[Qgis-developer] Area calculation issues

[Nyall Dawson]

On 3 February 2017 at 16:54, Neumann, Andreas <a.neumann at carto.net> wrote:
> Dear Nyall,
>
> Thank you so much for analyzing the issue and also thank you in advance for
> porting the fix to 2.18!

Done now. I also added one of those polygons to the reference geometry
test to ensure that all other stats (perimeter, centroid, etc) match
PostGIS' results.

Nyall


>
> I opened the sample project with 2.14 and I can confirm that the area values
> of both planimetric and ellipsoidic area calculations are very close for all
> features.
>
> It is very much appreciated!
>
> Thanks,
>
> Andreas
>
> On 2017-02-03 01:36, Nyall Dawson wrote:
>
> On 1 February 2017 at 19:55, Neumann, Andreas <a.neumann at carto.net> wrote:
>
>
> Hi Nyall,
>
> Thanks for the explanations. I should have read the explanations of the help
> text more closely. It explains it all.
>
> However, there still seem to be issues. Consider the following results
> below. First column is the result of $area, second column is
> area($geometry).
>
> Sometimes the results are very close, in other cases the difference is
> massive. Like in the very first row. 32'087 vs 190'605'589 square meters.
> This can't be explained with ellipsoidical vs. planimetric. In this first
> row, the result of area($geometry) (planimetric) seems correct, while the
> ellipsoidical result ($area) is massively wrong.
>
> How can this huge difference be explained?
>
>
> Ok - I've looked into this. It's not an issue in 2.14 or 3.0, just 2.18.
>
> It was most likely fixed by:
> https://github.com/qgis/QGIS/commit/abc0919 "Fix area calculation of
> compoundcurve rings if they contain a 2-vertex linestring"
>
> Committed to 2.14 (with unit tests), I forward ported to 3.0, but it
> hasn't been applied to 2.18. I'll do this.
>
> Nyall
>
>
>
>