A comparison of reconstruction techniques used in ionospheric tomography

Abstract

Abstract. This paper compares the performance of three different reconstruction algorithms used for tomographic imaging of the ionosphere. The test cases chosen make use of experimental electron content observations obtained from a meridional chain of receivers in Scandinavia at times when the European incoherent scatter radar facility provided independent measurements of electron density for verification of the reconstructions. The examples include ionospheres with layers peaking at both anomalously high and very low altitudes, in addition to a smooth horizontally stratified layer. The results expose the limitations of an algorithm based on use of a standard ionospheric model but demonstrate that with appropriate choice of background a priori profiles, reliable images are obtained from both discrete inverse theory and quadratic programing. For these two latter algorithms the resultant reconstructions are essentially independent of the algorithm used. The study highlights the critical role of the range of the background input to the algorithms. A density value at a single ghd cell provides sufficient additional a priori information to constrain the imaging of a horizontally stratified ionosphere.

Tomography Reconstruction of Ionospheric Electron Density with Empirical Orthonormal Functions Using Korea GNSS Network

“…Note that the elements of and matrices can be negative values bec absolute values of in the exponent to prevent divergence of the c changed to 0.01 in this study from 0.2 in previous study (Pryse et al 1998 two cases: (1) …”

“…To overcome this limitation, iterative methods like algebraic reconstruction technique (ART) and MART have been suggested in several studies (Raymund et al 1990;Raymund 1995;Pryse et al 1998). Especially, MART, a modified version of ART, finds the solution using ratio between an initial guess and calculated values.…”

“…Accordingly, we take absolute values of H in the exponent to prevent divergence of the computation. The relaxation parameter, λ', was changed to 0.01 in this study from 0.2 in previous study (Pryse et al 1998). The termination condition of MART was set as two cases: (1) the relative update value,…”

“…Accordingly, we take the exponent to prevent divergence of the computation. The relaxation parameter, ′ , was tudy from 0.2 in previous study (Pryse et al 1998). The termination condition of MART was set as e update value, ( +1 − ) × 100, is less than 1.0 (less than 1 %) or (2) the relative update value ersion of MART iteration reached to the above termination condition, the normal electron density transforming back to as a solution to CIT.…”

“…Therefore, a number of studies proposed various tomographic methods to mitigate difficulties in solving the ill-conditioned inverse problem. (Raymund et al 1990;Kunitsyn et al 1994;Raymund 1994Raymund , 1995Pryse et al 1998). …”