Why does QGIS get the whole connected component regardless of cost when calculating service area?

Question

I'm new to QGIS and I'm having a strange problem with its service area tool. I have a line layer representing a network of roads, imported from openstreetmaps. If I calculate the service area from a point using distance, then it works as expected. For example, if I set the cost to 1000 meters, it returns a line layer with all the paths of 1000 meters from that point.

However, if I try to get the service area based on travel time instead, then no matter what speed I set for the features, or cost I set in terms of time, it always returns a line layer with every reachable road highlighted. It's as if it's assuming infinite speed of travel, so every reachable point is reachable in any amount of time.

What am I doing wrong?

The data set in this example is from St. Pierre and Michelon, acquired here: OSM data

The roads were extracted from the raw pbf file using osmosis with this command: osmosis --read-pbf .\saint_pierre_et_miquelon-latest.osm.pbf --tf accept-ways highway=* --tf reject-relations --used-node --write-xml example.osm

Answer 1

There are at least two problems:

1. Travel costs for fastest travel type are hours - so a value of 0.5 means 30 minutes.

2. A just superficial look at your screenshot shows that your network layer is in EPSG:4326, so layer units are in degrees. If I run the tool with shortest travel type, the distance are calulated in ellipsoidal distances. So, whatever CRS your layer is in, you'll get more or less "correct" distances.

   Still, I would heavily advice against using EPSG:4326 for any measuremetns. A distance of 0.5 in this CRS means 0.5 degrees, what is quite a large distance (depending on the exact part on the globe you're working). Degrees never make sense for distance measurements. Reproject your layer to a projected CRS suitable for measurements in your area of interest - e.g. the local UTM zone.