Looking for advise on which Ntrip correction network to go with. I also have the D-RTK2 but still want to connect to a NTRIP server when not using it or to check accuracy. US based Connecticut
Hey @joestefano, welcome back. What are you using the drone and base for? There are several ways to set it up depending on want you want to accomplish and some very specific things that HAVE to be done to accomplish certain workflows.
Looks like you have a decent number of CORS sites so I would check access to them first. They may only be available via an NTRIP subscription though. An alternate in this method is to do PPK which allows you to get the data for free offline and correct it with better verifiable information.
Honestly though I can’t think of any reason other than wanting to use the D-RTK2 as a rover to collect ground control points as to why you would need any kind of service corrections. If you let your receiver average in and create the fixed point and set a monument then you have good coordinates to manually enter the next time to have repeatable surveys.
Another option that allot of people ask about and I won’t get into yet is coordinating your flights and data with a local grid like you would see in a pre-existing survey or construction site.
Hey Michael, thanks for the response. I am about to start preparing a 10 acre site for the construction of 2) 40K sqft buildings. I am the GC for the entire project. I have been flying the site since August while we took down a 106K sqft building that was on site. I own a Mavic 2 pro and Inspire 2. The site is in a flood plane and we have thousands of yards of earth to move around and import/export. I purchased the P4P RTK with D-RTK to keep track of all the changes almost daily. As of now I have only a few known points given to me in SPC without elevation (not sure what to do with these) and a catch basin with an elevation that I assume is correct. I want to be able to fly just the RTK sometimes and didn’t know if that should be with it connected to a CORS or not. My primary need is topo but I want to make sure that I take full advantage of the system. Your statement below fits this situation pretty well.
“Another option that allot of people ask about and I won’t get into yet is coordinating your flights and data with a local grid like you would see in a pre-existing survey or construction site.” I am using Dronedeploy Pro plan for processing.
Great explanation and cool project! We are finishing up a similar one here.
Are you trying to bring the drone data into design CAD files or are you just wanting to document with the map on the DroneDeploy online tools for cut/fill work performed from flight to flight? If just relative flights to be compared in DroneDeploy then you can take the following information as just food for future thought.
There are a couple of ways you can handle integrating with CAD. The preferred method is with GCP’s as they will take into account the site control localization and put the drone data exactly where it needs to be on the local grid. For those that aren’t familiar with localization, it is basically a method to locate the control points as they actually are on the ground and compare them to what record or CAD says it should have been. It then looks at all the control as a network and prorates that network in all axis as a best-fit scenario into what the CAD/record said it was supposed to be. For this size of site you should be good to go with 4 points safely in/near the limits of construction that “box” in the site. They don’t have to completely encompass the site, but maybe like this.
Once you set the points you could collect the Lat/Lon by setting up your base on each point and letting it acquire a coordinate. Take a picture or somehow document those coordinates for entry into the DroneDeploy GCP template. Then all you have to do is have someone level-loop those points using the site benchmark. If you have a survey crew available they can do all of this at once. As long as they are in a safe place you shouldn’t have to worry about it again. Make sure to overfly the site if you need to get the GCPs further from construction to be safe.
Another method is to just use what the RTK system acquires and have one centrally located control point on the site that you have the known ground elevation for that is on the same benchmark as your construction plans. This will allow you to do a single-point calibration in DroneDeploy and get your elevations inline with your site survey, but may or may not line up with the CAD horizontally. This is where you will need the capability to shift the drone data in CAD to make it line up horizontally. We can discuss this further if you like.
What would you do for finding out, on wich exact location the gcps are placed if you don’t have this dji rtk base? Using google maps coordinates is to inaccurate I guess? How do I make sure the coordinates of the gcps are proper accurate?
I don’t have a surveyor on my hand of course yet.
How big is the accuracy difference approximately between the first and the second method?
