I have a dataset (in excel) of species with the coordinates in which they were sighted, collected by many different people over decades. Aside from coordinates, the dataset also specifies a datum (Nad_27, Nad_83, WGS_84). I added the data to ArcGIS.

So my fields look like:

enter image description here

The HorizontalDatum column is confusing me. This one excel file says the dataset is composed of 3 different datums depending on the record. Although the original datum is said to be one of those three for any given record, can I set the datum for the entire dataset to Nad_83 in arcGIS and ignore this column? Will that still be accurate? I don't understand datums or transformations enough to be sure.

If not, will I have to split the dataset into 3 shapefiles for records of each datum?

These data points are all within the same province. I just want to keep all the records in the same shapefile.

To summarize and clarify: I will need to set a geographic coordinate system for this shapefile manually, as it's coming from excel data. Can I set the entire shapefile to Nad_83, or should I split it up into 3 shapefiles, and project those with the 3 "Horizontal Datums" stated in the excel field? This would be less than ideal... I'd like to be able to query the whole dataset in one go.

  • Wow, this is one of the more interesting questions I've seen on here. I don't think there's a really good or easy way to do this. My strategy would be to split the shapefile apart based on the datum, then reproject all the resulting shapefiles to the same coordinate system and datum. Then, merge the shapefiles back into one layer.
    – Thomas
    Oct 13, 2015 at 20:20
  • This depends on how accurate you need to be. The shift between NAD27 and NAD83 with those sample lat/longs is around 25 meters, give or take 10 meters.
    – Mintx
    Oct 13, 2015 at 20:40
  • 1
    @Mintx the shift depends on what coordinate system you're using...in UTM, NAD27-83 differences are around 200 meters in northing and 20-25 meters or less in easting.
    – mkennedy
    Oct 13, 2015 at 21:44
  • @mkennedy True, I was judging solely on the lat/long values in the table above.
    – Mintx
    Oct 13, 2015 at 21:50

1 Answer 1


As a rule, splitting the the Excel data into two groups would be better practice.

  1. NAD83 and WGS84
  2. NAD27

As mentioned in Mintx's comment, the NAD27 data will be identifiably different from the other two datums when the points are plotted using the Lat/Long values. This difference may or may not matter to you. As an aside I find it unlikely that the sightings were as accurate as the level of precision seen in your spreadsheet - listed as 5 decimal places. And the given "Coordinate Uncertainty" error in meters for the NAD27 sightings strongly supports that assessment (thousands of meters of error!).

Your final Shapefile ought to be given a projection meeting the current reporting standards for the project they support. In this case I expect either NAD83 or WGS84; with the latter preferred by a great many GPS and mapping interfaces available, so it might make sense to have all the data be in WGS84 when you are done.

TL/DR - It looks like WGS84 is what the most recent data are being collected in and, given the reported error of the NAD27 data, I would create one Shapefile from the whole data set in one go (the error introduced by not compensating for the datum shift is completely washed out by the original reporting error).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.