Update GMRT and Effelsberg ITRF coordinates to the VLBI / InPTA positions

[vhaasteren]

Summary

• Replace the legacy GMRT ITRF in observatories.dat with the InPTA / PINT
  coordinates (Update GMRT coordinates nanograv/PINT#2002, 2026-06-11): 750 m / 2.5 µs correction.
• Replace the legacy Effelsberg ITRF (shared by eff, effix, effedd, and
  leap) with the VLBI geodetic point used by PINT: 2.8 m / ~9 ns correction.
• No alias or clock-file changes; this is orthogonal to the recent clock-correction work.

Why

Both triples in Tempo2/TEMPO were of unknown provenance. PINT already carries
the published values:

| Site | Was (legacy) | Now | |Δ| |
|---|---|---|---:|
| GMRT | 1656342.30 5797947.77 2073243.16 | 1657059.36 5797913.14 2073026.71 | 750 m |
| Effelsberg (+ ASTERIX/EDD/LEAP) | 4033949.5 486989.4 4900430.8 | 4033947.146 486990.898 4900431.067 | 2.8 m |

GMRT matches the Observer’s Manual C02 reference (~19.1° N, 74.05° E, 588 m;
new height 580.9 m vs legacy 497 m). Effelsberg matches the May 2019 VLBI
az/el-axis point (~10 cm) and astropy’s EarthLocation.of_site("effelsberg").

On InPTA DR1 J1744−1134, aligning GMRT alone takes PINT–Tempo2 residual RMS
from ~833 ns to 0.82 ns on an otherwise unchanged .par/.tim pair.

I have tested this locally (Aug 6th, 2026 in my container on my macbook pro M1). But given the 'blast radius', it's probably best if someone independently does these:

Test plan

• Confirm the five updated lines parse in observatories.dat (whitespace-separated XYZ + name + code)
• Load a GMRT TOA with Tempo2 and check the site XYZ matches PINT’s gmrt
• Load an Effelsberg TOA (eff / effix / effedd) and check XYZ matches PINT’s effelsberg
• Spot-check that unrelated sites (e.g. ncy, pks, gbt) are unchanged

Discovery / acknowledgement

This was found when comparing MetaPulsar data combination results with the official IPTA DR2 data release

[SixByNine]

I think it would be best to confirm the Effelsberg position with e.g. David. I imagine that the VLBI position is correct, but I know we've had revisions of other telescopes to the "correct published position" which turned out to be incorrect. (I will ping David on Mattermost - lets see if he responds).

[SixByNine]

Just to record the status other than in slack messages - currently David Champion is checking with the geodesy team to confirm the best Effelsberg coordinate.