Direct Comparison of Delays in TP015

Figure 4: Comparison of S/X delays with ionospheric models for data from NL. The + signs indicate the S/X measurements. See the text for a detailed description.

Examination of the phase plots above, or images discussed below did not clearly demonstrate the effects of the use of the GPS models. So a program was written to directly compare the different ionospheric delays in 3 different ways. Figure 4 shows one of the stations for which the models give the best corrections -- NL. The upper left panel shows the total dispersive delay predicted by the JPL, CODE, and SATLOC models, adjusted to apply to 2.3 GHz. The CODE model truncates at 23 hours -- this was to avoid the problem with editing the IONEX files mentioned earlier. The effort was made to edit the JPL model files and the SATLOC model was derived in a different way so this was not a problem. The other panels are one per source.

The lower plot in each of the other panels is the data for this station with the interpolated data for LA, for the same source, subtracted. For this plot, the variations between the S/X data and the models should track if the models are good. But there is still an arbitrary constant offset because the possible offsets in the VLBI data, described earlier, are station dependent and are therefore not removed by the referencing. The S/X data have been shifted so that the average matches the average in the models, but still any constant offset should not be a concern. As can be seen, the slopes are generally correct, but none of the models track the S/X data exactly.

The upper plot in each panel is the data from the lower plot, but now with the values for the reference source removed. This is what would actually be used in a phase referencing observation. Now the constant terms, which are the same for all sources, have been removed by the referencing and the data from the S/X and the models should match if the corrections are good.

Figure 5: Comparison of S/X delays with ionospheric models for data from OV. The + signs indicate the S/X measurements. See the text for a detailed description.

Figure 6: Comparison of S/X delays with ionospheric models for data from MK. The + signs indicate the S/X measurements. See the text for a detailed description. Note that the SATLOC data does not cover MK.

Figure 5 and Figure 6 show similar plots for OV and MK. At OV, the models deviate from the S/X results in some fairly clear ways. At MK, the SATLOC model provides no coverage and the other two have even more significant deviations from the S/X data than at OV. But in all cases, it is clear that correcting the data using the models would be better than making no corrections at all. A rough impression from examining the plots, plus examining the RDV11 data, is that the models will provide a factor of about 2 to 5 improvement over not using any ionospheric correction.