1

I've created a map of Bordeaux using the QGIS map compositor by following this procedure:

  1. Used the OSM plugin to display an OSM map and zoomed to Bordeaux
  2. Used the print composer to create a map of Bordeax using A4 at full screen
  3. Added a scale bar, and compared the distance markers with a distance measured in GE between two points in Bordeaux

The two do not agree by quite a lot. The GE measurement for the test area was approximately 1100m, the scale bar suggested the same distance was approximately 1500m.

Is the QGIS scale bar in the print composer supposed to be only approximate? Anyone know if I'm doing something wrong? I don't believe the distance is long enough to worry about possible topography differences.

I've attached a screen shot to show the two measurements:

enter image description here enter image description here

Improve this question
5
  • Just from seing the images attached, I believe that the so called "big" difference is not that big, taking into account the resolution of both GE and OSM (within Qgis) maps. That is to say that the error stemms from the fact the the line and scalebar, are not measuring the same distances.
    – dof1985
    Commented Mar 19, 2015 at 12:08
  • Thank you for correcting my mistake with the scale bar, however I disagree with your comment. The difference is approximately 30% which is very significant. If you look at the scale bar, 0 - 1100m is only 2/3 of the way between the points measured in GE.
    – Single Entity
    Commented Mar 19, 2015 at 12:11
  • I guess that this link from a past thread might spread light on your question. Main notion of the thread, while reflecting back on your question, is that those two are not comparable, since many technical issues, source issues and processing that loose accuracy along it. (gis.stackexchange.com/questions/95133/…)
    – dof1985
    Commented Mar 19, 2015 at 12:54
  • That link suggests pixel accuracy and image accuracy. I've zoomed right in on GE and in QGIS to make this comparison to avoid that problem as much as possible. A pixel in both of these cases is a matter of a few meters, the difference I'm seeing is around 400 m which is 30%. Also, I believe having read the post you've linked, that they are referring to using GE for accurate measurements, not 1.1 to 1.5 km which is a significant difference. I question the accuracy of QGIS here.
    – Single Entity
    Commented Mar 20, 2015 at 10:26
  • I find it hard to believe that scalebar is incorrect. Anyway I tried it myself a couple of time, and notices that using a cartesian formulea to compute distance between to selected X,Y gives different results in GE and OSM tiles plugin in Qgis; I transform GE to UTM, and used default Pseudo-Mercator for OSM. It is weired since they both should be based upon the same datum, yet gives different results.
    – dof1985
    Commented Mar 20, 2015 at 12:02

2 Answers 2

Reset to default
1

I've spent a lot of time for this question, since it dealt with a peculiar phenomenon. I guess the solution found in the projection. Attached is an ESRI resource dealing with a similar issue: Measuring distances and areas when your map uses the Mercator projection.

First, it should be said that OSM tiles plugin in QGIS, and other web mapping services, using web-mercator, such as Pseudo-WGS84. This was mainly aimed to enable an efficient display of maps online and in common viewable resolutions. That is that it isn't accurate for measurments.

Google Earth is using WGS84, which makes it more accurate - thus it is not that qgis or its scalebar are wrong, but that the use of a tile map, from which to create a scalebar is wrong.

You might also be interested in this thread from the forum, which deals wih both projections. EPSG 3857 or 4326 for GoogleMaps, OpenStreetMap and Leaflet

Improve this answer
4
  • It sounds to me that you are saying that GE is more accurate and that essentially the scale bar is simply incorrect relative to the WFS tiles beneath it. In my mind this makes the scale bar wrong and misleading. Yes, indeed OSM does use Pseudo-mercator and this has precise measurements associated with it as its a well defined projection. I still contest that the scale bar is accurate and I think it should be used with caution.
    – Single Entity
    Commented Mar 20, 2015 at 15:13
  • @SingleEntity, This is definately WAS NOT what I was saying. Scalebar is fine, but when an inaccurate projection is under its base, it will give a "misleading" result. You can also try the ruler out of the map-composer, might it be that also the ruler is wrong? both are basd upon simple Math and coding. Have nothing else to say, except that measurments should not be based upon the OSM tiles, including scaling!
    – dof1985
    Commented Mar 20, 2015 at 15:16
  • The essential word in my comment there is 'relative'. I understand you are saying the projection is 'inaccurate' but remember that the scale bar is supposed to represent the map beneath it, consequently whether the scale bar is 'correct' or the map is 'correct' the fact is, the scale bar is not representing the scale on the map behind it. I should point out that I do indeed appreciate your efforts here, and the take home message is as you say, don't use OSM tiles + QGIS scale bar if you want to be 'accurate'.
    – Single Entity
    Commented Mar 20, 2015 at 15:58
  • I see, well if the map beneath, i.e. OSM tiles, is wrong; than scalebar would be wrong in comparison to the real world, but correct in comarison to the map.
    – dof1985
    Commented Mar 20, 2015 at 16:24
0

I seemed to be having this problem, but then discovered that the reason was that I had a second map included as an item in the Composer but with the tick-box empty so that it did not show. The scale bar was setting itself from this invisible second map, which had a different projection. Once I removed the second map from the Composer, everything worked correctly with the pseudo-Mercator projection. This need not be the explanation for your problem, but it may help others who are suffering similar symptoms

Improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.