The Atlas Mountains of Morocco are covered with paths, well-trodden but only barely visible unless you’re a shepherd who knows the way. I spent two weekends traversing some demanding terrain in these mountains, by recommendation of Alexandre from Sige, Mapillary’s local partner in Morocco. The journey turned out to be not only an adventurous exploration of a new country but also a learning experience on using street-level imagery for editing OpenStreetMap to add details such as:

• remote trails and pathways,
• road and pavement types,
• bridges,
• building and construction types,
• streams and waterfalls,
• marshlands, cliffs and other natural features,
• local businesses and related info.

My planned route was to cover three different mountain refuges: Refuge Tamsoult, Refuge Tazarhart (Jacques Lepiney), and Refuge du Toubkal (Louis Neltner). The latter two were founded by the Club Alpin Français, and have a local name plus an attached name of a French mountaineer.

While Toubkal is the most famous lodge, and most commercial, it stands in contrast to the quieter and less frequented Tamsoult location, which sits at lower elevation above a green valley. Tazarhart, meanwhile, is a small building with a sort of French alpine style of architecture, and doesn’t have staff who cook for and assist you upon arrival. All three can be explored on the same route. Additionally, all of them are visible on OpenStreetMap, along with the trails to reach them.

To trail or not to trail?

My first weekend went rather smoothly, and I didn’t add too many new sequences to the already existing 360-degree imagery of places like Refuge Tazarhart. However, I did attempt to take a shortcut, where two paths seemed to have a gap between where they connected. This inspired me to make some updates to OpenStreetMap.

In the map above, you can see the area called Tizi n’Tadat. It stands between trails on the west and east sides, and it is classified as bare rock. I added this to the map after approaching it in a grueling fashion; starting on a mellow trail at Refuge Tazarhart and eventually finding I had to duck under boulders and start wondering if I should have brought rope, a helmet, and a partner. I decided it was impossible for trekking, and was more of a mountaineering route. Indeed, it is exposed rock that goes to rather high elevation, offering a fast way to Refuge Toubkal along with a sense of danger.

To climb back down to the Refuge Tazarhart required some careful placement of hands and feet, so I left my phone in my backpack and didn’t capture any photos. However, using MAPS.ME to navigate, I could see that OpenStreetMap was clearly correct about the two paths failing to connect, despite my assumption that perhaps I could traverse the gap. My OSM edits as a result will hopefully help bring more context to others.

Paving the way

The following weekend, Mapillary was at the front of my mind as I started another long trek recommended by Alexandre. The route would start in the north, at Lraba Tirdouine (also called Tighdouine), and reach about 18 km into the Atlas Mountains where several villages awaited. I was advised to simply ask around for hospitality on arrival.

After a night in the high villages, the second day would take me a further 15 km or so down to Ourika Valley in the south, by way of Tizi (or Tizi n’Oucheg). There was no path on OpenStreetMap from any of the central villages down to Tizi, so I intended to map this myself. In general, there was a great lack of OSM features anywhere in these highlands, so Mapillary was to serve as a tool for me to record what I saw and later reference it in my OSM edits.

To begin with, I noticed that the roads near Tighdouine, on which I began trekking, were recorded in OSM as Unmaintained Track Road, which OSM defines as for agricultural and forestry use. This maybe seemed accurate to me, as far as the use case, but also the road appeared to be paved (although loose dirt tended to end up on top of it).

In fact, the pavement was very high quality, and certainly had been maintained. A car would pass me once every 20 mins, comfortably driving over 60 km/h. I started recording parts of the road with Mapillary in order to provide a reference for the road condition in my own edits, as well as evidence for anyone who wanted to double check.

As far as I could tell, Minor/Unclassified Road appeared to be the best fit for this route, especially considering the reference photo provided by OSM. I could also provide evidence for marking it as paved. After my hike, I was able to make these edits and have a strong visual reminder.

As I continued walking the road, I took more Mapillary sequences at points where the road type changed, or where other notable objects appeared. I didn’t want to record the entire route nonstop because even with a backup battery I wouldn’t have enough charge to use the maps on my phone over the course of the two-day adventure. In this case, it was best to be strategic about where I captured images. In another example further down the road, I walked over a concrete bridge, and started capturing photos to remind myself. In OSM, I was able to split this road section off and mark it as a bridge.

Later down the road, I started capturing Mapillary images again where the road changed from being paved to unpaved. It was still quite maintained, as it had no potholes and seemed like heavy equipment had been used to ensure it was smooth. I left the type as Minor/Unclassified Road and changed the surface to be unpaved at this point, by splitting the line.

An hour’s walk up the road, I walked through a village, and found that afterward the road truly became an unmaintained track. I captured images once again and marked this on OSM where my Mapillary sequences served as an indicator.

Discovering local knowledge

The villages in this region tend to have brightly colored paint on the walls around school grounds, plus a Moroccan flag in the schoolyard. I captured a school with Mapillary as I walked past, and was also able to add it to OSM later. The satellite images helped me see the building outline, but there wasn’t any indicator that it was a school without the street-level imagery as context.

This school marked the beginning of the last village where I could seek hospitality. It was a bit nerve wracking at first, but everyone in the village noticed me coming and several of the local kids were quick to strike up a conversation in any language possible. While they knew very little of English or French, we found we could communicate some in my broken Arabic.

