Scanning a Mountain

[SvSz]

Hello everyone!

Until now I have only been to ParrotPilots and Phantompilots, but I have a larger project coming up.

We are planning to map out a large mountain with roughly 0.9-1.5 km² (222 – 370 Acres)
using the DJI Zenmuse P1 35mm mounted on a Matrice 300 RTK to create a 3D Model from it.

The structure of the mountain is rather complex with all kinds of surfaces: flat, vertical and tilted.
To have an idea of the surfaces I am talking about, compare it to


Kleiner Watzmann or Small Watzmann (Germany)


Mount Triumph (Washington State)


Cerro Torre (Argentina/Chile)

To scan skycrapers and other tall buildings, you can use UGcs, DroneDeploy, Hammer, Dronelink etc.
to preplan vertical and horizontal flights.
Unfortunately, these services do not provide reliable DEMs for mountains. I see three problems with that:

- You do not exactly know where to place the routes, since some elevations are poorly, falsely or not resolved.
- You might end up with a route that has lots of obstacles in it that intervene with the obstacle avoidance system, making the flight useless.
This can even be the slope itself or some larger rocks that were not resolved correctly.
- Your GSD to (esp. for vertical walls) is not consistent, making it harder for the software to create 3D Model, resulting in a suboptimal final product.

A Manual flight seems impossible due to the size of the project.

My solution for all this was to use the Agisoft Metashape Professional workflow, creating a Mission Flight Plan tightly wrapped around an existing or in-situ generated DEM.
Can anyone confirm my idea as feasible or has a better idea? Should we just stick with UGcs? We might also be able to import better satellite derived DEMs into the software.

Cheers and thanks for reading!
SvSz

 

[John Githens]

This is not my area of expertise...
Have you read The Shuttle Radar Topography Mission (SRTM) Collection User Guide? More than the Guide is available on the Earth Engine Data Catalog.
I don't know if you would get the desired level of granularity of surface data for your flight planning. I would think that planar views of a mountainous DEM at different altitudes could be used to guide flight planning.
Given the 3D features of Mount Triumph as can be seen through Google Earth Pro, perhaps there's a way to export that DEM data and plan flights to be further away from surfaces than you would prefer, to avoid anomalies that could cause a crash.

With a lot of trial and error, you might be able to create Polygons in Google Earth Pro at progressive altitudes above sea level, then use the break points in the polygons to determine a "safe" distance away from the 3D points you marked on the mountain. See attached screenshots 1 - 5.

So here's a note of caution. See screenshot 6. The road is flat. It does not undulate. And those treetops and bushes are not uniformly sloping down to the flat area next to the concrete seawall.

You may want to look at Site Scan for ArcGIS. You have an RTK drone system. The video suggests a way to prep your flights. The original Site Scan technology appears to have been developed by 3D Robotics.

 

[John Githens]

Also worth reading to consider your options:

Optimization of UAV Flight Missions in Steep Terrain​

Note the cautionary comments about relying on SRTM DEMs.

 

[Gagey52]

Greetings mate, welcome to the forum. Sounds like an exciting job. Look forward to hearing how you go with it. Not much in the way of mountains in my area.
Regards

 

[AMann]

Haven’t read John’s article yet, just the abstract, and looks like a really good resource.

One thing I would consider is using a UAV fixed wing vtol type of aircraft to do this. You could have it follow contours on the mountainsides with a side facing camera or lidar and be able to have it scan in a lot less time and fewer battery changes than with a rotorcraft.

I heard of a similar study done in Yosemite Valley. People used a high resolution DSLR to make megapixel images using a long lens (600mm) with the camera mounted on a tripod to map out the sides of El Capitan. They did this from various positions in the valley in order to make a detailed three-dimensional map of its face.

Edit: I guess it’s been been a while since you posted this, how did your study go and what did you use?