This will all depend on what you really need. We have several different scenarios that can take place that will provide different accuracies on two different scales of relative and absolute. We also have a factor of some that just want repeatable maps that can be compared against each other.
Centimeter accurate drone surveying requires special GNSS equipment to either place GCP’s on the ground and/or supplement the drone’s GPS tagging of the images. Without GCP’s you are likely +/- 1m relatively accurate and 1-3m absolutely accurate. One is the map against itself and the other is the map against the real world values. There would be different tolerances horizontally and vertically as you see with a drone’s specs, but I’m trying to keep it simple spatially.
The next level is to have GCP’s alone, but in my testing there are many ways to establish GCP’s for different purposes. One of the crudest, but actually quite effective for setting a basis to fly maps the a highly relatively accurate to one another is to perform an initial flight and use a program like QGIS to extract GCP information from that first map to be used in future maps. I use this method when all the client wants is a way to see change over time. An example is a roadway that I flew for a neighboring municipality that just wanted to track degradation. I did there initial flight which we could analyze to get an idea of what they really were facing in the way of remediation, but I also used this flight to identify enough markers on the ground such as corners of paving and manholes to create a rough GCP network that could be used in each map moving forward. With this we could do side-by-side map comparisons that lined up exactly with each other and also could quantify some of the rates of decay. This method is not good for doing any work related to real survey data.
Next would be survey-grade GCP’s. Here again you can get different rates of accuracy depending on how many you use. Using 4 or 5 will greatly improve your absolute accuracy to the world, but will not improve your relative accuracy as much. The more GCP’s you place the more relative accuracy increases, but absolutely accuracy only really increases on the Z-Axis. I’d say this level gets you to about 7cm relative accuracy and 10-12cm absolute.
Next group would be the use of RTK/PPK on the drone itself. This can greatly increase relative accuracy without the need for GCP’s, but the only absolute accuracy it helps is GNSS global location. You can get 3cm relative accuracy, but the absolute accuracy is unknown without getting survey involved.
Now we have a break in absolute accuracy. There is absolute global GNSS accuracy and then there is surface accuracy. This happens because the drones use one kind of datum and surveyors on the ground use another. Allot of the time they are very loosely related. There are transformations, rotations, scale factors and other minor adjustments that can all come into play so that it is rarely a 1:1 relationship.
This is where you combine drone RTK/PPK with GCP’s. RTK/PPK on the drone relieves the need for a large number of GCP’s and the GCP’s fix the problem of translation between GNSS and surface accuracy… for the most part. It continues to get deeper, but this is the level that you have to get to in order to guarantee 2cm absolute accuracy with drones and photogrammetry.
It’s late so I will have to proofread this in the morning, lol.
That’s a proper answer. Thank you!
I have to reread this as well for sure due to the fact it’s a couple of hours later here and I have nightshift.
But already got most of it. Thanks for the good understandable writing. Very useful!
edit Oh just one thing. I’m not sure if you answered the question how you would find out the exact coordinates of the gcps you actually placed, without having a ground station or an surveyor.
Good night Mr. Michael
I have an app on my phone that will get me within a meter after about 10 minutes with weighted averaging, but the real answer is - You won’t. The closest you can get without surveying it would be flying it and then getting the coordinate off the ortho map using QGIS or some other CAD/GIS software.
Ah ok! I thought when we are creating a project including gcps that we have to know the coordinates of them, put them in this template before processing/uploading the images and then it will
Process. But when it’s working without as well, than it should be fine.
What is the app you are using called, if I can ask?
I see what you were thinking. There is no pre-planning in DroneDeploy for the position of the GCP’s. Sometimes if I am running close on my flight window I will throw down the GCP’s real quick, fly and then collect the coordinates on them afterwards. They just have to be in the pictures right?
The app I mentioned is Precision GPS Pro on Google Play. I am sure there is something similar for iOS.
Ok, now I am happy
Thank you very much! Again and again and again