Using Reach RS for GCP

I would appreciate help from anyone who has a better understanding of my needs than myself. I am starting a drone surveying business and I am teaching myself as I go.

I want to be able to have percise data models generated from my phantom 4 pro processed with pix4d. I would like to be able to generate ground control points to increase my accuracy.

Could somebody explain to me in plain English how I can leverage reach RS to do this. I’m a little confused, do I need to purchase two reach RS and have one as a base and one as a rover? Or can I just use one reach RS to determine the location of my ground control points? What would my work flow look like?

Thank you in advance!

You could use one, but you need another service or device to get that sub cm accuracy. Otherwise it is only a gps with accuracy to about sub m.
Buying two RS would not help you get accurate ABSOLUTE position, it will give you RELATIVE position, like a blind man guiding another blind man… BUT another RS unit can be placed over a known point that has absolute coordinates and this can be used to send corrections data to another RS. That is a cheap solutions compared to alot of other paid services and more like a man with perfect sight guiding a blind man :crazy_face:


Hi, I see TB_RTK has already answered your question but as I was almost finished I’ll send it anyways.
Learning as you go is a great way to learn! First of all, you need to answer 1 question.

  1. (Simplified approach) Do you have an outside source of corrections? If yes, you only need one Reach RS. If no, you need two Reach RS.

For high accuracy, which is necessary for accurate models, you need to process your data with a base station’s data taken during the same time as your rover. This can be done “real time” rtk, or after the survey,"post processing " For either case, you can get this data from your own base station or from a third party via subscription or sometimes as a free service.
In my opinion, the simplest approach is to buy two Reach RS’s and use your own real time correction. One possible work flow is as follows.

  1. Arrive at site and set up base station.

  2. Turn base station on after set up(not during or before).

  3. Distribute GCP and turn on rover.

  4. Connect to Rover via Reachview app and perform your survey in Reachview(that’s right, you don’t need any other data collector)

  5. Save your survey data to your device and send to PIX4D.

  6. This is all that’s required for high accuracy data models. Now if you need high accuracy ABSOLUTELY georeferenced data models, then you need to do as TB-RTK suggested and use a known point. The workflow would be similar but you would have to connect to your base station via Reachview and enter its exact position.

There are lots of other resources on this forum about this, but if you’re not familiar with terminology, you can get lost in the details. This is a good starting place.


I’m off to the linked docs, but have a couple of q’s.
Generally relative accuracy is enough for my purpose but if I could geo-reference with a little more effort…

  1. A previously surveyed property corner can be used as the known location on which to set a base (Reach RS) over. Would the terms used to decsribe the point need to be translated into another form? ( like from northing and easting to lat/long ).

  2. If 1 is yes, can a second Reach RS rover be used to establish a new known location and then use that as the base location for subsequent marks? In oher words, using a corner marker to eatablish a known point more in the center of a parcel to then use as a base on a large parcel.


There is something called the propagation of errors.

A national benchmark may be accurate to mm within the national reference frame, a regional benchmark cm’s.
The corner of a property, well I recently found one that was 65m different in real life to that shown on maps and in the digital cadastre. That size of bust is unusual, but once out of the cities it really is important to understand the providence of the ‘mark’ you are occupying as a reference.

Here in Tasmania the survey manual says this.
Horizontal Order

This is a function of the Class of a survey and its conformity with existing network coordinates determined from a least squares adjustment. It involves an assessment of the sizes of the semi-major axes of relative standard error ellipses. The maximum sizes of the semi-major axis allowed for stations 1 km apart (95% confidence level) are as follows:

1ST 22 mm
2ND 44 mm
3RD 88 mm
4TH 147 mm

Two other statements of Horizontal Order are also used:

PRV Provisional. Coordinates determined using a survey method that does not contain redundant observations.
SCA Coordinates scaled from a map.

Some of the maps used for scaling may be a little questionable, and the above accuaracies are the points people care about, not plot boundaries.

So yes you can use a corner post and yes you can give it ‘provisional’ coordinates. If you always use these coordinates when you are occupying that spot all your relative measurements will be highly accurate. If you occupy a new spot from as a RTK fix from the corner and you spend an hour at that spot your resulting position is likely to be of a similar accuracy as your corner so you can use it as the new reference, but if you keep hopping along in little legs each one will be a littlle bit less accurate than the last. The uncertainty in the base position on each leg will be added to the final new position.

