sotl.as AZ outlines for US+UK

[AD8IS]

Greetings all,

I’m working on a PhD involving LIDAR and geodetics. In an effort to overlap my study and hobby, I’ve learned a lot about Digital Elevation Models (DEM). I refactored code from WA7BEN/Ben and N0QLR/Shane, to generate activation zone contours from high-precision DEM. It is efficient and robust to outliers, so I’m able to process large areas.

The results are uploaded to sotl.as under each summit’s page. I completed the US (including Alaska!) and have recently published summits for G, GW, and some for GM. The AZs are represented by a light green, dashed area. There is an “AZ” button at the top to download the data.

You can review my open-source code here. AI was used in development to ensure code quality; iterations with the LLM were short and based on my manual research and test cases.

Interestingly, the code ran over the entire UK with no reported errors, owing to excellent curation of those summits by the Association Managers - kudos! Alaska was much more complicated, with moving glaciers, high-latitude reprojection errors, and more. No DEM was available for much of Scotland; only 12% of summits are outlined as a result. I did not fall back to lower-resolution DEM. The minimum I accepted is 1m horizontal, 10cm vertical unvegetated precision at p=0.95. In vegetated areas, this is approximately a 95th percentile within 30cm vertical. No data is preferable to bad data!

Please have a look at your favorite summits. If you find bad data in excess of the stated precision in the US or UK, I’m happy to look into it if you just let me know. HB9DQM/Manuel (of sotl.as) can also direct these bug reports as appropriate to other curators in other areas. So far, six reports of bad data have revealed coordinate transcription errors in the official database, and terrain changes due to recent mining operations.

These AZ outlines are for your convenience in planning. For example: it was convenient for me to query that there are five summits in England and Wales whose AZs contain a pub. I am happy to provide that list, and buy the first round, in exchange for a plane ticket!

However - the AZs are unofficial. The activator on the ground must be the authority on the matter. As with Archimedes’ lever, an activator with a shovel and enough gusto can activate any summit from anywhere!

I also must discourage use of this methodology for measuring prominence or disqualifying summits. It is statistically difficult to disprove a prominence claim using this DEM, due to the nature of confidence intervals and lack of Gaussian point distribution in vegetation. AA3TZ/Tom and I wrote some on the subject. The extent of my project is to outline the existing summits.

Enjoy - I hope this is useful to you!

AD8IS/Mike

[GM5ALX]

Oh very nice!

@AD8IS

No DEM was available for much of Scotland; only 12% of summits are outlined as a result.

I started wondering which 12% and browsed the map randomly but I couldn’t tell what the reasoning is for the ones in vs. out. Then I see there’s a list in the git repo, so I can see what’s what there. Although still not sure what the reason for why some do and some don’t have DEM data…but I suppose people collect it for specific reasons only.

[AD8IS]

I have that data!

Region	Created	Active summits	Coverage
CS	3	124	2.4%
ES	8	87	9.2%
NS	2	155	1.3%
SI	48	223	21.5%
SS	73	283	25.8%
WS	12	347	3.5%
GM overall	146	1,219	12.0%

As for the reasoning - this high-precision LIDAR is collected from small aircraft flying at low altitudes. It is expensive to recalibrate the LIDAR and refuel the plane for large area coverage. In rural areas, 1m precision DEM doesn’t have a use case; 30m precision (from satellites) is sufficient for public purposes. So it’s an efficient use of government resources to leave the areas blank until someone expresses a demand for the data. Note that 1m (and better) commercial data exist, but I won’t be the one to write those grants

[F4LEK]

Nice !

I think France has such a DEM covering all the country.

[MM1ROS]

@AD8IS

In rural areas, 1m precision DEM doesn’t have a use case

That would certainly explain the difference in coverage between SS, where most of the population in Scotland lives, and NS with probably the lowest population density. While that loosely aligns with what I’d expect to see on the mainland, I am surprised at how comparatively high the number is for the Scottish Islands (SI).

[G5JFJ]

This is great, thanks. Good to check what the tool thinks is the AZ for tomorrow and I can compare with OS map/altimeter

@MM1ROS

I am surprised at how comparatively high the number is for the Scottish Islands (SI).

Just guessing but perhaps something to do with coastline monitoring?

[AD8IS]

Thank you for letting me know! I’ve published AZs for France, and Manuel has put them on sotl.as.

I do notice some data quality issues in the elevation model. It appears to be interpolated from coarser data in places. I decided to publish as-is; if you become aware of better data, please let me know or feel free to adapt my code to work better, and send the results to Manuel.

-AD8IS/Mike

[F4LEK]

Ohhh fantastic ! Thank you.

I don’t really see what’s the issue with the example you showed, I’m a bit confused.

