Errors with M300 RTK Georeferencing
763 5 2023-11-29
Uploading and Loding Picture ...(0/1)
o(^-^)o
pendzoil
lvl.1
Canada
Offline

Hello,

I am conducting mapping missions with the M300 RTK and P1. Base station is an Emlid RS2+ connected to NTRIP with CAN-NET. Data processing is performed by Pix4Dmapper ver 4.8.4. When processing the data, the output is in NAD83 (CSRS) / UTM zone 10N and elevations adjusted to CGVD2013 (CGG2013) height datum. Corrections are being sent from the base to the drone using the Local NTRIP method within the Emlid RS2+

I am running into errors with the Orthomosaic and DEM being shifted when checked against the GCPs. GCPs are placed around the survey area using the RS2+ and NTRIP FIX connection. When The Pix4D model is checked against the GCPs the results are:

Relative accuracy of the model is very good with
X Error = 0.6cm
Y Error = 1.1cm  
Z Error = 3.9cm

Absolute accuracy is very poor for an RTK enabled drone with
X Error = -1.13m
Y Error = 0.62m
Z Error = 0.39cm

I don't know where that error is coming in. Could it be that the conversion from WGS 84 imagery to NAD 83 model? Looking at image properties, it shows that the images had FIX RTK with the base station but could there be an error in adjustments from the RS2+ base to the M300? I have conducted multiple test flights and all are having horizontal errors just over 1m. It seems to be consistent inaccuracy.

Any help would be greatly appreciated as these errors are very concerning for a professional RTK drone.

2023-11-29
Use props
LV_Forestry
First Officer
Flight distance : 4726654 ft
Latvia
Offline

Are the coordinates of the base in the RINEX file correct?
What is your procedure for installing the base?
2023-11-29
Use props
pendzoil
lvl.1
Canada
Offline

I just processed the RINEX file in Emlid Studio and these are the results:

Base station with NTRIP Fix
Easting = 434368.26
Northing = 5440972.153
Elevation CGVD2013 (CGG2013) = 54.451

RINEX Base
Easting = 434367.9004
Northing = 5440972.2301
Elevation CGVD2013 (CGG2013) = 55.0384

That is a difference of
Easting = 0.3596
Northing = - 0.077
Elevation = - 0.5874

Looks like there is some error there but doesn't corelate to the model error.

Procedure for installing the base:
- Set up Emlid RS2+ on tripod (Level out, record pole height)
- Within Emlid Flow app I will go into the survey that I was using to record the GCPs and adjust the pole height and recording time. I will take a point for usually 2 min FIX in order to ensure that I have a point at the bast station to compare Local and Global Coordinates (I use this to manually adjust vertical datum later by comparing the two).
- Within Flow App I will adjust the base settings of antenna height and collection time (2 min with FIX solution) to get the base location for output. I have sometimes taken the previous 2 min fix point from the survey project as the bast station location but wasn't sure if that was causing some of the issues so I started just re recording the base location.
- Once I have the base location, I will turn off NTRIP corrections to the base and turn on local NTRIP for the drone.
- I will then turn the RC and M300 on and wait for RTK correction FIX before flying.
2023-11-29
Use props
pendzoil
lvl.1
Canada
Offline

I think I have found the issue. It is a projection issue with Pix4D. When output coordinate system is in NAD83 (CSRS) / UTM zone 10N we get a large error between the GCPs and the model. If we have the output coordinate system as WGS84 / UTM 10N, the GCPs line up excellently. RMS Error (m): X  = 0.0108, Y = 0.0132 and Z = 0.0678. The GCPs are collected in NAD83 (CSRS) / UTM 10N so I guess Pix4D has issues with the conversion of WGS84 images into NAD83? I suppose this might be a question for Pix4D instead. Ideally we would have the outputs in NAD83 as that is what most other companies use for geolocation.
2023-12-1
Use props
fanse68b7f05
lvl.3
Flight distance : 476867 ft
  • >>>
United States
Offline

Instead of WGS 84 for the >>>image<<< coordinate system, try using 6318. https://epsg.io/6318
2023-12-29
Use props
pendzoil
lvl.1
Canada
Offline

fanse68b7f05 Posted at 2023-12-29 14:19
Instead of WGS 84 for the >>>image

Solution:

I had to download my own .prj from spatialreference.org and within Pix4D under Advanced output settings I have to upload the .prj file in (3157). For elevation, I have to calculate the difference between Ellipsoidal and CGVD2013 from the Base Station's survey point in Emlid Flow and adjust elevation manually, (and as a negative) within the Advanced output settings in Pix4Dmapper.

Hope this helps others.
2-14 14:36
Use props
Advanced
You need to log in before you can reply Login | Register now

Credit Rules