1 Like

I understand. My thinking is I have a contact that just purchased 1500 acres in a mostly square shape and is looking for a “good” DEM with contours in order to determine how they want to lay out the blue berries they plan on planting. They recently had the corners surveyed and I was just wondering if it would be a good idea to think about using one of those fresh corners to try an plot a single location toward the center to use for the survey. I am not a surveyor and I make that clear to folks. But I also want to do the best I can to get them the best data to help in their decisions. Actually, absolute accuracy is not too important in this case but it is very interesting to me.
Thanks again.

1500 acres! Are you planning on flying fixed wing or multiple quadcopter missionss?

I would rent an eBee if I were to get this project. My interest in your DJI solution would be for much smaller projects!

1 Like

Good topic :slight_smile: I want to clarify something too. I have single Reach RS and will use it just for collecting points (GCPs) in survey mode. I have a subscription to a dense CORS network with reference stations never further away than 15 kms. Now… how do I set up my RS before collecting the points?

  • RTK settings (positioning mode, GPS AR mode)?
  • Base mode? What to do here? I’m assuming the CORS network is the base, sending correction input to the Reach RS, effectively making it a static rover… correct? So corrections output off. But what about all the other settings? Since it’s not really a base in my view…

Follow the reach documentation, it is pretty good.
If you have managed to set up your corrections into your reach, then treat it as a rover and configure accordingly.
Inevitably there will be some fiddling.

Thanks for your reply @Simon_Allen. I have read all documentation already, but it’s not clear what to do in my case… where you have a single RS and use it for GCP collection only, with corrections input from a CORS network. So… my specific questions then, following the menu items from the ReachView app:

RTK settings

  • positioning mode static?
  • GPS AR mode continuous?
  • GLONASS AR mode off?
  • GNSS select GPS + GLONASS only (CORS network only supports these 2)?
  • Update rate 5 Hz?

Correction input
Base correction or additional correction on/off? My RS is acting as a rover… the CORS network is the base. So what to do?

Position output

  • OFF

Base mode

  • Corrections output OFF
  • Now the most important question: what to do regarding the base coordinates? Again… my base is the CORS network, so what do I enter here? Do I need to find out what the coordinates for the nearest reference station are and enter them manually, or can I just put the RS somewhere elevated and let it determine the coordinates by itself? What would be the parameters in the latter case? And do I already need to have my NTRIP caster going before I do that? I think I do… would make sense… but I just want to double check. I’m a noob, sorry…

See answer i bold behind your questions.

Use base correction when your base acts as rover. The other option is for more advanced use


Thank you so much! Very helpful already :slight_smile:

Thanks a lot again!

Ok, in another way. If you hold the pole with your hand and no other support, use kinematic and 5hz, if NOT, use static and 1 or 5Hz.

You dont input any coordinates, CORS feed reach with its position. Reach uses CORS correction to calculate absolute position and displays this in the status window.

CORS is just another “basestation”. It could easily be your own unit placed somehwere and sending its position and correction data over the internet.

The main thing you need to remember is:
-For a GPS/GNSS to achieve cm accuracy, either its RTK/PPK/PPP or any other process, it need to compare its own observation with another set of observation to calculate and rule out error. And this is usually done with a base of your own or CORS etc.

I hope i got it right. :thinking:

Makes perfect sense! I’ll go and play with it now :wink:
Thank you very much!

Just one more thing then, regarding auto-save settings in survey mode. What’s the best thing to do here? Only auto-save point if solution status = FIX, default settings (40s, 0.005m, DOP 2)? Or add FLOAT as well?

I use 1-2 sec autosave in fix mode

Wow… that short huh…

I always use 1min for my “FIX” in survey mode. I always thought a little extra time for things to settle, gather info, and then save point was worth it.

Just a note on PPP. PPP ain’t like regular RTK, it relies on modelled ionospheric correction, precise satellite orbits and a sure knowledge that the receiver is static to achieve convergence. In theory it doesn’t need an explicit base.

1 Like