Riepl gave an update of the status of studies on tropospheric refraction, in particular the developments after the meeting in October 2002 (Appendix 19). There, a proposal was made for specific test case to test new dispersion models (zenith delay formulations and mapping functions). So far, two analysis groups (ASI and JCET) have participated in these investigations.
Luceri reported on the developments at ASI (Appendix 20). In particular, 3 different options were tested: (i) the standard Marini-Murray formulation, (ii) the combination of the Saastemoinen zenith delay and the Mendes mapping function, and (iii) the Saastemoinen model extended with the Ciddor dispersion formula, again completed with the Mendes mapping function. ASI has investigated the effects on LAGEOS-2 observations, taken in the period 1999-2001 and analysed in monthly data intervals. Coordinates were kept fixed at ITRF2000, which might affect the outcome of the residuals of course (i.e. favoring the model that was used for the input of ITRF2000, notably Marini-Murray).
One direct observation was that the far majority of the low-elevation SLR data was taken by 3 stations only: 7835, 7839 and 7845 (action item CB: encourage the global network to improve, wherever possible, this situation). Other direct observations were made on the effect of the various modeling options: for “green” SLR systems (wavelength 532 nm), nominal differences between any of the various options of up to 1 mm were observed, whereas at a wavelength of 423 nm (Zimmerwald) the nominal differences might increase to up to 6 mm at low elevations. Next, orbital computations were done and the residuals inspected to see which model is most favorable. Here, a mixed conclusion had to be drawn: 1) the model Mendes+Ciddor generally gives residuals less biased than the Mendes model, 2) at 532 nm the Marini-Murray gives better results (up to 2 mm) for observations below 30 degrees, 3) at 423 nm the Mendes+Ciddor is better (up to 2 mm) at all elevations. Clearly, the influence of the model for station coordinates (and its background) is playing an important role here. Finally, the situation for the MLRO system was investigated as a special case: this system obtains high-frequency range observations at green and UV wavelengths, allowing a completely independent assessment of the tropospheric delay. Based on actually observed delay differences and their model equivalents, it was concluded that the Saastemoinen/Ciddor/Mendes combination best approximated reality at the full range of elevations.
Pavlis reported on his investigations (Appendix 21). He processed full-rate SLR data on LAGEOS-1 and LAGEOS-2 in weekly data arcs, covering the years 1999-2002. As modeling options, Pavlis used (i) the standard Marini-Murray formulation or (ii) a modified Saastemoinen model for zenith delay in combination with the Mendes mapping function. The data on LAGEOS-1 appears to be slightly more affected by the choice between the two options than the LAGEOS-2 data is. The data distribution as a function of elevation may play a role here.