Automatic GPS Averaging

[LarryWiebe]

Is there a way to start GPS averaging without any user input when the survey is opened?  The would ensure that the user always uses the averaging.

[JamesTedrick]

Hi Larry,

Currently there isn't a way to trigger averaging manually.  Can you provide more information on your workflow's averaging requirements? 

[LarryWiebe]

Hi James,

I am setting up gas meter inspections for roughly 250,000 meters over Southern Manitoba. The inspector has a assigned inbox of meters to navigate to. When an inspector gets to the meter, they answer roughly 20 questions about the state of the meter and we would like to take photos (my other question regarding support for photos on inbox based surveys). We are hoping to refine our meter locations and have found that the averaging needs to be used (iPad) to get a position worth using. Currently the inspector will have to press and hold to start the averaging, which is fine, but will get missed knowing the nature of field work and that is where my request came from. Cellular data is ok in the cities, but nonexistent in the rural areas so everything is geared to be disconnected.

Let me know if you want more info…

Thanks,

Larry

[JamesTedrick]

Thanks Larry,

I was curious on what you were thinking of in terms of the number of location observations needed to get to a point worth using - is it a set number of observations, a requirement that a set accuracy threshold is within range of the average (say, that 5m is within a 95% confidence interval of the mean), or some other metric?

[LarryWiebe]

Hi James, When it really comes down to it, the inspectors will only be at the location as long as it takes to fill out the survey. Ideally that will take 20-30 seconds which I would like to use to get the best averaged position. There will be inspections where the user is there for only 5-10 seconds. The position being acquired to help refine the meter location within the land parcel, which is currently based on the parcel centroid. So the point is not the focus of the survey, but a very helpful piece of data for the next time an inspection is performed. I deal with the fuzziness of GPS locations all the time and with the iPad 5m is pretty much the best you can expect, but the averaging will help get us to the 5m more of the time in whatever time the inspector occupies the location.

That is a practical metric, but not very mathematical…

Larry

[JamesTedrick]

Okay, thanks for the info.  Aside from using averaging in this way, are you setting the horizontal accuracy requirement for the point (Geopoints—Survey123 for ArcGIS | ArcGIS )?  This will automatically continue to collect the location until the accuracy requirement is met.

[LarryWiebe]

Yes, I have it set to 5, but I prefer an averaged 5 over a spot 5…

[Anonymous User]

lawiebehydro Do you mind explaining what the difference is here? 

[LarryWiebe]

Hi Tom,  most GPS units, especially tablet based units are not as accurate as they may report. A single GPS reading from an iPad that claims to by within 5m by Survey123 has a much higher potential error that an averaged point over 20 or 30 seconds (positions).  I do not know the algorithm used in Survey123 to get the horizontal accuracy, if it uses averaging until it get 5 positions within the tolerance I am fine with the option.  If it calculates an accuracy on each point and then stops when it hits the first point that falls within tolerance without averaging the points it has been collecting, the position is subject to a poorer absolute accuracy.  If someone can comment on the Survey123 horizontal accuracy methodology, that may negate this whole post...

Larry

[JamesTedrick]

Hi Larry,

Survey123 uses the location information as provided from the operating system's location, including the accuracy information. Depending on the application, you can be receiving a variety of location signals to determine location (Apple's page on Getting the User’s Location shows some of the options)- Survey123 engages the device's actual GPS if present.