How to convert heights from WGS84 Ellipsoid to GRS80 Geoid with Data from DJI Mavic 3m?

Hi,

I use a DJI Mavic 3m drone with RTK to make orthophotos and 3D models. I was hoping to use the data without taking GCPs as the drone has a built in RTK Modul which is corrected using the German SAPOS system.

However the drone writes the data using the WGS84 ellipsoid and the results are approx. 50m higher than the standard coordinate system EPSG:25832/GRS80.

According to my research, I should just have to subtract the Geoid undulation (available from the national geodesic authority) but this still results in heights around 2 m higher than they should be.

Does anyone have experience with DJI Drones with RTK and the workflow to correct the vertical datum?

Regards,
Robert

Good question. The GRS80 is also an ellipsoid model which is very similar to the ellipsoid model used with WGS84. There are many geoid models available. For example, the latest geoid model for the US is GEOID18 which can be obtained from the NGS website or from your survey processing software. It is possible to add a geoid model in Pix4d but I have always used ground control as it is the preferred method.

2 Likes

Hi @r.buckley?

SAPOS broadcasts in the ETRS89 datum. So, if your drone receives corrections from this service, your geotags will also be in the same datum.

Where will you process your data? Is it in Metashape? If so, I suggest using the DHHN2016 geoid.

If your system doesnā€™t have this geoid yet, you can download it on Metashapeā€™s website and import it into the software.

2 Likes

The software we use is Opendronemap (WebODM).

by the wayā€¦the drone images do not contain ETRS89 coordinates in the EXIF Metadata.

The exif data from the M3E with RTK contains the corrected data in WGS 84 and Ellipsoid height. WebODM drops the results in the nearest UTM CORS and keeps the Ellipsoid height if you donā€™t use GCPā€˜s. Here in Germany UTM 32N or UTM 33N. If you want DHHN2016 heights, youā€™ll have to manipulate the exif data off the pictures by using the according undulation or manipulate the resulting pointcloud. By using WebODM there is no better way than using GCPā€˜s. Commercial Photogrammetry Software have tools for georefferencing the results. Iā€™m using 3dsurvey. Itā€™s worth to have a look on it.

1 Like

Hi @b.hofmann.pan and @r.buckley,

The drone doesnā€™t do any datum transformation during the flight, so if your drone receives corrections from SAPOS, your image center positions (latitude, latitude, and ellipsoid height) will be in the same datum, ETRS89.

The same applies to all RTK and PPK surveys. The position is calculated based on the CORS (base) coordinates system.

I havenā€™t used WebODM, so Iā€™m not sure if you can convert the height to orthometric. If itā€™s not possible, you can maybe try the raster calculator in QGIS to apply the undulation value to get the orthometric height, or use GDAL.

4 Likes

I think, the fastest way to solve your issue is to get Emlid Reach RS2+/RS3 as a base station for corrections for your flights. Also you have to consider the overlap of the RGB camera.

ā€¦and it has to be Matic, not Mapper.

1 Like

ā€¦.where can I see these corrected ETRS89 coordinates? In the exif metadata I can only see WGS84 coordinates in decimal degrees.

You are seeing the corrected coordinates. Disregard where the metadata states WGS84.

As stated, when you receive corrections (RTK or PPK) your metadata will now have its coordinates in the same datum/coordinate system that the base uses.

This is highly misunderstood and needs to be placed in bold in various places both in Emlid apps, instructions as well as in DJI and photogrammetry software.

2 Likes

So how can I view the ETRS89 Koordinates? As I said, all Data is in WGS84.

You just need to look at the image exif data. Those values inherit the CS values used by the base. If the base was broadcasting corrections in WGS84, then your image values will be based on WGS84.

If your base was broadcasting corrections in EPSG:25832/GRS80, then the values of the positions in the image exif also now reference that system. As @jaja6009 mentioned, the parameter value of ā€œwgs84ā€ in the image exif does not change to reflect this. But your values are now referencing the CS used by the baseā€™s CS. In this case, where the tag says ā€œWGS84ā€, just understand that that tag is incorrect.

Also to note. If you do not use RTK, but instead just use the drone in ā€œsingleā€ mode, then the values reflect wgs84.

3 Likes

My exif Metadata defines the CRS as WGS84 and displays the Coordinates in WGS84 decimal degrees (DD).

So according to your explanation the droneā€˜s normal GNSS GPS coordinates are corrected by my correction service ( which transmits in ETRS89) using the RTK module, and then were transformed back to WGS84 and written to the EXIF metadata?

The correction of the GNSS WGS84 coordinate values using ETRS89 values would require that the original WGS84 values be transformed to ETRS, then corrected, then transformed back to WGS84 for the exif data. Is this the internal workflow of the RTK module?

Hi @r.buckley,

Thereā€™s no transformation involved in this workflow. As @jaja6009 and @dpitman mentioned, your tags are referenced to the same datum as your base, ETRS89, even though the EXIFā€™s metadata shows WGS84.

I believe WGS84 is the default value in the EXIF because the drone has no way of knowing which datum the corrections itā€™s receiving are based on.

3 Likes