Near-real-time regional ionospheric data assimilation using the local ensemble transform Kalman filter

The high latitude ionosphere provides a challenging environment for space weather forecasting due to its highly dynamic behaviour and the sparsity of data. By design, climatological models cannot adequately capture the short-term variability of the ionosphere. To be able to provide the best possible understanding of the current state of the ionosphere, augmenting traditional empirical models with near-real-time data sources is necessary.

A number of instruments are able to provide data with less than two-hour latency, including Global Navigation Satellite System (GNSS) receivers, ionosondes, and satellite-borne altimeters and GNSS receivers. These instruments are both sparsely and unevenly distributed at high latitudes, which creates challenges for traditional assimilation approaches. The locations of ground-based GNSS receivers preclude tomography in much of the region, a condition which is worsened when assimilation is limited to those stations that provide near-real-time data.

We use a Local Ensemble Transform Kalman Filter (LETKF) to assimilate data from the above sources [1]. The assimilation is defined over a HEALPIX grid aligned with the geomagnetic pole, ensuring an even spacing of grid points and minimizing computational footprint [2]. The vertical electron density is parameterized as a modified semi-epstein layer, fitted from an empirical model. The Empirical Canadian High Arctic Model (ECHAIM) is used as a basis for the assimilation above 450 magnetic latitude, with the NeQuick model used below [3, 4].

We will here present results using real data collected during the geomagnetic storm during early September 2017. By limiting our assimilation data to those sources that provide data in near-real-time, we can test the capabilities of the assimilation during a large scale disturbance from climatology.