OpenSnowMap.org - blog

Some of you may already use the daily and weekly change view of Opensnowmap.org to check for changes on the ski pistes they mapped.
Today, this view have been improved and now gives more information about the actual change. Is this a way, a relation deleted, a simple node moved? A new processing associated with a detailed legend will help to better understand the changes and maybe spot errors.

See, yesterday apparently someone seems to have refined a nicely mapped skating ring.

This view coupled with the button will allow you to find more information about the modified element. It is still a bit frustrating not to be able to check on deleted elements, I will try to add daily and weekly .html change list to allow this.

In practice, each night, the pistes of the OpenStreetMap database at 00:00 GMT are compared to the pistes that were there the day and the week before. Geometries are simplified in order not to display to much information.

I spent a couple of hours to refresh the mobile version of Opensnowmap.org style and icons.

Let me know if you like it. Maybe I've seen the website too often myself, but I find the new one refreshing.

For the record, here was the old design:

L'Espace Nordique Jurassien partage le tracé des pistes de ski de fond depuis l'espace pro de sa carte interactive (carte dispo ici ).

Ces tracés sont d'origines diverses, et il ne fait pas sens de les copier dans OpenStreetMap. Cependant, cela permet une comparaison, et on peut voir ainsi que près de la moitié des pistes de ski de fond du Jura français ne sont pas mappés dans OSM ! Sans surprise, les coins les plus courus et les plus enneigés sont mieux cartographiés que les autres.

J'en ai fait une carte qui permet de savoir ou poser vos skis cet hiver pour compléter, c'est ici: OpenStreetMap.org et Espace Nordique Jurassien: Pistes de ski nordiques manquantes dans le Jura

Parmi les sites les plus mal servis, on trouve:

• Belleydoux
• Bolandoz
• Chaux des Prés
• Frasne
• Le Haut Saugeais Blanc - Hauterive La Fresse
• La Malmaison
• La Combe Saint Pierre
• Le Lac des Rouges Truites
• Les Combes Dernier
• Les Crozets
• Les Granges Dessus
• Longchaumois
• Le Meix Mussy
• Menthières
• Nods
• Saint Laurent en Grandvaux
• Val de Vennes

Je vous souhaite bien du plaisir à découvrir ces sites lorsque la neige sera en quantité suffisante !

I have ask myself for some time now how I could have an estimation of the completeness of OpenStreetMap when it comes to ski pistes. I’m certain that with our more than 100’000 km of pistes, downhill and cross-country combined, we are the biggest global ski pistes database of this type, but what exactly remains to map? I needed something to compare to.

There is this great site called Skimap.org where contributors can upload scans and photos of ski maps from their favorite ski area. They have more than 3000 ski area listed. These ski areas location are crowd-sourced by their contributors, and skimap.org offers an open API. So I can certainly find a way to compare this with OpenStreetMap data, no ? Yes, so I did. And many thanks to Russel for making Skimap.org !

The comparison goes both ways, first there is a long list called Pistes missing from OpenStreetMap.org.

Then we have a map showing Ski areas missing in Skimaps.org.

It’s not perfect, because there are OSM ways on one side that are not really (or rarely) grouped as ‘ski resorts’ and single points on the other side. So of course in some case the distance threshold chosen (5km) is maybe not the best, and in some case Skimap.org ski area localization could be improved.

OpenStreetMap ‘ski areas’

No, they are not built-in in the OSM database. There is the tag ‘landuse=wintersports’ to tag ski resorts, but it is seldom used. The rejected relation proposal ‘site=piste’ attracts even less contributions.

So for this comparison they are build in a database like this :

The pistes from OSM are buffered by 500m, then glued in big polygons. Very big polygons (>30km) are split in equal size rectangular chunks ranging from 20km to 40km. This is a bit arbitrary but allows to more or less efficiently separate interconnected ski areas.

In the end, this gives >3500 downhill ski areas, and the same number of nordic skiing areas.

What have I found ?

First a really good news : both projects contributors can still have a lot of fun contributing for a while. Skimaps.org excels in the USA, and OpenStreetMap.org is better in Europe.

At first glance around 800 ski areas can be found in Skimap.org but not in OpenStreetMap. For Skimap.org, a lot of small and micro downhill resorts are missing (apparently > 1500). For nordic skiing, it’s really worse, but it’s not really the target audience I guess.

Also, OSM seems more complete when it comes to lifts (aerialways) than for ski pistes themselves. I would risk an explanation here. There is a lot of gondolas enthusiasts (https://www.remontees-mecaniques.net/, http://www.funivie.org/web/, Lift-World.info https://lift-world.info/, ...), and I suspect people are mapping lifts systematically. Armchair mapping is also possible for aerialways on the contrary to ski pistes.

How does it helps ?

Both project’s data have open license, but not compatible (CC-BY-SA vs OdBL). So each can be used for the other as an inspiration, but not as a direct data source. Needless to say ski maps uploaded at Skimap.org are for the vast majority copyrighted and not be used for mapping at OSM.

For Skimap.org contributors, they can use the map produced to spot a ski area in OSM, investigate on the web and find documentation of interest for the project. For OSM contributors, well, as usual : find a place with ski pistes to map, take a train ticket, your gear, GPS and go !

More

Skimap.org also have some more information about its ski areas, for example if they are closed. This could be used to spot places in OSM where some map curating is needed. Also, I spotted on the maps some isolated pistes in OSM that looks strange. In short, some idea for ski-related QA in OSM.

Copyright notice

OpenStreetMap® is open data, licensed under the Open Data Commons Open Database License (ODbL) by the OpenStreetMap Foundation (OSMF).

Skimap.org by https://skimap.org/ is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 United States License.

