Base station - rtk/ppk

[last6]

Hi,
I’m new to photogrammetry, struggling to get my head around a few things and would like some advice please.
I have a Matrice 4E, a D-RTK 3 base station and access to ntrip corrections.
What would be the most suitable way to incorporate all of the above to provide the most accurate surveys? Struggling to get my head around RTK/PPK processing as all of my surveys are out by around a metre at least. I record GCPs using the D-RTK 3 as a rover (using ntrip corrections), I then use the D-RTK 3 as a base station, connect my drone and fly the mission. Maybe it’s something I’m doing wrong at the processing part. Confused.com!!
Thanks for any help provided.

 

[jaja6009]

You have everything you need. You just have to put together a workflow.

First.

You have correctly connected the DRTK to the NTRIP service? ( I use Emlids so I cannot help you there.) But, do you know what datum your NTRIP service gives its corrections in? In the US, most will send corrections in NAD83 (2011) and use ellsipsoid for elevations.

You state the GCPs are out by a meter? For better communication, lets call them Control Points. Are these Control Points used as GCPs or as Check Points?

What software are you using for processing? I am assuming Terra, since you got a free subscription with you M4E purchase.
I use Agisoft, but might be able to help with Terra.
You state a meter inaccuracy. Is this meter an average for all Control Points? This sounds like a datum transformation issue.

There are a lot of other questions, but these will make a good start.

 

[last6]

Wow, thank you for your quick response. I record Control Points with a rover using ntrip and standard deviation errors horizontally and vertically are in the 0.006m which I’m very pleased with. I then use the base station with the same ntrip corrections to fly the mission. It’s when I process all info in Terra, the resulting ortho map is out by a metre. For example, if I obtain coordinates from any given point on my ortho map and check them against Google Earth Pro, I am out by at least a metre. I guess it’s how I’m importing the Ground Control data and processing data. Note: 10 points over the site recorded. 5 used as Control Points and 5 used as check points.

 

[jaja6009]

Are you saying that when comparing a point on your Terra created map, it disagrees with Google Earth by a meter?

If so:

1. Google Earth's coordinates are not survey grade.
2. Google Earth's coordinates are in WGS 84, and most projects in the US want to use NAD83 (2011) and a projection with State Plane being the most common.

To assess your maps accuracy, you do so by the RMSE of the Check Points. If you want to compare against another map, make sure the map used is in the same datum and also is of a high accuracy.

I would start off by knowing what datum your NTRIP provider sends corrections in. If in the US, a good assumption is NAD83 (2011) with ellipsoid elevations. I am unfamiliar with DRTK as I use Emlid GNSS units. Give me the elevations of your collected control points.

I cannot help you with Terra on DTRK as I use Agisoft Metashape Pro and Pix4D matic and mapper. But I can help you understand datums, coordinate systems and geoids.

Finally what is your output datum and projection?

 

[last6]

This is so helpful. Thank you so much for the advice. See attached for GCP data. All gathered using ntrip corrections (WGS84). Outputs in Terra set for IRENET75 (Ireland). I think my gcp data is good, judging by standard deviation figures. It’ll be something really basic I’m getting wrong I’m sure. So much to learn

 

[jaja6009]

This is so helpful. Thank you so much for the advice. See attached for GCP data. All gathered using ntrip corrections (WGS84). Outputs in Terra set for IRENET75 (Ireland). I think my gcp data is good, judging by standard deviation figures. It’ll be something really basic I’m getting wrong I’m sure. So much to learn

I don't think you're doing anything wrong. Google Earth's points are accurate measured in feet in meters. Your outputs if done correctly will be in the centimeter range.

On your Terra Quality report what do you have for your RMSE on the Check Points?
As long as your Checkpoints are laid out correctly and measured correctly, this is how you gauge the accuracy of your map.

 

[last6]

Thanks for all the info. A great source of info!!
GCP errors attached.

 

[jaja6009]

Thanks for all the info. A great source of info!!
GCP errors attached.

Make sure you set some of them to Check Points, and you look at the RMSE for the check points to assess your accuracy.

GCP's are used by the software to correct the accuracy of your outputs.
Check Points are not used will aerotriangulating. They are only used after the initial sparse point cloud is made to "check" the accuracy.
Again, do not use Google Earth to assess your accuracy. While it can be in the foot or meter range for accuracy, your Matrice 4 E can output products in the low centimeter range.

If you would like to further check your results, see if you can find some authoritative Lidar LAS files. You can use Cloud Compare to compare different areas.

I do not use Terra for orthomaps or DEMs, as I use Agisoft Metashape. I only use Terra for its best in class (Lower Cost software) 3D models. So I am of limited use for the proper use of Terra, but all photogrametry software assess accuracy through the use of Check Points.

 

[last6]

This is all valued information so thank you very much. I’ll re process and set some GCPs as check points and check the results.

 

[jaja6009]

Make sure they are spread out. Since you used RTK, you can set all of them as check points. Run some tests, try half as Checks and half as GCPs. Then run all as checks.

If you used Altitude Optimization where the last flight line is a diagonal line of oblique images from an outside corner to the center, you can use all of them as checks. Otherwise there is usually a vertical error when only using nadir images. This does not happen with all obliques or nadir and obliques mixed.