1

I currently working on the data set "Wolfcamp Aquifer", but the coordinates in the data set is just a relative coordinates. So, is there anyone who has the original coordinates?

nmtoken
  • 13,355
  • 5
  • 38
  • 87
Nicolas
  • 11
  • 1
  • You may have to look into the sources given in http://wiki.stat.ucla.edu/socr/index.php/SOCR_061708_NC_Data_Aquifer – AndreJ Apr 15 '18 at 06:11
  • Thank you so much. I check the link and the reference but, unfortunately, it's still the relatively coordinates. I was wondering how can I get the exact coordinates if I randomly assign a reference point to the relative coordinates. – Nicolas Apr 15 '18 at 15:26
  • You would need the exact coordinates of the reference point to create a custom CRS like in https://gis.stackexchange.com/questions/83861/using-customized-coordinate-system-in-arcgis-desktop . Note that map units have to be kilometers. – AndreJ Apr 15 '18 at 16:55
  • If you want to type in (or image-to-text) the numbers, they are in the appendix of this PDF: https://www.researchgate.net/publication/236564521_Geostatistical_analysis_of_potentiometric_data_in_Wolfcamp_aquifer_of_the_Palo_Duro_Basin_Texas – Spacedman Apr 15 '18 at 19:19
  • I've OCRd and fixed it up: https://gitlab.com/snippets/1710174 – Spacedman Apr 15 '18 at 20:13
  • The map from the data in the geoR package shows one point clearly not in the same place as the map from the paper. Whether there's a transcription error in the original map from the original data or from the data to geoR I do not yet know... – Spacedman Apr 16 '18 at 15:19

1 Answers1

1

I found a map of the locations in an ICOTS8 (2010) Invited Paper by Harper & Clark called "AMARILLO BY MORNING: DATA VISUALIZATION IN GEOSTATISTICS" - you can probably search online and find this. I extracted the image, and georeferenced it in QGIS.

Then I identified a few of the sample locations on the map and got their lat-long coordinates. I then matched those up to the "relative" coordinates from the data set in R. So now I have a table of:

 x_lat y_lat x_relative y_relative

coordinates for seven of the points. I can then run ogr2ogr on the command line using the coordinate pairs as control points - it looks like this:

ogr2ogr \
 -tps \
 -gcp  -233.72172 -115.83894 -11636657   4003094 \
 -gcp -30.54492  115.72629 -11392916   4286809 \
 -gcp   133.79896   95.12698 -11190784   4260256 \
 -gcp   174.71182  -27.48198 -11143259   4107547 \
 -gcp   18.74859 -130.78953 -11331381   3985634 \
 -gcp   -29.96271  -37.89631 -11391316   4099389 \
 -gcp  169.09138   51.54656 -11147872   4205489 \
 -f "ESRI shapefile" wcreftps.shp wc.shp

On each gcp option, the first two numbers are the "relative" coordinates, and the second two are in EPSG:3857 (Google Mercator). wc.shp is the R data saved to a shapefile.

After running this, the shapefile wcreftps.shp seems geolocated correctly - here it is with an OpenStreetMap and county boundaries.

enter image description here

The coordinates, in the same order as the 85 coordinates in the wolfcamp object in the R geoR package, are in this pastebin:

https://pastebin.com/zdXMdnZ7

Caveat: These coordinates are as precise as my clicking on the map - they are not the original measured coordinates but should be close enough for most purposes that use the whole map - don't zoom in on one point and expect to see the exact location.

Spacedman
  • 63,755
  • 5
  • 81
  • 115
  • Wow...................Thank you so much .............It's amazing..I'm new to this field, your answer is really helpful. – Nicolas Apr 15 '18 at 22:06
  • I think I need to learn QGIS now. – Nicolas Apr 15 '18 at 22:12
  • I may try and compare the coordinates I got from this with the ones I got from the original paper (linked in my comment to the question). – Spacedman Apr 16 '18 at 07:23
  • I think they didn't give the exact coordinates in their original paper, right? – Nicolas Apr 16 '18 at 19:38
  • The paper I link in the comments gives both lat-long and offset-in-miles from a point coordinates. I'm not sure how "exact" these coordinates are, plus the lat-long appear to be in a 1958 military coordinate system and I can't find the technical manual online so I don't exactly know the coordinate transform parameters to get from that to EPSG:4326. – Spacedman Apr 16 '18 at 20:44
  • And I've now completed OCRing and correcting the "relative" coordinates from the paper and there is definitely a wrong location point in the geoR data set. Once I've tidied things up I might report it. – Spacedman Apr 18 '18 at 16:14
  • @Spacedman I get rather good results with +proj=tmerc +lat_0=0 +lon_0=-102 +datum=NAD83 +x_0=0 +y_0=-3696553 +k=0.9996 +units=us-mi +no_defs except for point ROO-014 which is off by -2,29437 miles in x. Might be a typo. – AndreJ Apr 18 '18 at 18:56
  • There's definitely one point, the 49th, that is way off in the geoR dataset compared to the data in the paper. In the wrong county. – Spacedman Apr 18 '18 at 21:30
  • @Spacedman, so the coordinates you post is the right one? – Nicolas Apr 18 '18 at 22:04
  • Yes, I think so. – Spacedman Apr 19 '18 at 06:32