It's been a few years since the last statistics about ski pistes mapping in Openstreetmap.

You know it: Openstreetmap has the biggest worldwide skimap available. And still growing !

From today stats:

• Crossountry trails: 67961 km
• Alpine runs : 37287 km
• Lifts: 17647 km
• Sled runs : 1376 km

While at a slower pace than in the glory years, the length of the ski runs mapped in OSM grow every year with several thousand of kilometers. This means that there is still some ski pistes to map, and a good question is how many ? I'm trying to have some clue in comparing other available ski areas-related listings, such as Russel's excellent Skimaps.org. I'll keep you informed.

Edit : Ah, it's more than 100'000 km of ski pistes without the lifts now !

OpenSnowMap.org renders ski ressorts at zoom 10,11 & 12. For this, it takes either site=piste relations or landuse=winter_sports polygons, and checks what type of pistes they contain (downhill, nordic, sled, etc...) in order to display small icons next to the ressort name.

Until now, these icons where pre-calculated images of all possible combinations.They were not so nice, and did not scale well with the big tiles used on the mobile site. So I built a very handy font file, called Ski.ttf, that looks like this:

This allows me to use Mapnik list-placement on TextSymbolizers to position the needed icons nicely.

It will takes some time for the tiles to re-render, so here what it looks like:

before / after comparison

When Mapquest layer disappeared, this was a problem for OpenSnowMap: its cold colors and less detailed style were great to overlay ski pistes onto. But the end of this freely available map forced me to build a simple style with plenty of room for ski pistes: the OpenSnowMap Base Snow Map.

A winter topo map

I used OSM-Bright to start building this style. Using colder colours more suited to a winter map, of course I also added relief. In particular, relief hillshading led me to desaturate highways to avoid they pops out the shadowed side of the hills . Ski pistes are not rendered on this layer, the OpenSnowMap remains an overlay. In fact, they *are* rendered on the Base Snow Map with 12 pixels wide transparent labels that forbids other labels to take place where ski pistes from the OpenSnowMap overlay will land.

I want to see it!

This style is live on opensnowmap.org. The mobile part of the website is shown with 'retina tiles' by default, but you can choose this high-dpi option or not from the menu. For these mobile-friendly tiles, I choose to render 384px tiles and display them with a 1.5x scale factor. That please my eyes on my own phone and the server seems to be OK with them, so please tell me they also looks good on your phone or tablet.

Of course, Openstreetmap-carto is still available, although without high-res tiles but a simple scaling for the mobile version. Here also let me know what the default should be: I find it more usable with the scaling on my device, although a bit pixelized.

That's great, but ...

Unfortunately, the use of the layer outside OpenSnowMap is discouraged by referer magic and so on. Also, tiles containing ski pistes are pre-rendered, but not the others. So while the Imposm-powered DB is up-to-date and a few minutes behind Openstreetmap, don't expect super-performances in tile refresh. Yes, 10 minutes so you can check your mapping mistakes and correct them before I take them into account, isn't it clever ? This new base map won't change much the OpenSnowMap overlay tiles that will remains pretty much the same and updated daily.

When you enter www.opensnowmap.org, there is a small counter that shows you the length of ski pistes and lifts there is in OSM each days. On june 6th, the total length of alpine and crosscountry ski pistes, skitour, snowshoeing, lifts and sleds pistes crossed the 100'000km line.

Not bad, isn't it ? This number seems important, so I searched and found a few flaws in my count, you can find the queries here. I notably found a few thousand of km of crosscountry pistes mapped as relations, without pistes tags on ways that were missing on previous calculations.

Anyway, now, we can claim that on 2016-06-06T00:03:02Z, there was 108'084 km of winter sports pistes in Openstreetmap ! Two and a half tour of the planet !

Before OSM, does anybody figured out that there is so many pistes dedicated to winter sports on the globe ? And how much is still to be mapped in OSM ? Why does some people map alpine ski pistes as area without a median line that would allow routing?

A new version? Wait ... nothing has changed !

Well hopefully it's the case. Opensnowmap.org website has been rewrote with a cleaner code separating HTML, CSS and Javascript. Also, a few good practises in website design have been implemented. For instance there is no blocking calls to the server any more.

There is some changes for the user, though:

• Language change can be done on the fly, no need to reload the page.
• Mobile version has been redesigned a little, allowing you to keep the map in view while looking for pistes with the 'Search' input.
• A 'Donate' button landed on the menu, it's not only here to look nice.

As the code is now cleaner, don't hesitate to propose some changes yourself on the website, it's here on Github . Be kind enough to report bugs, I made some tests on a few browsers, but there may things I missed.

Yves

The relief (hillshading) layer on OpenSnowMap.org have been updated. On the major part of the world, resolution jumps to 30m instead of 90m beforehand, thanks to the ongoing release of 1 arc-second SRTM data by NASA.

To mak a long story short, the relief is more precise by a x1.5 zoom factor.

For more details, here are the data used on the serveur to generate the hillshading.

Screenshot from USGS explorer on Google-map background

30m SRTM data are used where available. Up north, data comes from the EU-DEM for scandinavia, ASTER for Alaska, Canada, Greenland and Russia. In an area extending from Greece to Iran, I had to kept the 90m SRTM data from CGIAR, I hoipe to be able to update this area soon.

To assemble those datasets, I use a script based on python-gdal to merge the data by small 1x1° tiles, filtering SRTM data by a 3x3 average filter and ASTER by a 10x10 filter to reduce noise. Gdal utilities then re-assemble everything in a big Tiff. For rendering, I took the opportunity to pré-calculate smaller Tiff at lower zooms to reduce the load on the tile server. The map should show up more fluidly.

Contour lines will be updated from the same datasets, if I can optimize them enough to fit my disks.