Hi,

the hillshading layer that can be added to the OpenStreetMap map in Marble only covers a part of the globe. (It covers most of Europe up to about 60° 30′ N, some parts of the Middle East and North Africa north of 30° N, the US, Canada mostly up to 55 °N, the Americas south of the US up to the Equator.) Could this be extended? As relief maps based on OSM show, the data seems to be freely available for the whole Earth (but I haven’t investigated any details).

Hi,

In Marble we have the policy not to create/edit data ourselves (and we rarely make exceptions there).
So the data we use is a compilation of existing data and services.

For the hillshading we are using

http://hikebikemap.org/

So if you'd like see more coverage of the hillshading layer then please ask the author of the hikebikemap server

Best Regards,
Torsten

tackat wrote:Hi,
For the hillshading we are using

http://hikebikemap.org/

So if you'd like see more coverage of the hillshading layer then please ask the author of the hikebikemap server :-)

Couldn’t you try substituting this with another data source with global coverage? As I tried to indicate earlier, I assume that such a source must exist. OpenTopoMap seems to have one for example. Also, this is quite an striking gap in Marble’s data so I’m surprised that this source was deemed acceptable in the first place. Anyway, thanks for the response, and please indicate if there really is no better way than contacting the hikebikemap authors, which I would be willing to do then.

> Couldn’t you try substituting this with another data source with global coverage?

If somebody comes up with a viable suggestion we'll surely trade the current hillshading for a better option
However so far nobody has suggested a better source.
While the data itself is most likely based on the free SRTM DEM it requires somebody to compile and distribute it

> As I tried to indicate earlier, I assume that such a source must exist. OpenTopoMap seems to have one for example.

... which reminds me that it would be great if we had OpenTopoMap as a Marble map theme

Unfortunately in OpenTopoMap the hillshading is not separated from the topography.
We'd need it in a separate layer on a server where we have the promise that the data will still be there in the upcoming future.

> Also, this is quite an striking gap in Marble’s data so I’m surprised that this source was deemed acceptable in the first place.

Yes, we have pretty high quality standards in the Marble team (also especially with regard to licenses involved).
However at some point we have to work with what we get.
In this case the hillshading is available for much of the most populated areas - which we have considered good enough.
It would be awesome if we got better data. However this requires somebody to look into it
We'd be glad about any help with compiling data. That part is actually much easier than doing the C++ programming.
But it's still something that needs to be done by someone

The whole Marble project is a community project - so any help is welcome

tackat wrote:>
... which reminds me that it would be great if we had OpenTopoMap as a Marble map theme

It's not directly related to your hillshading request but I've just uploaded a new map theme which features OpenTopoMap.

You should be able to access it via

File->Download Maps

inside Marble. See also

https://marble.kde.org/maps-4.5.php

Have fun.

For the curious I've also created an illustrated step-by-step tutorial for creating and modifying map themes such as the OpenStreetMap, Hike&Bike or OpenTopoMap map theme.
It can be done with a simple text editor. And here's how this works:

https://techbase.kde.org/Marble/OSMSlippyMaps

tackat wrote:

tackat wrote:>
... which reminds me that it would be great if we had OpenTopoMap as a Marble map theme :-)

It's not directly related to your hillshading request but I've just uploaded a new map theme which features OpenTopoMap.

Thank you, also for the tutorial. This is a major step up from the maps for ÖPNVKarte and OpenTopoMap that I had created manually using the menu with very little knowledge what the further wizards steps really can do. :-) I will have to study the tutorial in detail some time.

I did make some more adjustments to your OpenTopoMap DGML file though. First, I adjusted the maximum zoom value because OpenTopoMap doesn't scale that far. Ideally I would have liked zooming to stop exactly at the level where the most detailed layer would be loaded, but I found that the value for that would have had to be somewhere between 2911 and 2912 and fractions did not seem to be supported. Now the maximum possible zoom is slightly "beyond" that last layer so that using the mouse to zoom in to the maximum gives a somewhat blurred view of that layer and pressing "-" gets you back to the same layer but without blurring. I hope that these values are not system-specific anyway. Second, I added some options for the layers that are offered in "Atlas" (would have liked to also add boundaries but I wasn't able to find out how to include them).

The result is at http://pastebin.com/TmFaw4T9 and maybe you will want to use it in the version you distribute.

tackat wrote:

sebalis wrote:Couldn’t you try substituting this with another data source with global coverage?

If somebody comes up with a viable suggestion we'll surely trade the current hillshading for a better option :-)
However so far nobody has suggested a better source.
While the data itself is most likely based on the free SRTM DEM it requires somebody to compile and distribute it :-)

[...]

