4

I have been working on GPS data and I do see observe some grid or square like patterns from GPS.

Any reason why this could be?

enter image description here

PolyGeo
  • 65,136
  • 29
  • 109
  • 338
Roger Ganga
  • 51
  • 2
  • 1
    Pics or it didn't happen aka please attach a screenshot showing what you're talking about. – Erik Jul 26 '21 at 07:53
  • 1
    @Erik Pl check the image above. – Roger Ganga Jul 26 '21 at 08:04
  • 3
    Values of your coordinates are rounded to a specific amount of decimal places. Simple as that ;) – MrXsquared Jul 26 '21 at 08:17
  • Thank yuo @MrXsquared but what could be the underlying reason the co-ordinates are rounded? Is it a common GPS error that we observe? – Roger Ganga Jul 26 '21 at 08:20
  • 2
    Can't tell without more details. Most likely software related. – MrXsquared Jul 26 '21 at 08:46
  • 4
    It could be done to anonymise the data (ie. to prevent you to link a point to a specific adresse) – J.R Jul 26 '21 at 08:55
  • I have recently observed this when extracting geotags (i.e. coordinates where the photos were taken) from photos taken with an Android mobile phone. I didn't dig deeper into the matter, but it was clear to me that this was due, as others have noted, to coordinate rounding. I might add that the resolution was sufficient for my application, where the photos were taken inside buildings and the points, although displaying the grid pattern were still inside the relevant building. – Techie_Gus Jul 27 '21 at 08:09

1 Answers1

7

As mentioned in the comments, this looks like a rounding issue at a first glance. Raw GPS coordinates are stored in WGS 84 coordinates (a geographic coordinate system). Very roughly,you need 5 decimal to have a precision of approximately 1m. See here for more detail. And be aware that the size of a degree of longitude changes according to the latitude. With a bold estimate of the scale of your map, the rounding would be to the second decimal but the spacing does not precisely correspond with the distance that I would expect from a rounding in this area. You could test this hypothesis by looking at the lat/long coordinates of successive points.

On the other hand, there are a few points near the word Chennai that do not have this systematic spacing. So either those points were collected with another precision, or the pattern is simply due to a systematic sampling on purpose (statistical inference, spatial aggregation for anonymity...). I don't know the projection used, but in a Plate carree projection (often used by default with geographic coordinate systems), the pattern of rounding errors should be vertical and horizontal lines. More context needed to be sure of the cause.

In any case, the pattern is not due to the fact that GPS is used: either something happened to the coordinates after being collected or the data was collected with a specific spatial pattern.

radouxju
  • 49,636
  • 2
  • 71
  • 144