Validation of the NeQuick model, Ensemble Kalman filter and SMART+ based estimations of the topside ionosphere and plasmasphere

Abstract

In this work, we analyse the performance of an Ensemble Kalman filter (EnKF) approach, the simultaneous multiplicative column normalized method SMART+ and the NeQuick model to estimate the topside ionosphere and plasmasphere. The slant total electron content (STEC) measurements of 11 Low Earth Orbit (LEO) satellites are used as input for the EnKF and SMART+ to update the NeQuick model which serves as background model.

Our comparative case study is implemented globally for altitudes between 430 and 20 200 km for two periods of the year 2015 covering moderate to perturbed ionospheric conditions.

The performance of the methods is investigated regarding their capability to reproduce electron densities. For that purpose, the independent electron density measurements of the Van Allen Probes (VAP) twin satellites, Swarm Langmuir Probes (LP) and FORMOSAT-3/COSMIC ionospheric radio occultation (IRO) profiles serve as reference.

The results reveal that in median the NeQuick model underestimates the IRO electron densities in the moderate period and for high latitudes in the perturbed period. For the NeQuick model the median of the absolute relative residuals is about 33 % in the moderate period and around 25 % in the perturbed period. SMART+ reduces these residuals to about 24 % and 23 %, respectively, whereas EnKF doesn’t provide an improvement.

The validation with VAP electron densities shows that in both periods and for all altitudes, the electron densities provided by the NeQuick model, the EnKF and SMART+ are in general significantly lower than the VAP measurements with a median of the relative residuals equal to about 80 %. The lowest median and RMS values of the VAP residuals are produces by EnKF, reducing the NeQuick statistics by up to 9 %.

Further we observe that the electron densities provided by the NeQuick model are in median lower than the calibrated LP in-situ measurements for all three Swarm satellites in the moderate period and higher in the perturbed period. SMART+ shows the lowest median and standard deviation for the absolute relative residuals. These are up to 20 % lower than for the NeQuick model. For the LP in-situ measurements, the EnKF has the worst performance.

Overall, the results underpin that even though the data of 11 satellite missions has been assimilated to adjust the a priori information of the NeQuick model, only limited improvements can be achieved especially for the plasmasphere.