NavList:

David Pike, you wrote:
"One thing this discussion shows is how remarkably accurate the lat and long readout on Google maps, which is presumably tied to WGS84, can be."

Yes, it's amazing, isn't it? And as another example, you can take the latitudes and longitudes in that "Table 1" in the article and visit all those former transit instrument locations all around the globe. In the bar graph I posted, you'll note that there is one location where the change in longitude is 14.5 seconds of arc. That's a lot, amounting to nearly one second in time. The estimate of the east-west tilt there is a nearly perfect match. So where would we get such a "tilted" platform? This site is listed as "San Juan". That's column SJ on the barchart. Its latitude and longitude are 31°30'34.0"S, 68°37'25.6"W (the link takes you there in Google Maps). If you zoom out, you find you're in the foothills of the Andes nearly due west from Buenos Aires. That's exactly the sort of place where you would expect a significant gravitational anomaly.

In addition to the issues Robert Wyatt mentioned, it's also worth remembering that the "satellite images" (often aerial photography rather than actual satellite data) are updated rather frequently, as often as once a year and with increasing frequency in recent years. New images are added by largely automated processes, and sometimes they have incorrect registration data. Last year there was a funny error just a few miles from where I live. It was easy to spot since a major sandy beach was split and offset about 100 feet south of its correct location. There were some weird "rainbow" effects where the behind-the-sccenes software tried to merge the misplaced photo with nearby frames. This happened to be right next to a small nuclear reactor just across the bay, and I had a few people convinced that the rainbows on the map were caused by radiation (oh yes... I did do that).  The road along the beach was misplaced, too, so for a while you could walk on water on the cove near the beach. Such miracles are ephemeral. It's gone now...

If each patch of highest-resolution imagery in Google Maps is 1 km across (for the sake of discussion), I would estimate that the photographic map data in 99.9% of those patches are correctly placed within one meter or better in terms of true (WGS84) latitude and longitude. But please count that as an educated guess! Note that vector mapping data, consisting of road networks, rivers and streams and all that, are another matter entirely. It's not hard to find roads drawn on top of images of roads where the network and the photo are clearly displaced relative to each other by dozens of meters. This is just flawed data, and it will take a long time to replace. The accuracy level here is considerably lower. You can find at least one displacement larger than 10 meters in as many as 10% or more of those 1km patches.

Frank Reed
Conanicut Island USA