I've just been checking out a few mine locations on the excellent NMRS Google Earth application and have found some reasonably large discrepancies.

For example, the location of Sowden No.4 in New Farnley near Leeds is way off. It should be on the other side of the road directly opposite the southernmost of the three semi-detached houses just before the junction of Low Moor Side Lane and Walsh Lane. (I used to live in that house!).

Just how reliable are the coordinates?

Cheers
D.

BTW: the map is here:
http://www.nmrs.org.uk/mines/coal/maps/england3.html 

The 3D version might take quite a while to load on slower Internet connections. However, it does show the mine names automatically and a double-click shows a pop-up with information - on the 2D version you have to click on the symbol to get the name and information.

All depends on whether they use GPS or actually add a push pin to Google Earth map at the location. Seems SatNav is sometimes way out!

There's a discussion on UKC on GPS accuracy at the moment.

I grid ref co-ord is just the southwest corner of a square and everything in that square has the same co-ord.
Thus, by default a 6 fig grid ref (i.e. a 100m square) can be correct, but an actual position of something can be more than 100m away.

Then it would depend on how they got the grid ref and how accurately it got converted to long/lat or put on to a map.

---

There's a discussion on UKC on GPS accuracy at the moment.

Wonder how long it'll be before it turns into an argument?

---

There's a discussion on UKC on GPS accuracy at the moment.

Wonder how long it'll be before it turns into an argument?

I would speculate we'll continue to remain quite gentlemanly on here, UKC is anybody's guess though. :lol:

I grid ref co-ord is just the southwest corner of a square and everything in that square has the same co-ord.
Thus, by default a 6 fig grid ref (i.e. a 100m square) can be correct, but an actual position of something can be more than 100m away.

Then it would depend on how they got the grid ref and how accurately it got converted to long/lat or put on to a map.

Thanks for the info - I just measured it in Google Earth, the shift is around 120 metres, but way out of the square. It would probably fit if the ref was the NE corner of the square
D. 😉

Cheers
D.

100 metre grid square is 142 metres diagonally across corners. 🙂

Google Earth is not the best for its spatial accuracy which depends upon the maps used for X,Y location and the Shuttle radar digital terrain model for Z location. Conversion from grid coordinates to Lat/Long should also take account of elevation and the local deviation from the Earth's ellipsoid (as reflected in the datum). In my experience Google Earth can be up to 100m wrong in its location. Your GPS should be better accuracy (2-5m). So generally it is not the map/GPS that is wrong but Google Earth.

Martin

..Conversion from grid coordinates to Lat/Long should also take account of elevation and the local deviation from the Earth's ellipsoid (as reflected in the datum).

When I produced my Bath Freestone GoogleEarth map
http://www.bristolunderground.com/dpexplorer/dpexplorer.html ). I noticed Bill Chadwick's javascript code took a height parameter and so I did some experiments and systematically varied the height from sea level to 1 kilometer and was not producing noticeable differences on my map for my data around Bath.

This error does depend up your location but I would suspect your observed errors are not due to using the wrong height.

Dave

---

I just took the coordinates given by NMRS for Sowden No.4 Pit and entered them in "Old Maps" and got exactly the same shift from the actual position as in Google Earth.

The Coal Authority map

http://mapapps3.bgs.ac.uk/coalauthority/homeiframe.html 

seems to have the correct coordinates for the three shafts at the junction of Low Moor Side Lane and Walsh Lane in New Farnley.

All very puzzling.

Whatever the technical problem is it also applies to the 'Disused Rail Tunnel' database.

Edit: Just checked the Rhondda Tunnel and find the East portal is marked at 25 metres North of actual portal, West portal would be about right if the line marker was 180 odd metres longer!

I suspect it is not the conversion from National Grid to Lat/Long that is the problem but the accuracy of the geometric correction of the satellite imagery by Google.

My professional job involves working with satellite imagery and it might help if I explain some of the issues with Google Earth.

Raw imagery obtained from a satellite is usually only positionally accurate in the range 100m to 1km (depending upon the satellite being used). Also, many of the images are not obtained vertical scans but from adjacent orbits which gives problems of distrortion in areas of topographic relief. In order to get better accuracy requires geometric correction of the imagery using specialised software with ground control points (of known location) and a digital terrain model. Google use a limited number of ground control points around the world and the 90m resolution Shuttle Radar Digital Terrain Model to correct the satellite imagery.

There have been serveral studies on the positional accuracy of Google Earth and most give figures in range of 25m - 100m average positional accuracy (see www.fig.net/pub/fig2011/papers/ts05i/ts05i_becek_ibrahim_4947.pdf? for one such study)

The Mining Heritage Trust of Ireland is presently producing a Google Earth based map viewer of the mines in Ireland. We have accurate locations for the mines from field surveys, air photos and maps. The mines database is stored in ARCGIS and converted to KML for Google. Often when brought into Google Earth the mine locations don't match the satellite imagery. Our solution is keep the mines location correct in the GIS but to relocate the lat/long coordinates to fit the Google Earth image. Unfortunately this involves a visual check and if necessary the moving of the mine location in Google Earth. Also there is the possibility that a corrected mine location may not match the imagery in future versions of Google Earth if new imagery is being used by Google.

That explains a lot regarding how it all works and the accuracy issue.
Would the accuracy issue also effect the red highlighted positional info at the bottom of the image? Pin was placed manually (so is in correct place) but your explanation seems to indicate positional info may still be in error?
Apologies for ignorance in advance.

🔗Personal-Album-1695-Image-88519[linkphoto]Personal-Album-1695-Image-88519[/linkphoto][/link]

Yes, the Lat/Long position shown in Google Earth is where it thinks it is based upon the imagery but if the imagery is wrong then the Lat/Long will be (slightly) wrong. In our MHTI we will store 'real' National Grid coordinates (plus Lat/Long) and the Google Earth Lat/Long from any visual relocation.

Martin

Thanks. I always wondered about accuracy whilst tranferring push pin locations to MemoryMap using the Degrees, minutes, seconds function.

Fascinating!

I now have absolutely no idea exactly where the hundreds of mine locations I've researched so far really are. (From north of Bradford down to Penistone and from Ogden across to Wakefield)

On a lighter note. What do we learn from this?

Never walk backwards from where your coordinates say there's a mineshaft. 😉

(see www.fig.net/pub/fig2011/papers/ts05i/ts05i_becek_ibrahim_4947.pdf? for one such study)

Interesting this study used the start and end of runways as features with known locations. It reported inaccuracies of
http://sdrv.ms/139pfo9 ). In the early days of Google Earth these errors were common and there was one near where I lived.

If the disparities are known at a collection of discrete locations then this could be used to predict (interpolate)
the disparity at other locations. In fact I am surprised this
has not already been done and put online e.g. Your mine location data could be used for this purpose.

Unfortunately this involves a visual check and if necessary the moving of the mine location in Google Earth. Also there is the possibility that a corrected mine location may not match the imagery in future versions of Google Earth if new imagery is being used by Google.

You need to use features that can be detected using image processing and so you can track when Google Earth changes. These features are unlikely to be mine locations, other studies have used road intersections.

---