They showed me around the village, asked how I was enjoying Morocco and the local foods, and were curious what brought me out to their remote village. I also attempted to ask them for directions to the next village—Tizi n’Oucheg—but they didn’t seem to know much about what was over the next valley. This kept me on edge, as I also didn’t have a map of the trail.

I went past the village to the next ridge to scout the route for the next day’s hike, and couldn’t see much. Gazing upon a vast canyon, I attempted to hike along the north side, keeping to high ground. Across the canyon, however, I could hear the bleeting of goats—squinting, I could also see the microscopic figure of a shepherd moving along a path atop the canyon walls.

I was amazed that such a path existed, and soon realized it would carry me all the way over the mountain pass that was at the west end of the canyon. My work was cut out for me. Back in the village, I asked around more for hospitality, and was quickly invited into a family’s home for a dinner of bread, fig preserves, honey, olive oil, and tea. I slept soundly in a Berber hut, and was awoken at sunrise with another meal that was just as hardy.

Shelters in the wild

I started my journey through the central highland, which was truly pristine nature. I marked several locations on the map where streams and waterfalls crossed my path, and opened Mapillary to capture images of the trail condition (where I could even see the trail!).

As I made the climb uphill to the final mountain pass before the next village, I saw cattle grazing on green grass, which stood out strikingly against the red earth. Moving closer to investigate, my feet started to sink, and I realized the verdant surface was actually something of a marsh because of springwater soaking the ground. I captured some Mapillary images of the marsh, and used satellite imagery to mark the boundary of this area.

I hit the highest point of the pass shortly after this marsh, which seemed to coincide with the location of a freshwater spring flowing both to the east, where I had just been, and to the west. Shepherds must have seen this as a great reason to make a basecamp nearby, for as I descended to the west I suddenly noticed stone huts built into the mountainside. Without pausing to take a look, it could be easy to miss their presence, as they blended in with the landscape.

I approached the huts, and found that around every turn, behind every gargantuan boulder, there was another and another. These are very difficult to see using satellite imagery, but combining Mapillary and the satellite views, I was able to demarcate a few of them and also add them to OSM.

Some of them seemed to have low walls served as a sheep and goat corral, plus a structure for sleeping. In a remote area such as this, a trekker may find it useful to know that shelter is here—maybe not for casual use, but with the common afternoon thunderstorms it could be that the Berber hospitality includes offering shelter from the weather in these huts.

Some of the areas surrounded by low walls were also easy to mark, using the wall key on OSM. It appears that these are used either as wind shelter for sitting and resting, or as mentioned, some kind of corral for goats and sheep, although with lower walls than the ones near the huts.

Back to the city

Toward the end of the journey, I began to approach the highway where I could catch a bus back to my accomodation in Marrakech. As I descended from the high plateau around the village of Tizi n’Oucheg, I noticed that the dirt track road had returned, but also that sections of it were paved—primarily on the switchbacks, where perhaps winter snowmelt caused erosion of the roads.

I was able to mark where the paved sections started and ended, usually in sections of no more than 100 meters, and use this as reference once again on OSM. This could come in handy later on for someone assessing road quality and planning any future improvements. I left the type as Unmaintained Track Road and changed the surface type to paved for these sections.

Overall, the journey went very smoothly for me and yielded quite a bit of map data to add to OSM—plenty of buildings, streams, and natural features such as cliffs. Back in Marrakech, I also captured several Mapillary sequences that were useful because of the density of local businesses to be found in many images.

While the old city (the Medina) of Marrakech and interferes with GPS signals, the more modern district (Gueliz) is far more GPS friendly and has plenty of signs over shops that help to add them to OSM when captured with Mapillary. For example, see the sequence below that I took facing sideways out of a taxi.

From the image above we can clearly add Labo Photo, Atlas Voyages, Cash Plus, and Haouz Annonces to the map. I’m not sure what the last one is—something about advertising, perhaps—but I can clearly see their website in the image, so I checked it out! They actually own the .info domain and not .com, so I added that to their OSM location and the category of Shop.

This is all from a single photo. 20 seconds of driving can get me 20 photos, which probably yields 20 to 60 businesses I can add to OSM. All of this I could do on my computer from the comfort of the co-working space down the street, instead of requiring me to stop by the shops and add them to the map on my mobile phone, one-by-one.

From street level to OpenStreetMap

Mapillary serves as a powerful tool for editing OpenStreetMap. It doesn’t compete with other resources for OSM, such as high resolution satellite imagery or mobile apps for mapping on the go, but complements them instead.

Images provide concrete evidence for what’s on the ground, in a way that even human-confirmed map data can’t quite provide. Reviewing other contributors’ edits on OSM is easy using Mapillary, as is getting plenty of detail for your own edits. In my own travels, I have started to take more interest in my surroundings, as I look at capturing not just beautiful landscapes but also important pieces of infrastructure, names of places, and conditions of the land.

To learn more about using Mapillary to edit OpenStreetMap, check out this great video tutorial from our ambassador Nuno Caldeira. Be sure to keep on eye on our world map as we continue to expand coverage, with over 235 million images on the map today. Finally, if you’re not a contributor already, you can help map your local community by signing up for Mapillary today, and getting started on mapping with your mobile phone!

/Chris