Errors in the Correlation Analysis of Ionospheric Drift Records

Abstract

The errors in the computation of the ionospheric drift parameters by the correlation analysis of spaced fading records are discussed on the basis of the observations at Thumba. To obtain a reliable correlogram, the optimum sampling rate is suggested to be 4–5 values per fade and the length of the fading record should include at least 15 fades.

The errors in the computation of drift direction and speed due to the anisotropy of the ground diffraction pattern are estimated. Even though the patterns obtained at Thumba are highly elongated, the errors are very small due to the drift being always in a direction perpendicular to the elongation of the irregularities. Average correction is about 5° for direction and 2 per cent for speed.

It has been found that when V²_ccomes out to be negative using Phillips and Spencer's method, the correlograms are found to be asymmetric, non-gaussian or having very poor correlation coefficient. A wrong conclusion that the apparent drift is less than true drift in these cases can be avoided if one uses the peak of correlogram and not the cross-correlation coefficients at zero time shift. The occurrences of such negative V²_c at Thumba are most frequent when fading is fast and pattern sizes are small in comparison to antenna separation. By recording fadings at a number of antenna, it is shown that large separation along E-W may give rise to correlograms with negative V²_c. Similarly, the insufficient separation of antenna in N-S direction would lead to the derivation of a smaller value of elongation ratio.