Support Website Contact Support Blog

GCPs issue for coordinate system

I am new at this so bear with me.

We have this old horse farm we have previously field surveyed via total station and have the buildings, boundaries etc located,  we wanted to do a fly over to get all the locations of the fields ponds etc on the parcel of land (~150 acres of land total) so we flew it @200’ with a dji phantom 3 advanced(~1500 images total)- not worried about the topography, but just would like a good mosaic picture of said area so i can insert it into my CAD at 1-1 size (yes image is freaking huge ~1gig ) and sketch -within a foot or so- where the tree line is, the bodies of water etc. site took 4 flights, and i have good coverage, the issue is. i got my control point coordinate on the mass. state plane coordinate system within tolerances. using google earth, fema and survey- and it is all checking …however when i put said coordinate in for a gcp, and then using the geodetic listings from my cad to get  the mass state plane coordinates for the other gcp’s…all EXCEPT the intial gcp are off by a good 30 to 50 feet, it is like it is rotated a bit counter clockwise

the drawing i am using is based on the 1955 magnetic north(mass state highway layout), and i adjusted the cad file  to get grid north (mass state plane reference) and the declination angle to grid from my bearing system…(i am using 2017’s declination angle i looked up for the town i am in)

so which coordinates should i use, my cad x(E)y(N) coordinate or the geodetic results. and if i need to use the geodetic, which declination angle should i use- the 1955 one or one from 2017 (about a degree and a half difference in angle)

one possible thing it might be- the system MIGHT be on the mass state plane grid north now and my declination angle is all wrong… (i’m gonna have to fiddle with my cad file to check that out- list out what the coordinate in pix4d says is a gcp and then put that coordinate into cad then see what the angle is…

 

another issue i am seeing- and i see this on google MAPS as well

i have physically located most of the main fences and buildings on the parcel via survey.

if i overlay a screen shot of google maps in the area,(and a partial mosiac i have run via the cloud software) blow it up to real size, position it so that the buildings, road edges etc overlay perfectly(within an inch or two) for placement on the west side of the parcel, as i go EAST…everything is off… by ~50 feet east once i get about 900 feet from the west road. buildings, , fence corners that we can see fine on google map image are all skewed east from where we located them, YET…once i get to the far east boundary of the property(~2500’ roughly from the west roadway boundary) (a River)…it all lines up again(property lines are pretty much center of river).

any idea why? both google maps and pix4d mosiac of the same area shows this same issue

i imagine there should be a way i can use the record distances i have and the surveyed distances i have to manually adjust the pix4d image to match properly…i’m still figuring this all out…

 

Hey Leonard, 

We are not sure we completely understand the issue you are mentioning and we would like to ask you some more questions: 

  1. What was the accuracy of the acquired GCPs?
    If the GCPs were acquired from a Web Map Service (like Google Earth for instance) then the accuracy needs to be properly adjusted. 

Check this link for more info: https://support.pix4d.com/hc/en-us/articles/202560149

  1. How many different GCPs do you have? Did you also acquire some on the field as well?  
    Which ones are the initial ones that are correctly placed on the map? 

  2. So you are using the orthomosaic that was produced in Pix4D software and then importing it into a CAD software for digitization, is that correct? 

  • To reply to your question, a priori you shouldn’t need to transform the coordinates and correct for the declination, you can simply import the GCPs in Pix4D software, mark them and then reoptimize. In the drawing you have, if they indicate the declination then this should be used to define the magnetic north if a compass is used, which in this case should not be necessary. 

Moreover, GPS, as well as Web Map Services, most likely use WGS84 which is based on the true north. Here you can also find the .prj file of the NAD83 projection:  http://spatialreference.org/ref/epsg/26986/ 

This article has more information on magnetic and true north: http://geomag.nrcan.gc.ca/mag_fld/magdec-en.php 

  1. Have you marked the GCPs and adjusted their accuracy in Pix4Ddesktop? If you want the GCPs to be taken into account for the overall reconstruction of the project, then the GCPs should be marked on the images.  

Check this article for step by step instructions: https://support.pix4d.com/hc/en-us/articles/202560769 

 

  1. Can you send a screenshot of the drawing? 

  2. Could you also share with us the project that was processed on the Cloud? 

  •  Since you would like to overlay the google area then we would not recommend the screenshot of google maps since there might be some distortions that could explain the displacement. On the other hand, if the orthomosaic does not have GCPs marked then it might not be georeferenced correctly. 

 

  • Alternatively, you could export the Google Maps Tiles and .kml  and open it in google earth for instance to compare. You can see here how to generate them: 

https://support.pix4d.com/hc/en-us/articles/206860936

  • Since you would like to use the recorded distance, then you can add a scale constraint. You will find all the instructions here

Hope this helps,