Add alternative DEM sources (SRTM/NASADEM/ALOS) as fallback or secondary elevation data

[wilhel1812]

Context

Currently LinkSim uses Copernicus DEM (GLO-30/GLO-90) hosted on AWS S3, proxied through Cloudflare Workers. While Copernicus DEM provides excellent 30m global coverage, there are reasons to support alternative sources:

Why

1. ToS concerns: Copernicus Data Space Ecosystem terms have ambiguous restrictions on commercial use and redistribution
2. Data source diversity: Having alternatives provides resilience if Copernicus access changes
3. User choice: Some users may prefer specific datasets for their region (e.g., NASADEM has better void-filling for certain areas)
4. AWS Open Data: SRTM, NASADEM, and ALOS World 3D are Public Domain / freely usable with no commercial restrictions

Candidate Datasets (AWS Open Data)

Dataset	Resolution	License	Notes
SRTM	30m/90m	Public Domain	60N-56S coverage
NASADEM	30m	Public Domain	Improved SRTM reprocessing, global
ALOS AW3D30	30m	JAXA Free Use	Excellent global coverage
AWS Terrain Tiles	Variable	Attribution required	XYZ tiled, widely used

Research Needed

• Compare tile formats (COG vs XYZ) — current implementation uses COG, terrain tiles are XYZ
• Assess if client-side changes are needed to support XYZ tiles (vs current COG approach)
• Evaluate data quality differences per region
• Check if OpenTopography API is viable as API-only fallback (requires user API keys — likely not)
• Attribution requirements per dataset
• Storage/caching implications

Out of Scope

• Replacing Copernicus as primary source in this iteration
• Self-hosting tiles
• UI changes for dataset selection

This is a research-only issue for now. Implementation to be scoped separately.

[oddstr13]

| Geonorge/Kartverket/hoydedata.no | 1m/10m/50m | Public Domain? | Coverage: Norway. Both DOM and DTM, available via several different API types and with several formats, in addition to downloadable archives. |

Looks like GeoTIFF in the UTM33 reference grid is the most versatile.
The DOM version includes objects (buildings, trees etc) in addition to the terrain (clutter included in the dataset).
The full DTM1 dataset seems to be about 1.4TB based on the file listing, DTM10 looks much more reasonable at 20GB.

Likely best to proxy with cache if using the APIs as I've hit ratelimits quite quickly when using different maps as reference layers when editing open street map.

• https://www.kartverket.no/api-og-data/terrengdata
• https://kartkatalog.geonorge.no/metadata/kartverket/hoydedata-laser/f297e948-8a34-4e6c-9740-54b3a657f8d5
• https://hoydedata.no/LaserInnsyn2/