Allow projects to require GPS accuracy

Platform(s), such as mobile, website, API, other: website, mobile, platform.

URLs (aka web addresses) of any pages, if relevant:

Description of need:
Scientific projects, including the network of Bee Atlases supported by Oregon State University’s Master Melittologist program, require a level of precision on coordinates in order for the data generated through iNaturalist observations to be useful. Adding a rule to a traditional project requiring gps accuracy to be within a number of meters would move the problem of ensuring accuracy to as soon as possible to the time of observation, when the observer is in the best position to correct the problem.

Feature request details:

  • make a new project rule available in traditional project settings called ‘Observation must… have coordinates accurate to…’ with a field to enter the number of meters
  • adding an observation failing the rule to a project that requires it, on web or mobile, fails with a message explaining the rule

If approved, I will develop an implementation plan and build it.

This is a reasonable attempt at quality control, but as an observer using my iPhone to take photos that I later upload to iNaturalist via the inaturalist.org website, I don’t generally exercise any control over the accuracy/precision measurement of the observation. As far as I can tell it is something added automatically at the time the photo is made. It does not show on the photo or app and would only be visible once the observaiton is uploaded to iNaturalist, and even then not conspicuous. I could edit it if I happened to notice it had an unrealistic value, but I almost never look at this metric. Wouldn’t sending the observer a notification that the accuracy bubble must be reduced just lead to spurious accuracy measurements?

У меня андроид и его приложение камера регулярно искажает координаты. Я решаю эту проблему так:

  1. Приложение GPS Logger пишет трек всего моего маршрута.
  2. Сбрасываю фото и трек на компьютер.
  3. На компьютере, с помощью программы GeoSetter проставляю к фото правильные координаты.
    У меня ещё есть два фотоаппарата, которые я беру с собой по ситуации. Схема та же, только в начале маршрута необходимо синхронизировать время смартфона и фотоаппаратов.

If you upload through the webpage, then you have full control of this. Simply click at the location of the observation, it shows you the location with a circle around it, which is the uncertainty. You can drag the circle to change its size as you deem appropriate.

The problem is you’re likely to get folks just changing the accuracy to meet the project requirements, without actually looking at whether the circle encompasses the location where the observation was made.

When folks type in a location name, and then simply accept the lat/long suggested by iNat, I’m not sure how the accuracy gets set, but I see plenty of such observations where I’m 100% positive that the observation didn’t occur within the accuracy circle shown. The accuracy may fall within a specified limit for a project, but that’s no guarantee that the coordinates + accuracy figure reflect where the observation was made. In other words, the feature request may be an attempt at quality control, but it may just provide a false sense of security.

You are lucky, an iPhone adds accuracy/precision measurement. When adding photo’s from an Android phone this value is not included in the EXIF data GPSHPositioningError

i’ve never understood why people impose this sort of requirement on themselves. if you can control and standardize the methodology for how people record locations, then that sort of requirement is fine as part of defining that methodology, but there is no standard methodology reflected in the locations of iNaturalist observations generally. so imposing such a requirement without any other context of methodology seems pointless. in my mind, it simply reflects a fundamental misunderstanding of how iNaturalist data is actually recorded.

Thanks, I didn’t realize this wasn’t as workable on Android.

I think you’d be better off just excluding observations with very large accuracy circles rather than ignore all those without recorded precision. In my experience pins with no accuracy set seem to quite often be fairly precise. Sure, you can’t know exactly how precise, but this is true for observations with accuracy as well, where in practice the true location is fairly often outside the accuracy bubble anyways due to GPS issues or user error. I think you either have to accept there will be some noise in locations if you use iNat data or eschew it all together.