We'd need it in a separate layer on a server where we have the promise that the data will still be there in the upcoming future.

[...]

It would be awesome if we got better data. However this requires somebody to look into it :-)
We'd be glad about any help with compiling data. That part is actually much easier than doing the C++ programming.
But it's still something that needs to be done by someone :)

The whole Marble project is a community project - so any help is welcome :-)

As a first attempt to help I tried to get in touch with HikeBikeMap and was able to establish contact with its maker Colin Marquardt, who has kindly sent me his thoughts (in German) and allowed me to summarise them here. He also said he will look at this thread after the weekend. I hope I'm not misrepresenting what he wrote:

All Colin had at the time was the low-resolution STRM data, whose coverage ends at a certain N latitude (and beyond that there's only ASTER which has a different, somewhat more restrictive licence). It does require a significant amount of storage space which was kindly offered by Wikimedia. The setup for generating the files used to be on a Wikimedia toolserver and was later moved to Wikimedia Labs. The Labs setup is not simple and the tools used then aren't all available now. They would have to be modernised anyway, Colin does not have the time to involve himself in that now.

Another interesting source that has appeared now is clusterGIS, announced in https://lists.openstreetmap.org/piperma ... 75717.html. They host GeoTIFF files that would have to be downloaded, converted to PNG (according to http://wiki.openstreetmap.org/wiki/Hill ... f_an_Image or the current state of the art would need to be researched) and hosted somewhere. (I [sebalis] looked at http://www.clustergis.org/ and its download site http://www.theearthsrelief.com/ and found that it too does not include latitudes above about 60° N nor the Antarctic. The licence is CC BY 3.0.) Colin could ask Wikimedia for donating the space and connectivity but he is not entirely sure how to initiate this or whether at some point they might decide they need to make savings. He did have some other leads for possibly finding a donor but also wondered whether KDE/Marble could provide hosting.

Further places to look at, according to Colin, could be a tool to download, merge and generate tiles from digital elevation data (https://github.com/mapzen/joerd), a page on Advanced Relief Shading using GDAL (http://giswerk.org/doku.php?id=de:r:r-m ... adedrelief), and a transparent hillshade cooking guide using GDAL (https://gist.github.com/cquest/8179870).

Re storage, how much would be needed roughly?

> The result is at http://pastebin.com/TmFaw4T9 and maybe you will want to use it in the version you distribute.

Thanks a lot!
I had also been wondering about the maximum zoom level - however it turned out that the maximum zoomlevel in the map seems to vary (it seems to be higher in the alps). Also I don't know whether there will be updates to the maximum tile level - so kept the current setting for now.

Originally I disabled the placemarks on purpose since they duplicated lots of labels. However you added really nice checkboxes for the Continents and polulated places. So I cherry-picked the checkboxes you added from your diff - but set the populated places to false by default (which still leaves the option to the user to turn them on - which makes a lot of sense for areas that have no labels yet at all.

I also did a few minor improvements to the legend.
As a result I uploaded version 0.2 of the OpenTopoMap to the server.

Again thanks for your feedback and patch - that's exactly the kind of community work that we are looking for.

Earthwings wrote:Re storage, how much would be needed roughly?

Difficult to answer of course, for everyone including Colin Marquardt I guess since we haven't exactly defined what data and formats we need. But for a first approximation I downloaded two tiles from http://www.theearthsrelief.com/ (at max. resolution, 15 m / pixel) and used GIMP to convert them from TIFF (I found that GeoTIFF is TIFF plus extra metadata which I chose to ignore) to PNGs with an alpha layer and saved them, trying to minimise disk space (RGB → greyscale, max. compression, no meta information saved). All tiles cover an area of 1° longitude × 1° latitude (i.e. a graticule, a term I take from http://wiki.xkcd.com/geohashing/Main_Page) and their sizes are 13 and 15 MB. So let's assume 15 MB per tile. The data set goes from 56° S to 60° N and covers all longitudes, so that's (60+56) · 360 = 41760 graticules. I then took a screenshot of the world map on the download site, scaled down so that each graticule was approx. 1 pixel in size, and used GIMP to select and count the blue pixels. Through GIMP's histogram I found that about 40 % of the graticules are blue. That gives us about 16.700 tiles. At 15 MB each that yields an estimate of 250 GB of PNG tiles. This is just for the highest resolution, I don't know how many layers with lower resolution you would want to add. Also it would of course be great to add the missing latitudes, and there is a question whether the oceans should be covered as well. Both these additions would require a different data source of course and more processing than just converting from TIFF to PNG.