This is the basic case for all network sites. Twice a year iNat sends exports of observations made in that site’s region of interest to the appropriate network partner (eg ALA in Australia, CONABIO in Mexico), and it includes true locations for all observations that are obscured due to taxon geoprivacy. It also includes true coordinates for obscured/private observations made in the region by users who are affiliated with that network site and who have manually obscured/made private those observations.
2 Likes
Thank you for clarifying that it’s the GBIF version of iNat data you are concerned about. I agree. I also don’t know how the iNat data set compares with other GBIF data sets where, for example, museum specimen data may have location data withheld beyond the second or third administrative level for threatened taxa. I imagine there are all kinds of GBIF data where one would need to contact the data source to get sufficiently precise data for a particular research use.
1 Like
Is there an iNat standard for how granular location needs to be for research grade / research purposes? My locations are not granular.
I’m quite sure my new camera doesn’t have GPS, and my current camera has it turned off because my understanding is that data can be pulled from photos*. As a result, my locations are all approximate unless I’m at a marked intersection, but I do try to place the observation in the general area. I also sometimes having trouble locating where I was and moving the dot (which is not super sensitive to navigate) where I need it to be. I usually decide after a bit of effort I am close enough. Animals move. Birds move a lot. And I don’t see any way around this, really, although I’m open to suggestions.
For people who are worried about sensitive species, you can enter the exact location and switch it to “obscured” even if iNat won’t automatically obscure it. I often do this when I have a cluster of observations from (clearly) the same location, and one is an endangered bird.
*Does anyone know if that’s true from iNat-uploaded photos, btw?
There is no requirement for how precise a location needs to be for the observation to be verifiable (eligible for Research Grade on iNat). The only important thing is that the true location is somewhere within the “accuracy circle”. Place the center point where you think the subject of your observation was, and expand the circle so that it includes everywhere it may have been. This varies from observation to observation; I can locate some of mine to within a meter, while others require several kilometers.
4 Likes
i don’t think this is actually a thing.
people use the accuracy value to represent all sorts of different things. if someone wants to use the accuracy value to represent the outer limit of possible true locations for the subject of the observation, that’s fine, but i think it would be wrong to hold others to this particular usage standard.
I don’t understand what else it could be used for. It’s meant to represent the precision, or “accuracy”, of your stated location, so how is using it in that way an unreasonable standard?
1 Like
the truth is that the accuracy value is used to capture many different things.
1 Like
Just because it can mean those things doesn’t mean they are correct use cases. The only two out of that list that I use are “a radius of uncertainty within which the subject is located” and “radius within which a subject moved”, and only the former of those is the actual intended use. Many people do use the accuracy circle incorrectly, but when someone asks a question about it there’s no reason not to tell them how to use it correctly.
2 Likes
the only real choices you have are whether you want to reveal information or not. if you want it to be impossible for others to discover the location of a particular observation, the only way to really achieve this right now is to never post the true location of the observation in the first place. it doesn’t really matter if the location is unobscured, obscured to 0.2 degrees, or obscured to some other arbitrary boundary. if you post your true location to iNaturalist, it is possible to discover your true location.
iNaturalist does its best to balance privacy / information security and sharing knowledge, but at the end of the day, this is a really difficult – if not impossible – task.
where did you learn that these are the “correct” usages?
i don’t know the exact context of this, but hopefully this wasn’t someone suggesting that iNaturalist itself wouldn’t allow you to obscure. if there are other factors that demand this like some sort of freedom-of-information regulation, then i guess go with the laws / regs…
if you only changed the accuracy value in the observation as opposed to changing an actual setting your GPS device, then just make sure that you didn’t accidentally also load more accurate coordinates in the metadata of your images.
3 Likes
I don’t know. Probably as the result of accumulated knowledge from reading the iNat help pages and the forum. The important fact is that an observation is about an organism, not about the observer or their device, so any information that doesn’t relate directly to the subject of the observation is irrelevant. This excludes the majority of your possible use cases for the accuracy circle.
2 Likes
I think you are both making different and equally valid points - what the accuracy value means in concept vs. how it gets populated in practice.
