Comparison of the Methods for Recovering Ionozonde Data Gaps

Short-term ionosphere prediction of the F2-layer parameters is very important problem. A number of algorithms has been developed up to date. Meanwhile, all of them suffer from the existance of gaps in ionozonde data. We have verified the following techniques, most suitable for recovering of data gaps: polinomial interpolation, tension splines, method of principal curves, running median and multiquadric method. The goal was to fill gaps in f_0F_2 data. It is shown, that the relative quality of recovering varies significantly, depending on the sparsity of the gaps and on their structure, namely, whether the temporal structure of gaps is random or not. The results for various levels of solar and geomagnetic activity as well as for seasons and LT are presented.