Error In Z(elevation)

[chandansingh23396]

We are performing geotagging using a PPK drone where we are geotagging the image using our solution. But when we process the data using “Pix4D mapper” we find that x and y coordinates are right whereas the z coordinate shows error in initial and computed position. For this, we are not using GCP but are using checkpoints and we are using ellipsoidal height in our overall project.

So where are we missing also in a quality report?
Please find the attached quality report and screenshot of the initial and computed position.

pix4d.pdf (1.9 MB)

[Nikoleta_Dimopoulou]

Hi @chandansingh23396 ,

I had a look at the quality report and I noticed that your GCPs that are used as checkpoints have 0,50m difference between initial and computed Position more or less.

It looks to me more like a systematic error that this can be related either to the collection of your GCPs or the image geolocation.

Are you sure of the exact heights of your GCPs? May I ask how did you collect them?
I would also suggest changing the Horizontal and Vertical Image accuracy to 0.02m and process again.

Best,

[chandansingh23396]

Hi @Nikoleta_Dimopoulou
I have collected the GCPs through DGPS in static mode.
Can you tell me about the red indicator in the absolute and relative geolocation variance?

[chandansingh23396]

Hi @Nikoleta_Dimopoulou
As you said in your rply,we have changed the horizontal and vertical accuracy then after processing we got the 0.3m error in Z and I have also cross checked mine GCPs as you said then the GCPs is completely fine.
Please check the quality report and screenshot also and tell me what should i do? and tell me one thing why are you suggesting me to change horizontal and vertical accuracy?

PIX4D.pdf (1.8 MB)

[Nikoleta_Dimopoulou]

Hi,

In the first quality report, you shared with me the RMS error of your images was around 3-4 cm that is why I suggested changing the horizontal and vertical image accuracy.

In your last report, the accuracy changed to 6-8cm. I would suggest changing again the horizontal and vertical image accuracy to 7cm. Have in mind that this way you put different weights to the image geolocation. I would also recommend using 3 of your checkpoints as GCPs to georeference the project and maybe keep the other 3 as checkpoints. Make sure that the 3 GCPs are well distributed in the project.

I hope you find this helpful!

Best,

[chandansingh23396]

Hi @Nikoleta_Dimopoulou
As you said in your previous reply, we have corrected the Absolute and relative geolocation errors. I have also cross-checked my GCPs as you said then the GCPs are completely fine. but after processing, we got the 1.8m error in the Z.Can you tell me how to correct this error?
One thing I told you, we have used PPK solution, that’s the reason we don’t use a GCP because I think, there is no need for using GCPs.
report1.pdf (1.5 MB)

Please check the quality report and screenshot also and tell me what should I do?

[Mike_K]

Hello,
Thank you for providing your quality report. It appears that you have high camera optimization errors. I would suggest enabling All Prior under step 1 processing options. This should correct your project.

[chandansingh23396]

Hi @Mike_K
Hi @Nikoleta_Dimopoulou
As you said in your previous reply, I have changed the setting All to All prior under step 1 processing options but after the processing, I have got the same result.
Please check the quality report and screenshot.
report21.pdf (1.4 MB)

[Mike_K]

I’m sorry but I am a little confused right now. How are you postprocessing a P4P?

I also see that in some quality reports you are using GCPs/CPs and in others you are not. The only way to truly verify your outputs is to use some kind of control on the ground. Otherwise we cannot make a full assessment of the model.

[chandansingh23396]

Hi @Mike_K
In all previous reports, we were using checkpoints instead of GCPs because we are using a PPK solution for our project.
I am sharing a quality report with you. If are you still confused? can we share our project detail with you?
Please check your side. we were wrong in the post-processing .Pls, correct me and share your google drive link with us.
report22.pdf (1.5 MB)

[Mike_K]

Sorry, I am still a little confused how you are using PPK workflow with a DJI Phantom 4 Pro that is not typically equipped for this.

On your last report, which included the use of checkpoints, indicates that your model has an RMS error in the Z axis of approximate 1.8 meters. Interestingly, all of your checkpoints are within 6mm of each other. I believe this would indicate a discrepancy between the vertical coordinate systems of your checkpoints and your images. Can you verify what vertical datums you are using for both your checkpoints and images?

[m.lammers]

Hi @chandansingh23396

I’m having the same issues with RTK tagged photo’s, taken with a Phantom 4 RTK. Have you found a solution?

[Mike_K]

Assuming you did not use an external correction source aligned to a different datum, the geotags in your P4R data are Geographic (Lat/Lon) coordinates WGS84 datum, height above ellipsoid vertical. But if you are using an NTRIP then it is most likely correcting on a different datum and this datun needs to be correctly specified in the image coordinate system. If you are in the US then it is likely that it is NAD83(2011) but you should verify first.

[XanDerALeX]

Hello,

I have similar issue with the elevation of cp.I have flown with P4P and as you can see on the report, the z values of the checkpoints have something wrong.I don’t understand why i have small error values on gpc’s and high error values on cp.
Any suggestions?

[Nikoleta_Dimopoulou]

Hi,

Could you please share with us the pdf file of the quality report to have a look?

Best,

[Chris_Putnam]

What is the distance and the orientation of your surveyed GCPs relative to your survey?