[Qgis-user] Help with scales in print layouts for geographic CRS

[Nyall Dawson]

On Thu, 7 Apr 2022 at 10:26, Greg Troxel <gdt at lexort.com> wrote:
>
>
> Nyall Dawson <nyall.dawson at gmail.com> writes:
>
> > In all cases (projected OR geographic) the scalebar logic is:
> >
> > 1. Create a horizontal line across the width of the layout map
> > 2. Calculate the length of that line using great circle/ellipsoidal
> > calculations, based on the project's ellipsoid settings.
> > 3. Compare the length of the ellipsoidal line vs the map width to
> > calculate the corresponding scale
>
> That is a very straightforward approach with easy-to-understand
> semantics.
>
> > So qgis scalebar calculations are ALWAYS based on great
> > circle/ellipsoidal lengths*.
>
> And more precisely, a true geodesic between two points varying in x and
> with constant y in the layout CRS.  For north-up (at least ish)
> projections, that means a line of constant latitude.
>
> For a lat/lon CRS in the layout, at mid latitudes, it's going to mean
> that the scale in the y axis is smaller by about 110km/70km; a square in
> lat/lon is taller than it is wide (and plus it's not a square, but
> that's less obvious).  Equivalently, a square on the ground has a larger
> difference in longitude than it does in latitude.  Or maybe I have that
> backward, but using lat/lon in a layout does not preserve shapes.
>
> This is what I was trying to get at: a single concept of scale is really
> only valid for a projection that has the same scale in x and y.  Which I
> think is true iff the projection is conformal.  And it's only truly
> valid if the scale doesn't change over the layout, which is probably
> ~never exactly true, but many projections aim to have it remain close
> enough to be treated as constant (e.g. UTM).
>
> Which is a long way of getting around to "if you are using lat/lon as
> the axes in your layout, either you are doing something odd on purpose
> or you probably should rethink your approach."  If you really do mean
> it, I would find the explanation of why interesting.

All correct, indeed!

I've been thinking about introducing a new warning when exporting
layouts, which would look something like this:

1. For each scale bar, look at the referenced map
2. Measure the scale using multiple lines, each across the
top/middle/bottom horizontally and left/center/right vertically
3. Compare these scale values, and if any differ by more than XXX %
then show a warning to the user. Something like "The scale for map 1
varies from 1:xxxx to 1:yyyyy across different parts of the map.
Consider using an alternative form of representing the map scale (such
as a grid overlay), as the scale bar may be misleading for this map."

What do you think?

Nyall


>
> > * Unless the project itself is set to not use ellipsoidal calculations
> > and only use planar calculations, in the project properties dialog
>
> Presumably those calculations are in the project CRS's xy space, and
> would typically (when used sensibly) be grid distance in UTM, some State
> Plane Coordinate System, or similar.