I am new to 3D modelling using the Zenmuse P1 and Matrice 300 RTK.
In the example below, I'm collecting some test data on a small building using a Smart Oblique mission. The workflow summary is as follows:

- Setup and connect  to DRTK 2 base station
- Fly the mapping mission with M300 using a P1 with smart oblique on
- Process the raw imagery using ESRI Drone2Map (Pix4D with an ESRI skin)

The models themselves render fine, but when I turn on the base map I notice they are drawn about 10 or so meters into the ground. I didn't set ground control points because I'm not looking for cm level accuracy here, I figured that with the base station I would get within the ball park at least but that doesn't appear to be the case.

Any guidance or hints on where I'm messing up would be greatly appreciated.

Equipment:
- M300
- Zenmuse P1
- DRTK 2 Base Station
- ESRI Drone2Map

Haven't used D2M but I'm a long time Pix4D user and hydrographic/terrestrial mapper.

Sounds like a vertical datum issue, probably an ellipsoid/geoid separation one (shifts like this usually are, unless they're small enough to be antenna offsets). Could have been introduced during acquisition or processing.

When you entered the coordinates for the D-RTK 2 before the survey, what vertical datum did you use for Z?

Also please describe your settings for the input and output coordinate system vertical datum in D2M.

patiam Posted at 5-9 13:16
Haven't used D2M but I'm a long time Pix4D user and hydrographic/terrestrial mapper.

Sounds like a vertical datum issue, probably an ellipsoid/geoid separation one (shifts like this usually are, unless they're small enough to be antenna offsets). Could have been introduced during acquisition or processing.

I have extensive experience with 2D GIS mapping and fixing all kinds of coordinate system issues in that context, but the 3rd D is a whole new ballgame.

My hunch as well is that it has to do with a settings discrepancy but I don't have enough background with the vertical side of things to troubleshoot it.

I've attached screenshots of the settings from the DRTK 2 synced with the M300 on the day the data was captured, as well as a screenshot of the D2M settings.

Did you enter those Base station coordinates into the D-RTK 2 or are they from its independent  position solution? Those should be precise coordinates in X Y & Z that you enter yourself, and the Z should be in Height Above Ellisoid (HAE) and include the antenna height above the benchmark over which you placed the base.

You have the Image Coordinate System parameter in D2M correct.

What are the D2M choices for Project Coordinate system vertical reference? It appears that as set you asked to put you project in HAE. Is there a choice for Orthometric Height or NAVD88 or something? If so, and you choose that and enter the Geoid Height (difference between ellipsoid and Geoid heights) for your location, then you will get your results in a more commonly used veritcal datul for mapping, instead of HAE.



There are a ton of thread here discussing this:
https://forum.dji.com/forum.php?mod=redirect&goto=findpost&ptid=264744&pid=2741050
https://forum.dji.com/forum.php?mod=redirect&goto=findpost&ptid=264115&pid=2733492
https://forum.dji.com/forum.php?mod=viewthread&tid=181111&extra=page%3D1%26filter%3Dtypeid%26typeid%3D693%26typeid%3D693
and others

patiam Posted at 5-14 09:28
Did you enter those Base station coordinates into the D-RTK 2 or are they from its independent  position solution? Those should be precise coordinates in X Y & Z that you enter yourself, and the Z should be in Height Above Ellisoid (HAE) and include the antenna height above the benchmark over which you placed the base.

You have the Image Coordinate System parameter in D2M correct.
[Image]

Thank you for your reply. The base station coordinates you see are just what the base station reading was, it was not on a known survey marker but I think that should be accurate enough for my purposes. I don’t need to generate a survey report or anything like that, but I would still like to obtain some level of accuracy so it can be combined with other GIS products from different sources.

The D2M vertical reference options are

EGM 84
EGM 96
EGM 2008
Ellipsoid
Height Above Ellipsoid

I noticed in the image metadata from a recent flight (not the one from the screenshots in the original p), that the AGL altitude was 99.98m which is what the mission altitude was set at, but the ASL altitude (this is what is pulled into D2M) was 96.26m. On the controller during flight the ASL was reading over 100m.

Are you aware of any software applications that would apply an accurate geoid correction to each image based on individual locations rather than just a general correction for the entire area

You should be aware that when you do not "tell" the base station exactly where it is, and instead rely on its autonomous coordinate fix only, you might as well not even be using RTK, unless you are also using GCPs or are going to use PPK.

This is because the D-RTK 2's autonomous solution is subject to the same uncertainty as any L1/L2 GNSS receiver w/ no source of corrections- namely, it is ± a couple meters on average, and maybe in the decimeter range at best. So the corrections given to the aircraft are derived from an imprecise base station coordinate, and while they may be internally consistent, the real-world accuracy is not survey-grade.

That being said, if you want to map using high precision GNSS-equipped aircraft, You need to educate yourself on how RTK/PPK works, as well as the basics of geodesy- how both horizontal and vertical datums work and how to convert between different ones.

The "Above Sea Level" GPS altitude ref in the geotag is BS. The GPS altitude in your RTK DJI aircraft is always going to be HAE unless they dramatically change the software/firmware. Raw (and corrected) GPS Z solutions are in HAE. Unless something applies a shift or conversion during acquisition, they remain in HAE.

Looks like D2M has changed the nomenclature a bit from the underlying Pix4D software on which it was based. Again I'm not a D2M user, but my take from the choices you listed is that you should choose "Height Above Ellipsoid" and enter the Geoid height in the appropriate box. I can't figure out a better option from those given that will get you to orthometric height, which I think is the goal here if you want the drone data to line up vertically with other existing data. The D2M documention should explain this clearly, so I suggest you look there to make usre what I've guessed at here is correct.

Pix4D (and software using its engine) allow for only a single Geoid Height to be applied for a given project. Unless your site is very large, the Geoid Height varies by only a few cm across it, so while not perfect, using a single value works pretty well for most sites, and only intoduces a couple cm error across them. It would be better if Pix 4D would allow the user to specify a raster or other file that models how the Geoid Height varies over space, and that functionality has been requested, but they have not implemented it. You can apply a spatially-varying Geoid Height to your DSM in GIS instead of during photogrammetric processing, but I don't know of a photogrammetry application that does this.

Again, if you want to get the most out of this kit and workflow, it will serve you well to spend the time to understand that with which you are fiddling. I've provided a bunch of links here that should help, but really a more formal course of study would be the best option.
Hope that helps, good luck.