Make it possible to store NaN values for Z values

[Nico_Burgerhart_Sweco]

Make it possible to store NaN values for Z values. This is possible for M values.

If this is possible for Z values, there can be made a difference between a Z value 0 and a Z value which is unknown.

The documentation says a Z value can be set to NaN but this isnt't true.

https://pro.arcgis.com/en/pro-app/latest/help/editing/edit-vertex-z-values.htm

See also BUG-000152196 Clearing a z-value results in zero, instead of NaN

[KoryKramer]

Thank you for submitting this idea @Nico_Burgerhart_Sweco It looks like that documentation is incorrect and will be updated with the Pro 3.1 release to remove the sentence about setting Z to NaN.

So I think that this idea is that you want to be able to store <Null> values for Z. If you have z-enabled data, can you give some more details about your workflow that requires storing null z values? This will be great to help us understand why. 

(Also, see https://community.esri.com/t5/arcgis-pro-ideas/store-blank-m-values-as-null-not-negative-infinity/id.... It appears that m values are not actually stored as null/unknown).

Thank you!

[Nico_Burgerhart_Sweco]

I want to merge two datasets, one dataset having Z values an and one not. For geometries from the second dataset I want to set the Z-values to NaN. At this moment they get a Z = 0 (they are merged in a FME ETL process). Z = 0 is an existing value in the vertical reference system (NAP = EPSG 5709) we use in the Netherlands.

[KoryKramer]

So when you have the final dataset that is the result of merging two datasets, you will have a z-enabled dataset, with some of the features having valid z values and other features in the same dataset having null z values. Correct?

Do you draw that data only in 2D? Regardless of z, you would see the features drawn using the x,y values. If this data were to be drawn in 3D, what is the expectation? With no z values, those features should not draw? Interested in learning more about this.

[Nico_Burgerhart_Sweco]

I realized me also the problem of drawing the features in 3D (also because in one polygon feauture there could be vertices with and without a Z-value present). I've no solution for this.

For a point dataset I could imagine just not to draw points with an unknown Z-value.

An alternative could be the possibility to set a NoData value (e.g. a value outside of the 'normal' domain of the data like 99999, see also https://www.ian-ko.com/ET_GeoWizards/UserGuide/getZCharacteristics.htm) and have the possibility to exclude this value in visualisation of the data and in calculations.

In the original situation of combining datasets with and without Z-values, we try to change the process and assign a Z-value to each vertex (by interpolating, using an external elevation source, and removing vertices without Z-value).

[KoryKramer]

Thanks, Nico.

[Bud]

Edited Aug 16, 2023

It'll be strange to have M-values stored as NaN and Z as null. To me, M and Z dimensions are similar concepts (additional dimensions to X and Y) and so I would have expected blank values to be consistent: both Nan or both null.

Related: 

• Documentation about empty strings, nulls, and NaN in ArcGIS
• Store blank M-values as null, not negative infinity/NaN