Hi Scott,
So I looked at the data and the output file from NRCAN. On the bright side, the warning about the difference between the a priori (rinex header) position and the NRCAN estimated position is based on the height difference between those two (the lat/long were both less than 10m). Non-corrected GPS height can be significantly worse than 10m depending on ionosphere and troposphere characteristics. Another possibility is the log started up right when you turned on the unit and it hadn’t stabilized on its solution (thus it logged its initial a prior point with few satellites and bad resolution on the height). When I run your observation file through RTKLIB in single mode it gives me an uncorrected height of 358.096m, about a meter off of the NRCAN estimate. This leads me to believe the issue is really with the initial position written to the file by the GPS. Let the GPS run for a few min before starting the log next time and see if it clears up the issue.
Now, how you move forward on getting your absolute position really depends on what you are doing and the accuracy you need. A lot of GIS work is okay being 1-3m accurate. If this is the case, you can run PPP and get the coordinates in ITRF2014 and assume they are equivalent to WGS84. You mentioned GCPs, which makes me think you might be running drones for orthometric photos or photogrammetry. If you need high relative accuracy, but are okay with absolute being 1-3m, you can still use the above. If you need survey grade accuracy throughout all of your work, you are going to need to do more to make that happen. WGS84 and ITRF2014 are reference frames that are not tied to any one tectonic plate so they are only valid in the moment of time that they are collected and your local plate will continue to move away from that position. You will need to use the tectonically tied NAD83(PA11). Since I don’t work in your region, I am unsure if NRCAN automatically selects NAD842(PA11) for the pacific plate -> you will need to research to make sure this is the case and its not giving you a result in the north american plate.
I see that there is a JPL operated CORS near your location (named MKEA). It has its coordinates in NAD83(PA11). I see its sampling rate is low (30s), but you could PP the location of your base station against that in static mode and get a high quality position for your base station in NAD83(PA11) and your work is then tied in to a CORS monument. Great for repeatability of your accuracy. Once your base station is in NAD83(PA11), then all of your GCPs will be in the same reference frame (make sure you choose it in the emlid survey function). This will require likely two visits to the field, one to establish the base station monument and the next to set your GCPs once you have PPed the base station to the CORS. The other alternative is to do it all at once, GCPs relative to the base, while you log the base (starting in average single mode), and then you will have to correct the base by PP after the fact and then adjust each GCP by the deltas between the original average single base point and the PP point you get later. If it’s a few GCPs and you have good PP software, this might save you field time at the expense of more office work and doing more adjustments/conversions. Well after writing this section about this CORS, I looked up the actual distance and it seems you’re approx 70 km from this station, which is starting to hit the upper range for reliable PP. You’d need ideal conditions, patience and probably a good PP software package to help out with this. And I would make sure you are able to actually PP the base before trying to set GCPs.