If you are refering to the shape of the AZ and the contour lines of the sotlas map, I think the basemap used for sotlas is not very representative of the reality sometimes. It’s quite coarse.

Here is the IGN official 1/25000 map, with the same AZ :

Both the AZ from your program and the IGN map match perfectly with what I experienced from my very recent activation of this summit.

[AD8IS]

I’m glad you like it - and I’m happy to report I found the issue.

I had requested the data advertised by IGN as “High Resolution” in my chosen projection; I got the highest resolution available which was approx 3-5m but I then treated it as 1m precision. Instead, I am now requesting the 1m-precision data in their chosen projection, and reprojecting it myself.

See how the contour now looks smooth, and traces every meter of terrain? I knew this summit would be a good way to get your input Congrats on the first activation!

I will regenerate the whole country, to be safe. I am also working on Japan now, since it uses a similar data source.

-AD8IS/Mike

[F4LEK]

Oh I see. Always better smoother

However, I would question the significance of such a precision and smoothness. On this summit for example, it’s full of holes and large boulders (more than 1 m) all together on a cliffy ridge. So I’m not sure 1m precision is really significant, in a scientific point of view.

[AD8IS]

It is very interesting to consider the Coastline Paradox in situations like this. The more precise our models become, the longer the AZ perimeter will get, as it follows every nook and cranny.

LIDAR point clouds capture vegetation and rocks and other occlusions, but in 1m-precision elevation models, the “deepest” point returns in each square meter are recorded as the surface elevation. Since the rocks have gaps between them at least every meter, the surface beneath is recorded.

Since the outline is smooth, we know we’re not tracing around individual rocks. The terrain frequency is still at least double our quantization rate, so the outline is aliasing the rock structure. Thanks be to Harry Nyquist!

My point: there is a strictly scientific limit to this precision, and in this case, we haven’t reached it.

I choose to believe that’s why the ASPRS 10cm vertical accuracy class (which is the standard used for most 1m tile DEMs) is well-received globally. It’s right at the scale of natural human movement.

[HB9DQM]

Thanks, Mike (and all the others who have contributed), for the new AZ polygons – it’s really great to see the momentum that has built around this topic!

On a related note…

Aside from their use in determining the exact extent of an activation zone, I find it remarkable how DEMs, and especially their derivative, the slope angle map, have changed and influenced the way I (and many others) plan summit ascents.

For many summits, even with no route description, just looking at the colored “slope classes over 30°” overlay provided by swisstopo, I can often tell immediately what the most likely/reasonable routes are. Combined with information about the terrain type (grass, scree, ice etc.) given by the map, I can draw a track with what I perceive to be the most practical route before I leave for an activation, and then usually just follow that track in the field. It is rare that I find a better route once I’m on the ground. Of course it requires accurate and detailed maps, which we are blessed to have in our small country.

Example of a summit that I recently activated, and about which no information was available on the web (just a very brief route description in an old book by the Swiss Alpine Club, accompanied by the note ‘arduous, very rarely climbed’) – planned route in red, actual GPS track in green:

Where this fails, of course, is on narrow ridges, or in very steep, ragged terrain, where a small ledge half a meter wide can make all the difference. Very narrow couloirs are also difficult to judge given the resolution (10m) of the swisstopo slope class map.

As an experiment, I loaded a 0.5m resolution DEM of a savage summit into QGIS, combined it with a topographic landscape model, assigned a compound “cost” value based on steepness and terrain type (e.g., higher cost for scree or steep ice), and ran Dijkstra’s algorithm to find the least-cost path. The results were interesting, but not the ultimate solution for that particular summit. It turns out that once you exceed around 45° in steepness, what matters more than the precise slope angle is how well the terrain is structured. For example, a vertical wall (which we don’t have any information on in a regular DEM created from directly above anyway) can be quite easy or extremely difficult to climb, depending on the availability of hand- and footholds.

But I wonder what might be possible in the future with even more accurate DEMs? Perhaps we can have AI find new routes to summits?

Anyway, I think it would be very valuable to have slope class maps for other associations too. They’re also a great tool for assessing avalanche risk and planning a conservative route in winter.

[G4TGJ]

These sotl.as AZ outlines are really interesting. Looking at my local summit G/NP-028 I can see that it extends the AZ beyond Keighley Road. I have always wondered if the Buck Stones were in the AZ but had always thought they probably weren’t. The summit is 402m and the contours either side of the col are 380m so I had no way of knowing if it dipped an additional 3m - on the ground it looks like it might. So do I accept the sotl.as AZ as correct or do I have to continue to go near the actual summit just to be sure my activations are valid?

