Source tiles for new New Zealand DEM

[quinnhornblow]

Excited to see the new composite New Zealand collection!

Just questioning what other collections or process has be used to make this one? I expected it would include all of the individual collections but have noticed a few dependencies. For example this hole on Banks Peninsula near Kennedys reserve:

That area is covered by both the 2018-2019 Christchurch and Ashley River, and 2018-2019 Banks Peninsula DEMs, so I'm unsure why there is a hole in this new collection... am I missing something?

[dwsilk]

Hi @quinnhornblow, this has also been picked up during internal QC and will be fixed soon (I'll provide another update and close this issue when done). We were overzealous when it came to removing older surveys that were thought to be fully replaced by newer ones... Once that is sorted we will also be publishing the composite set to the LINZ Data Service.

[dwsilk]

For a bit more of an overview on the process:

basemaps.linz.govt.nz uses json documents to configure tilesets that are collections of a large number of surveys to create national services for Aerial Imagery and Elevation. This is basically a top-to-bottom list where data at the top is replaced by data referenced in the list after it (if there is overlap). Over time some surveys will be completely reflown by other surveys, and it's inefficient to continue to hold those surveys in the configuration list, so we try to keep only those that are necessary in the configs.

This new national New Zealand 1m DEM composite dataset is re-using the same configuration to build a merged GeoTIFF source instead of a basemaps.linz.govt.nz service. We have also created larger tile sizes that align to the Topo50 Map grid and included a capture-dates.geojson that references the source capture metadata.

The workflow used to make it is here if you're interested, but probably needs additional context on how that all works: https://github.com/linz/topo-workflows/blob/master/workflows/raster/national-dem.yaml (we use Argo Workflows)

[dwsilk]

Fixed by linz/basemaps-config#1031 and #340

No gap in BX24.tiff anymore:

[quinnhornblow]

Nice, thanks for looking into that. Still noticing a few small gaps in various places. I'm using the capture area geojson and looking for unexpected holes in the polygon.
There are three ~60m2 holes along the Selwyn/Christchurch boundary near Gebbies Pass. Banks Peninsula 2018-2019 seems to fill these. Another spot is in the Motueka Valley - lots of small slivers here. I don't think there is a solution though, looks like subsequent surveys kept leaving gaps.