What Projections and Transformations Happen Under The Hood

[DougGreen]

I'm tying to understand what's going on under the hood with projections and transformations when I use Field Maps for high-accuracy data collection. I've read through this article in detail with the lure of being able to walk away from post-processing workflows by being able to collect in high-accuracy using external Bluetooth GNSS receivers and mobile devices.

I've noticed that I can publish a hosted FeatureLayer in a variety of coordinate systems and enable them for editing so that I can add them into a map used for data collection in Field Maps. However, when setting up a location profile in the settings, the documentation indicates that the Map coordinate system is determined by the basemap it uses.

So that leads to my first question, if my layer is in a projection other than that of the basemap, and my NTRIP provider is in yet another coordinate system, is my data being projected/transformed twice?

Specific example. Let's say I have a hosted FeatureLayer that was published from pro in  GCS NAD 1983 2011 (EPSG: 6318) and then I add it to a map with a basemap that is in WGS 1984 Web Mercator Auxiliary Sphere (EPSG: 3857). Then, I set up a location profile in Field Maps using EPSG 6318 as the GNSS coordinate system (the one used by my NTRIP correction service that I don't have ability to change) and EPSG 3857 as the Map coordinate system (because that's what the ESRI basemaps are in), and the horizontal datum transformation as ~WGS_1984_(ITRF08)_To_NAD_1983_2011.

In my understanding of this, the following steps would occur when I collect a point:

1. Corrected location is received from the receiver in EPSG 6318
2. Field Maps uses the transformation supplied in the location profile to project the point to that of the basemap (EPSG 3857)
3. Field Maps sends the point to the FeatureLayer service
4. ArcGIS Online receives the point in EPSG 3857 and detects that it does not match the coordinate system of the FeatureLayer

This is the point at which I'm unsure of what's happening and I'm wondering if someone could answer my second question: Does ArcGIS Online use the top transformation from the list to project the point back to  the coordinate system of the FeatureLayer (EPSG 6318) or does it use some other default? Without knowing what's going on under the hood, I can't truly report my accuracy or the method in which my data was projected.

Thanks in advance for supplementing the available documentation so I can make this switch to more efficient data collection without loosing some of the important details.

[ZacharyUhlmann1]

@DougGreen Did you ever find an answer for this question.  I am curious not only about your question, but also generally on whether the accuracy is better when data is collected in a local/regional projected coordinate system (SP or UTM) as opposed to sticking with Web Mercator?  Or even lat/long for that matter.  So two questions - 1) any interesting discoveries on your original question and 2) do you have any insight into why collecting in say EPSG 6318 is preferable to Web Mercator or even lat/long.  

Thanks!

Zach

[DougGreen · Accepted solution]

@ZacharyUhlmann1, I have not received any additional info.

My initial concern was that ESRI indicates that generally the software will project on-the-fly for data that does not match the basemap; but cautioned "Although it is possible to edit data that is in a different coordinate system from the data frame, when high levels of accuracy are critical, it is better to project the data to a common coordinate system before editing." So I was trying to determine how much error could be introduced by not heading that caution. 

I have since discovered a simpler way to eliminate the project-on-the-fly step. I can use ArcGIS Pro to generate a vector tile package with very simple reference data in the projection I'd like to collect in and then share it to my ArcGIS Online Organization and use that as the basemap. This way, I'm only doing one project/transform from the coordinate system of my NTRIP provider to the coordinate system of the map and this method is known.

I still don't know which on-the-fly transformation method ESRI is using under the hood when you use a one of ESRI's Basemaps. I'd love to just use theirs since they are wonderful and then I could just report on the methodology of projection/transformation. As of right now, I'll have to use this method.