[GIS] Understanding strange measure distance results from QGIS

distanceopenstreetmapqgisweb-mercator

My colleague just told me about incorrect measure distance tool results in QGIS. For reference we used this pond.

While our local map provider Mapy.cz

and Google Maps:

both show about 970 meters distance, QGIS shows me about 1500 meters:

I first though it is OTF reprojection issue so I switched it off. But problem remains. The results are same for various Openlayers map services (OSM, Bing etc.) but also for OSM derived offline data.

I have my data in EPSG:3857 – WGS84 Web Mercator in order to be able to let the beginner users switch between offline and online data without slowing down the map loading and avoiding the need of OTF reprojection.

Is there something wrong with this particular projection? I would not expect such a huge differences for this small area.

Best Answer

By using the Web Mercator projection the data gets more and more distorted the further you go away from the equator. There is also a ESRI Blogpost where this problem is described.

And this is also the reason why the measurement on the OSM Layer is that much off, because you use EPSG:3857 in the project from your third screenshot. When your reproject it either manually or by OTF (e.g. to EPSG:4326), the measurement should be correct. Google and Mapy.cz probably solve this issue by converting it in the backend before calculating the "real" distance.

Related Solutions

[GIS] Offline Map Service with Mapnik and Leaflet

Your approach looks ok to me. On my local system, I skipped the TileCache part. Mapnik fills the folders in the way Openlayers reads them from disk using file:///... I don't know if leaflet can do it the same way.

For the satellite tiles: You have no chance to get them legally. Google and bing do not like storing their tiles locally. If you have mapnik-rendered tiles form OSM, they will be not transparent, so you will not see the satellite picture underneath.

If performance is an issue, you can render on a faster machine, and copy the tiles to slower ones for the end user.

I have not tried storing data back into the database.

[GIS] NNJoin, MMQGIS and Nearest Neighbor QGIS failed to measure shortest distance

If you have to measure the distances once again, the following Python code may save you some time.

Have your points and lines loaded as map layers. Change the names 'point' and 'line' in line 4 and 5 resp. to the names of your layers. Copy the code into the Python console.

The function iterates over all points and finds the closest segment of all lines, and writes the closest position along the segment to a list. Then takes the position with the minimum distance to a points and builds a connecting line. The length of the line goes to the field distance.

d_lyr = QgsVectorLayer('LineString?field=distance:float', 'dist', 'memory')
QgsMapLayerRegistry.instance().addMapLayer(d_lyr)

p_lyr = QgsMapLayerRegistry.instance().mapLayersByName('point')[0]
l_lyr = QgsMapLayerRegistry.instance().mapLayersByName('line')[0]
d_lyr = QgsMapLayerRegistry.instance().mapLayersByName('dist')[0]
prov = d_lyr.dataProvider()
feats = []      
for p in p_lyr.getFeatures():
    minDistPoint = min([l.geometry().closestSegmentWithContext(QgsPoint(p.geometry().asPoint())) for l in l_lyr.getFeatures()])[1]
    feat = QgsFeature()
    feat.setGeometry(QgsGeometry.fromPolyline([QgsPoint(p.geometry().asPoint()), QgsPoint(minDistPoint[0], minDistPoint[1])]))
    feat.setAttributes([feat.geometry().length()])
    feats.append(feat)

prov.addFeatures(feats)
d_lyr.updateExtents()
d_lyr.triggerRepaint()

To test the code initiate 2 memory layers, a point layer named 'point' and a line layer named 'line', digitize some features, and run the code.

An example:

Best Answer

Related Solutions

[GIS] Offline Map Service with Mapnik and Leaflet

[GIS] NNJoin, MMQGIS and Nearest Neighbor QGIS failed to measure shortest distance

Related Question