How to stop Survey123 to attempt to get user location? And how to use predetermined XY on a survey?

[Gustavo_Souza]

For almost two years, we managed to create a map where the user clicks on a point, which opens a Survey123 form. The geopoint was set to read-only, and the URL had the center on XY coordinates (&center={Point_Y},{Point_X}). It worked flawlessly.

Until recently this system became unreliable. Not only new surveys do not work, but older ones are having the issues. They often try to get to the user's location. Or just center somewhere near Los Angeles. Or in Null Island. In different computers and web browsers. 

I figured out that the "&center=" URL thing is not reliable anymore to add points on that location (despite always working that way). So I read a bit about the calculations for geopoints, but it still does not work. 

Even by typing a coordinate on the default collum, for the geopoint field on the .xlsx form, (such as "48.858328 2.294513") , I get the message on the survey with no point selected:

- Can not find your location. Please check your browser to ensure that your location is shared

On top of that, there point And if I allow the location, then it points to my house.

I have then two questions:

1. How can I ensure that the survey 123 will never attempt to overwrite the pre-defined values with the user location? Disabling the web browser permission is not a good option because many people use it, and its unrealistic to expect that everyone does it (and again, even when disabling, sometimes we get the error message).
2. Even when writing a coordinate on the default value for the geopoint, the survey still does not ready it, sends me to LA. (and if I try to submit, it will go to 0,0). I tried using calculation too, but even if the simplest test does not work, I wonder what I am doing wrong.

Thanks for your help!

[SMH-Rio]

Taking advantage of the topic: is there any way to send a response without a geographic point (attribute without geometry)?

I have a workflow where the location of the main layer is not necessary, since we collect the locations on repeat. In the various tests I did, Survey123 always sends a geographic point to 0,0 (default=null, appearance=press-to-location, esri type=null, among others).

Is this the expected behavior or is there any way to prevent the creation of a geometry associated with the record sent by Survey123?

[DougBrowning]

Since the data is stored in a AGOL Feature layer it must always have some sort of geometry.  Over the years I have found that when you go to use the data in maps, dashboards, whatever it is always easier to have geometry in the end.  Even if the geo is in a repeat it can be a pain.  Esri has never been very good with tables so it starts to limit you fast.  I would just default the form to where they opened it or pick one of the repeats.  Also If you have all these parent points at 0,0 your spatial index falls apart and performance gets bad.  We use geometry for QA all the time.  We have set locations where collectors are to go also. With the form having a location we find issues all the time where they tapped the wrong preloaded point, drove to the wrong location, said they were at x when they were really at y, and we have proof they actually went.  I personally see no disadvantage to having location on there we just use it so much.

Hope that helps

[SMH-Rio]

Yes I agree. This limitation of tables within ArcGIS is old. For example, WAB was terrible with them.

I also think that taking one of the repeats (the last one, in this case) is the best option. However, this issue of performance loss due to everyone being at 0,0 is new to me. I will certainly take it into consideration in future forms.

Thanks for your help

[DougBrowning]

Spatial index can grind you to a halt for sure.  Last job I had to automate checking extents in our SDE since it would cause so many issues and people kept adding 0,0 all the time (cause spatial in NOT important lol).

I cannot find great graphics but I will try.

Arc takes your data extent and chops it up into a checkerboard like this (kinda).

It does this so it can draw just one of these boxes at a time.  So if you are zoomed into a above it just draws box 02.  This is very efficient - why draw box 15 when it is not even on the screen.

What happens when you have 0,0 points and work in North America is now ALL your data is jammed into one single box.  This means that even when zoomed in Arc has to draw the entire dataset!  So basically it is unindexed at that point.  You are drawing every single point or vertex every single zoom or pan.  Gets real slow real fast and I think it is most of users performance issues when in Esri land.  I have brought many old datasets back to life just fixing this while consulting.

I hope that made sense.  For sure 0,0 points are a killer.  I look for and fix them then recalc extent, then recalc the spatial index.  There are ways to do this in SDE and AGOL.

[SMH-Rio]

Yes, this explanation makes perfect sense. Even working in South America and with "small" sets of data, this is certainly something to be taken into consideration when building forms. Thank you for that.

However, I still think that Survey123, with geopoint, should be able to send records without geometry (null geometry), as is possible with geoshape and geotrace, especially considering the issue of respondents' personal data (in addition to the issue of spatial index that you very well mentioned).

I hope this will be considered in the future.