduction in error for the DMSP satellites. The reduction in error is distributed evenly over the assimilation region, including in data-sparse regions.