Crowdsourcing a global terrain map

OpenStreetMap is a crowdsourced online map that's an alternative to the commercial maps like Google Maps.The map was built by users who used devices with GPS receivers to record tracks (a series of points that the device covered) which were then used as a basis for a map. This was necessary because cartographers can't use existing maps to create a new map, they'd either have to license the existing map or be liable for copyright infringement. By creating a map from the ground up, OpenStreetMap is free to do what they want with the map and people are free to use the map with fewer restrictions than a commercial map.

There's one big thing that's missing form all online maps and that's detailed elevation data. People can not currently source detailed elevation data even if they were willing to risk breaking copyright and that information isn't available in a format that's easy to adapt for other projects.

 Why Vertical Accuracy Is Important  

Clearly everyone would like any tool to be as accurate as possible but it's even more important for vertical measurements than horizontal. If a horizontal measurement is off by, say, 10m it's good enough for navigation purposes even if it's not ideal, however 10m vertically can make a massive difference to the appearance. Imagine a hill which is surveyed at three points, the bottom (0m), the middle (10m) and the top (20m). Now imagine that the first point is recorded as 10m (+10m) the middle as 10m (+0m) and the top as 10m (-10m). That 10m difference has turned an incline of 20m into a flat surface, albeit one with a 10m shear drop at the start

Unlike horizontal maps, vertical maps don't change. A map needs to be updated every time a new development is finished, a new road is built or an old one closed, in many cities across the globe thing have changed significantly in the space of a few decades or even less. However the land itself rarely changes, that hill or valley has been there for thousands of years and will be there for thousands more

 Solutions

Requirements

Any solution must be relatively affordable and easy to use. There should be a direct correlation between cost and speed/accuracy, so the lowest cost solution should be slower to gather data and/or less accurate and that should improve with more expensive solutions because there's no point having something that costs more that's as good or worse than the cheaper one.

Dead Reckoning

One possible way to model terrain is by recording horizontal movement and slope and combining the two to create a profile. The slope gives you the angle and the horizontal movement the length of the slope or flat part. The key to this is synchronisation between the slope recording and the horizontal recording, whether by time or Lat/Long position

GNSS

GNSS is the proper name for what most people refer to as GPS. It encompasses GPS which is a system created by the US and other systems such as Galileo, BeiDou and GLOSNASS.

GPS has been available to the consumer for decades and almost everyone with a smartphone has a GPS receiver of some description which gives a low barrier to entry. However it is not very accurate which, as discussed above, is very important when it comes to vertical accuracy. Some devices support more than one GNSS system and as 

Post-processed GPS accuracy 0m average over 15m with a range of -2 to 2m vs an accuracy of -3m with GPS

Lidar

Some phones such as the Apple iPhone 12 and 13 pro feature lidar sensors that are able to be used with apps to create 3D models that are accurate to an average of 6cm (4 to -12cm range) at least for interior walls and 8 cm for exterior floors. It also seems to handle slopes and steps correctly rather than turning them into flat surfaces.

Drones can be fitted with Lidar sensors to quickly take Lidar scans of large areas

Physical 

 

Publishing format

Data should be available in a format that's easy for users to adapt for their own requirements. A digital elevation model (DEM) seems to be the ideal format. This is provided in the form of images with each pixel representing the height of that area with a colour that represents the elevation. This format is limited by how much area 1 pixel represents (1m²/pixel vs 10m²/pixel) and how many colours are used, a 256bit greyscale image can only represent 256 unique values which, if used on a global scale would be very limited, if 1bit = 1m in height that'd limit the model to a height that had a range of 256m. On a local scale it might be possible, large cities aren't usually right next to mountains so different areas of the map could have different scales, 0-256m on the cost and maybe 50-306m in the mountains. However there'd be areas that have height changes of over 256m and having to convert scales each time a section of map is imported into a project would be an added annoyance, not to mention that it'd limit changes in elevation to 1m intervals which could flatten smaller changes in elevation

Accuracy

Accuracy, or rather resolution, in DEM is usually expressed in CM or M. The lower the number the better, a 1m DEM would be more detailed than a 10m DEM which is more detailed than a 50m DEM, but what do these numbers mean? Does 10m mean that a height represents 10m²

The future

As always technology will improve over time and will lead to better tools that'll lead to lower cost or better solutions. One development is the launch of the High Accuracy Service (HSA) by the Galileo GNSS network of satellites. The service is expected to be free but will probably require equipment that's compatible and thus more expensive than the standard receiver. The specs point to a 40cm or better vertical accuracy 95% of the time as opposed to 8m on the standard system.

GPS is adding more satellites and by 2022 they the augmentation system on L2C is expected to be fully operational. I'm unable to determine whether any devices are able to

 More smartphones will support SBAS that increase accuracy at no cost. EGNOS in europe offers <4m vertical accuracy on devices that support it at no extra cost and the other services fare even better.

Potential uses

Build the earth is a project that aims to recreate the entire world in Minecraft using 1m³ blocks. Being able to convert a detailed DEM to this format would be easier and more efficient than doing the opposite. It's easier to throw some data away to decrease resolution than to increase it

If you have any suggestions, please leave a comment!