In concept, the pop-up help when editing an observation says this:
Accuracy of the coordinates in meters. You can edit this by click the “edit” link above or by clicking on the map marker and dragging the handles. Try to make the circle big enough that you’re sure you were somewhere inside it.
I agree that the intent is to express the error radius around the stated coordinates, such that it contains the “true” location of the observation.
That said, I also agree that in practice that value gets estimated in many different ways, and some of those ways may not be compatible with the intended use of the value. There has already been discussion in this topic, for example, of photographs taken at a distance, and that the intent is to encompass the location of the photograph’s subject, not of the photographer, when the two are very different.
2 Likes
can you point to where this is expressed?
if this is actually the intent from those who designed the system, then the system itself often seems to be doing something very different. even in the tooltip text that you quoted above, “Try to make the circle big enough that you’re sure you were somewhere inside it” seems to me to clearly indicate something different than what you’re describing.
https://forum.inaturalist.org/t/deliberately-posting-false-or-inaccurate-information/42082/16
https://forum.inaturalist.org/t/deliberately-posting-false-or-inaccurate-information/42082/17
https://forum.inaturalist.org/t/deliberately-posting-false-or-inaccurate-information/42082/36
these are just users expressing what they think the accuracy value should represent, right? is there anything that confirms that this is actually the intent from the overseers of the system?
Thanks.
Hence the legitimacy of people posting deliberately false coordinates: not a “fraud”, but rather the only sensible option. With false coordinates + no obscuring, they have the best of both worlds: true location not discoverable + false coordinates conveying useful information.
if you need as close to 100% location protection from poachers and stalkers as possible, you just need to never post an observation at all.
if you can accept some risk of discovery, manually setting coordinates that are false by themselves but when paired with a large accuracy value could be interpreted as encompassing the true coordinates is a reasonable compromise, in my opinion, as long as you make sure you don’t accidentally still retain the true coordinates in your image metadata or observation notes, etc. taking measures to protect the exact date of the observation would also help keep the location more secure.
whether this makes the data useful or unusable is debatable and probably depends on the specific use case.
i think researchers who have or can get access to the true coordinates would probably prefer that you post true coordinates and use the system’s obscure functionality. but if i were a poacher or stalker, i personally would also prefer this.
the obscure function is not entirely useless for protecting locations, just as closing a door without locking it still offers some deterrent over leaving a door wide open, but my concern is that it can make people think their locations are more protected than they actually are, leading them to be less critical about how and what they share.
1 Like
I think this is a case of less-than-optimal wording, rather than an attempt to say the observer’s location matters. The impression I get is that whoever wrote this assumed the observer would be standing very close to the subject of the observation, so the two locations would be interchangeable. They used “you were” rather than “the organism you were looking at was” in the interest of concision and to communicate more easily with casual observers.
4 Likes
That sounds like a job for a spreadsheet or data sheet more so than iNaturalist observations.
iNaturalist obs go into Biological DataBase of South Australia, and from there to Atlas of Living Australia, NatureMaps etc etc. From of these, apart from pretty dots on a map, I can download a .csv file, ie a spreadsheet. There is a field for location accuracy and in some cases I know both location & accuracy are wrong, eg fish on top of a mountain to 10 meters. But unknown unknowns, I can’t always be aware.
I have just checked and that is what is happening for my case. Thank you.
But then how could this be solved?
I would say that it is relative.
Users often post observations with a rather broad radius but centroid is precise. So, do they act this way for a precise scope? Maybe it is because the organism has been photographed in a private property and they hope that doing this could protect their privacy. And this could be good.
Alternatively, they could hope to hide that the organism is not wild just by broadedning the radius. This is far less good.
A some kilometers radius in a wild area presumably without certain reference point can be acceptable. Differently, for observations in urban environments, I think that they should not have a radius larger than 100 m. Otherwise, It is conceivable that the user does not put the necessary care in providing acceptably precise informations.
It can also happen that the centroid is unprecise to very unlikely. In these cases feel free to flag the observation. If these users are advised once, twice or more times but go on acting this way, consider the possibility to ask the help service of iNat.
To be honest, I think that in cases of multiple wrong (as very unlikely) position or date, a suspension of the account could be acceptable (and even desirable).