Please document converting geographic coordinate systems

[czender]

updated description

• currently there are two pages:
  • https://github.com/terraref/documentation/blob/version1/geographic-coordinate-data.md
  • https://github.com/terraref/documentation/blob/version1/spatial-reference-system-data.md

need to describe the multiple coordinate reference systems at Maricopa

There is a lot of confusing content in these related issues:

• Convert gantry coordinate system from right handed (x positive to left) to conventional (x positive to right) Convert gantry coordinate system from right handed (x positive to left) to conventional (x positive to right) computing-pipeline#109
• converting lemnatec scanner metadata to geographic positions Converting lemnatec scanner metadata to absolute geographic positions reference-data#32
• Create polygons for sites table in BETYdb Create polygons for sites table in BETYdb  computing-pipeline#174
• Create custom spatial reference systems for PostGIS Create custom spatial reference systems for PostGIS reference-data#51
• point cloud origin (@ssolmazhajmohammadi) The origin of the coordinate system of pointclouds reference-data#44
• georeferencing hyperspectral data Georeferencing hyperspectral data computing-pipeline#144
• Store location and field of view metadata associated with each file in Clowder Store location and field of view metadata associated with each file in Clowder computing-pipeline#101
• A protocol for checking that scans align with plots A protocol for checking that scans align with plots  computing-pipeline#153

---

original issue

How do we (@hmb1 and @FlyingWithJerome) convert position on a gantry sensor to geographic location?

For example, this file
$DATA/terraref/MovingSensor/SWIR/2016-03-05/2016-03-05__10-08_49_485/e2eaa60d-9f3d-4af9-9fe0-8f01731b70d5_metadata.json
contains these metadata:

"gantry_system_variable_metadata": {
      "Time": "03/05/2016 10:08:49",
      "Position x [m]": "195.1725",
      "Position y [m]": "7.95",
      "Position z [m]": "0.599",

Where "exactly" is x=y=z=0.0 and what approach should we use to convert offsets from that origin to absolute geographic location (lat/lon)? And, for imagery, how do we determine the location of each pixel (Sample and line in ENVI-speak) in (x,y,z) coordinates within the image whose nominal position is described (I assume) by x,y,z above?

A worked-through example would be tremendously helpful.

Thanks!
Charlie

[nfahlgren]

Hi all,

@solmazhajmohammadi and I were just talking about this this morning. I cannot remember, but does the gantry have a GPS system in addition to the xyz positioning system? If not, what do you all think about the possibility of putting georeferenced markers across the field (now or before sorghum planting) and using them to calibrate the xyz system? I think Solmaz has some ideas about what kind of markers we might be able to use (potentially make) for both this purpose and for camera calibration/testing.

[dlebauer]

where is the sensor?

[0,0,0] is the SW corner of the gantry system. The [x,y,z] you give is the location of a reference point on the camera box. In addition to the location of the camera box, we need the [x,y,z] offset of the specific sensor relative to the camera box reference point.

@rjstrand may be able to provide some guidance (and perhaps the actual coordinates) of each sensor within the camera box

@pless, @ejdelp33 may also have some suggestions

[dlebauer]

There is an additional section in the metadata file with relevant fields ... some currently unused, and we probably need to add a field for "location in camera box Z [m]",

    "sensor_fixed_metadata": {
      "sensor manufacturer": "Headwall Scientific",
      "sensor product name": "SWIR",
      "sensor serial number": "Todo",
      "sensor description": "Todo",
      "sensor purpose": "Todo",
      "location in gantry system": "camera box, facing ground",
      "location in camera box X [m]": "Todo",
      "location in camera box Y [m]": "Todo",
      "optics manufactuer": "Stingray optics",
      "optics name": "SR-1200-030 F/1.3 0.48-2.5um",
      "optics focal length [mm]": "25",
      "optics focus apperture": "2.0",
      "Output data format": "hypercube with preview image and additional information provided by sensor",
      "Calibration available": "todo, mail to bob"
    },

I've pasted the entire metadata file here: https://gist.github.com/dlebauer/b47218fc842e1e2bb46a

[dlebauer]

@rjstrand just provided approximate field coordinates (from his cell phone GPS)

Here are locations from my cell phone. There is a little noise in it. Additionally, the north readings could be off a bit because there was a big chunk of metal in the way... ;)

-Bob

SE Corner 33° 04.470' N / -111° 58.485' W
SW Corner. 33° 04.474' N / -111° 58.505' W
NW Corner. 33° 04.592' N / -111° 58.505' W
NE Corner. 33° 04.591' N / -111° 58.487' W

The SW Corner is the [0,0] local reference point so that North and East are positive.

[dlebauer]

@solmazhajmohammadi not sure why this was closed, but I don't think it has been resolved.

[TinoDornbusch]

We will provide reference coordinates this week gantry coordinates <>GPS. Mathematically two are enough, but we will do more. Is that enough for you to derive the transformation matrix / look-up table?

[solmazhajmohammadi]

I accidentally closed this one, and reopen it here:
#9
For hyperspectral data, I think using external trigger is the best approach to convert the pixel positions to the GPS/Gantry coordinate system.
For converting only gantry's location to geographic location, we have not had GPS location of (0,0).
However, I think it would be more accurate to use only gantry coordinate system, instead of recording rail temperature and modelling the location based on the temperature change.

[solmazhajmohammadi]

In a case you need initial conversion matrix between GPS and Gantry Location you can use : Gantry_Location=S * GPS
(Note: it is an initial linear model, and might have up to 5 cm error)
Linear modelling between GPS coordinate and Gantry location:

Gantry_Location=[ Gantry_y ; Gantry_x]

GPS=[ Lon ; Lat ; 1]

S =[
-93.5378801063054e+003 0.00000000000000e+000 -10.4738844723827e+006
0.00000000000000e+000 110.853602042202e+003 -3.66643212115991e+006]

Gantry_Location=S * GPS

[solmazhajmohammadi]

Swir data with external trigger "ON" in the metadata, should be corrected. Pixel dimension is 1mm in the long axis and between 1-2mm in the short axis depends on the working distance.
I took a quick look at the data, I guess the setting has not been changed in the metadata. @smarshall-bmr can you please make sure that the "sensor_variable_metadata" is corrected before the new scans.

[ghost]

Documentation https://github.com/terraref/documentation/blob/version1/geographic-coordinate-data.md

[max-zilla]

https://terraref.gitbooks.io/terraref-documentation/content/user/geospatial-information.html

Conversion methods are described in detail here.