Bug: wrong error message

[VikingScientist]

Requesting to download an area of less than 16 km² still gives the error message that this is too big. My current guess is that it won't accept long_thin areas, but anything less than 4km x 4km is fine.

[TheBB]

This is because the bounding box in UTM33 axis-aligned coordinates is bigger than the actual polygon, which is axis-aligned in latitude and longitude.

[VikingScientist]

Really? Is the difference this great?

I tried creating a stupidly long thin region which reportedly is less than 6km², and this also failed.

EDIT: When it was thin enough (0,7km²) it actually did work.

[VikingScientist]

Your reasoning makes sense, but we'll still keep it as a bug since the error message is misleading. Potential solutions

• Write out in clear text in the error message that area on right side is approximation due to coordinate transformation and the actual value is something else
• Show the returned polygon on the map by transforming coordinates
• Compute the actual area on the right-hand side and show the correct used value

[TheBB]

I agree it's misleading. Just to be clear though, the area on the right is not an approximation and it is the genuine area of the polygon. Hoydedata.no then computes a bounding box in the UTM33 coordinate system, which will generally be larger. It's this bounding box that must be limited by 16 km2.

[VikingScientist]

So the "square" bounding box is given in (lat,lon), which translates to a parlellogram(ish) region in UTM33 which is the one returned from hoydedata.no. This is also the reason for the cut-off no-data regions in the seemingly rotated rectangle which is returned as a tiff file.

You can actually see this distortion with your naked eye if you look at for instance Nordkapp in google earth (lat-lon) vs hoydedata (UTM33) where the angle of the coastlines go in different directions

Høydedata.no

Google maps

[VikingScientist]

If the "preview" button on #43 is implemented, then this will also be fixed