Use global DEM instead of or in addition to SRTM

[dittaeva]

See also grundid/relation-analyzer#2

1. ASTER v2: Covers almost the whole world with high 30 metres resolution, but there are still some issues with anomalies and voids. Used successfully by turkompisen.no. Limits on distribution of original data, no limits on use and distribution on derived data that cannot be converted back to original data.
2. GMTED2010: Only one with worldwide coverage, good accuracy and quality, but best resolution is 7.5 arc-seconds (~232 m). Should be public domain. Suitable for routes of several km, but not good for (at least) sub-1 km routes.

[yvecai]

Ok, so Aster v2 it will be.
The Reverb interface is a bit more appealing than former Wist to retrieve the data, but it will take some time anyway.
I hope this can be done until the end 2012, however, Aster will need quite some processing for the hillshade layer, seems to have lot of bumps in planes.

[dittaeva]

Ah, you make your own hillshading layer! That's great, but yes it does need some processing. You know, hillshading and contour line layers for the whole world is such a requested feature that I think it could be made a project in itself that could receive support from many. I've been thinking about kickstarter, but maybe there are other better alternatives.

[yvecai]

Yes, I do so because at the time I started with pistes-nordiques.org, there was no world-coverage SRTM layers available, or none good to my personal taste.
A centralized tile-server for these purpose looks like a good idea, however the management around is over what my shoulders can bare.
If you want to start something, I can however help with the technical part. But it is beyond the issue here.

[yvecai]

Contour line layer (Isohypses) done.
The Aster dataset is not usable as is for contour lines. There is a lot of noise in the DEM, thus contour lines computed with gdalcontour are only useful in mountainous area, where the slopes are steep. In flat areas, this lead to completely useless information: 50m hills everywhere even if there is none, small hills <50m not visible.
Thus I recreated the contour database from the following sources:

• Between -56° and +60°, I used the SRTM DEM with void filled from Aster data downsampled to a factor 3. Now contours lines are renderd for Mount Everest.
  • -60° to -72° and +60° to +72°: Aster DEM downsampled by a factor 3. The contours seams to be useful in mountains, but cannot be trusted in plains.
• beyond 72°: no data. I don't have enough disk space to generate the contour lines.

Next steps:

1. Use the same datasource for the script showing elevation profile.
2. Compute the hillshade layer, either form the same datasource or from blurred Aster data, not sure yet.

[dittaeva]

I have "unlimited space" and transfer on Dreamhost, and could host there. It is shared hosting though, so not a very scalable solution I suppose.

[yvecai]

Anyway, the data are not very meaningful beyond 72°. Also, representing a
map in a web-style mercator projection does not make very much sense at
these latitudes :).
And last, if someone goes skiing up there, I would prefer he doesn't rely
on my website for is trip planning ;-)

Yves

2012/9/9 Guttorm Flatabø [email protected]

I have "unlimited space" and transfer on Dreamhost, and could host there.
It is shared hosting though, so not a very scalable solution I suppose.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/1#issuecomment-8408802.

[yvecai]

New hillshade up to 72°N is available for some weeks now.