I originally asked the question almost exactly 8 years ago:

Rombalds Moor (G/NP-028) Activation Zone

I’m trying to work out the extent of the activation zone on Rombalds Moor, in particular whether the high ground by the Buck Stones is included. The summit is 402m so the activation zone is formed by the 377m contour. The OS maps show two separate 380m contours around the trig point and around the Buck Stones top. So it isn’t possible to tell if the top is in the AZ as I can’t tell how low the col is. But the Open Topo Map (linked from the summit info page) shows a single 380m contour enclosing …

[MW0PDV]

This is a fantastic tool, and having had a play with just UK activation zones, can appreciate the effort that has gone into this.

@G4TGJ

So do I accept the sotl.as AZ as correct or do I have to continue to go near the actual summit just to be sure my activations are valid?

I took cautious approach for the UK, using the OS DEM with a resolution of 10m for elevation. I round up to the 10m contour nearest the summit. This way the shown activation zone will always be within the activation zone. Although the actual activation zone could be larger.
This is what I get for G/NP-028

But just to add you should always check the activation zone for yourself, as my tool is also only a guide.

[GM5ALX]

@G4TGJ

So do I accept the sotl.as AZ as correct or do I have to continue to go near the actual summit just to be sure my activations are valid?

Given the marginality of the uncertainty and the fact you have to climb ~15m to get to the Buck stones vs. the historically acceptable 3-5m climb to the wireless station to be in the AZ, I’d say you are good to go! Especially if the trig is busy and you can keep out the way. Of course, my opinion is worthless.

[M0LEP]

@MW0PDV

This way the shown activation zone will always be within the activation zone. Although the actual activation zone could be larger.

15 years or so ago I took the then-current OS 1:25,000 paper map and tried to figure the extent of the activation area for my then closest summit G/SE-005 Botley Hill using the 250 metre contour. The area shown on sotl.as does seem to stray quite a way from the OS contours in places, and there are certainly parts where one approach puts a point within the activation area while the other puts the same point outside, both ways round. Who’s to say which approach is more accurate…

[G4TGJ]

@GM5ALX

you have to climb ~15m to get to the Buck stones vs. the historically acceptable 3-5m climb to the wireless station to be in the AZ

I’m not sure I understand the logic of your argument. Whether or not a point is in the AZ has nothing to do with the car park which, in this case, is in the AZ anyway.

I have never parked in the car park. I live in Ilkley so it would be a long drive (without a 4WD). I have always walked from home. My interest in the Buck Stones is because it would allow me to activate the summit from a completely different spot and approach it a different way.

@GM5ALX

if the trig is busy and you can keep out the way

The AZ is so large it’s easy to find a quiet spot. I always activate a short distance from the trig point. I see few people walk past even though the main path is quite popular.

What I’m after is someone from the MT saying that they consider the data used by sotl.as to be accurate enough that I can rely on its AZ maps, at least in this case.

[AD8IS]

The LIDAR methods are validated to 30cm of vegetated vertical accuracy (VVA) on the 95th percentile of data, and 1m of absolute horizontal accuracy. “Who’s to say” - ASPRS, generally, and in this case the UK Environment Agency.

A 250-meter contour line is a big quantization. Due to the Nyquist limit, it cannot capture terrain changes finer than twice its frequency. Linear interpolation of a 25-meter AZ contour is 14dB below that limit! Like trying to read a license plate with my glasses off - the resolution just isn’t there, and no amount of squinting will bring forth the information.

I will repeat that these AZ outlines are in no way official. Use them for convenience. The operator is ultimately responsible for making sure they’re inside the real AZ on the ground, on the day of the activation. No maps or survey can excuse us from this responsibility.

[G4AZS]

There are some suggestions for on-site checking of AZs further down this thread, which might suit Richard’s use case:

Activation zone boundary annotation on maps

Is there somewhere on-line where one can view the Activation Zone (AZ) boundary for UK SOTA summits drawn on an OS Map [or equivalent]? Or does one need a special software tool / app to produce the annotation? It’s not something you can work out by hand as contour lines are at 10m intervals. If say, the summit is at 159m ASL then the AZ boundary runs along the [unshown] 134m contour. Until now it’s never been an issue for me because I’ve either activated at or close to the summit [usually the …

[F4LAD]

Hello Mike, thank you for this nice addition to sotl.as!

If you are using the RGE ALTI dataset from the French IGN, it is using elevation data compiled from heterogeneous sources, including old LIDAR surveys with only 1 m vertical accuracy, and updates to this layer have been on hold since 2024 (see the official dataset page, in French: Jeu de données - RGE ALTI® | data.gouv.fr). Data from the recent LiDAR HD campaigns, which offer ≤10 cm vertical accuracy, has not yet been incorporated into this layer. National LiDAR HD coverage is about 80% as of late 2025, with full coverage expected by the end of 2026.