Improving Location Accuracy on observations

When you’re taking photos in the app, you take the photo and then the GPS coordinates are grabbed, at least on Android. After you take the photo, you can see the meters going down in number (20m…15m…12m…etc), so it’s not too late to improve precision after taking the photo. I pause a few moments for the number to go down–and tell others I’m teaching about the app to do the same.

I personally use a separate app when I’m iNatting so that I don’t have to wait so long for the GPS precision to refine. I could see adding a button to go into “iNatting mode” and having location services on temporarily, so you don’t have to use a separate GPS app. Having location always on would probably drain the battery pretty bad though right?

3 Likes

The current generation of phones seems to manage location services energy usage fairly effectively. I have location services enabled at all times and I am aware that my phone is periodically updating its location. I am also a runner and while running the GPS is continuously tracking my location. I am not sure I would characterize the drain as “pretty bad.” As noted elsewhere in the forum, I find that I can improve location accuracy on observations on my Motorola G7 Android phone by opening Google maps and pressing on the “my location” homing button prior to taking images of the plants around me with the iNaturalist app. Telling Google Maps to locate “me” seems to improve location accuracy.

2 Likes

huh? as far as i know, Google stores its data in EPSG 4326 (WGS 84). Google Earth displays data in EPSG 4326. Google Maps displays stuff in EPSG 3857 (Web Mercator), but it’s just a different projection. iNat probably does the same thing, storing data in EPSG 4326 and projecting it to EPSG 3857 when displayed online. if everything on your map is projected the same, i don’t know of any reason this would cause the Google map to “be a little different”.

1 Like

probably not what they were referring to but when it switches into oblique view when you zoom into some areas, that does shift the points around a bit

1 Like

I think a lot depends on where you are. When I’m in remote, hilly, forested locations it often takes a long time to get a GPS fix at all (tens of minutes or even hours in some cases). Once you have a fix, it will maintain it as long as an app using location services is open (e.g. Google Maps). But if only the iNaturalist app is open, it’s common to lose the fix and it then takes a long time to get it back.

This is also how it works in iOS, but the updating seems to only run for a few seconds after the photo is taken, I think? So if the precision value only gets down to 253 metres within those few seconds, it stays at 253 metres. What I’d like is either a way to keep my device updating the GPS fix all the time, or to be able to extend the time that it has to refine a new fix.

I sometimes use a separate app too. If I have an eBird list running, with tracking on, at the same time, I get very good locations without having to wait or check the accuracy each time. Otherwise it requires a lot of attention to location quality which I’d rather was directed towards local biodiversity! It does drain the battery a bit faster, but there are situations where it’s worth it. Or I open a mapping app so I know when the GPS location has been updated - again, this adds an awkward extra step. I don’t think we should need to use a second app to make the most of iNaturalist.

2 Likes

yeah, the android app has a ‘find current location’ button that re-zones in to where you are pretty well, having that might help. A roaming mode, would be nice too, maybe one that also shows/saves the air photo and shows where you have observed things that haven’t been uploaded yet… which is another thing on my wish list I think I have requested before…

3 Likes

EPSG is a system of codes for different Coordinate systems,Datums etc.
Do not confuse Datum with a projected Coordinate system(Mercator etc.).
A Datum defines the shape of the earth (3d). A projected coordinate system(projection) is a 2d reference system based on a starting point. If a different Datum is used, then the starting point for the projection will be different. This is referred to as Datum Shift.

ok, fair enough. i still don’t see anything that indicates that Google is using a modified anything. here (https://developers.google.com/maps/documentation/javascript/coordinates), they say “Google uses the World Geodetic System WGS84 standard”.

Maybe this was mentioned somewhere up-thread, but there seems to be two issues being addressed here: accuracy versus precision. This applies mainly to cases where you might not be GPSing as you are obtaining your photos (e.g., my camera lacks GPS capability) and thus might have to fix your location after the fact.

If I plot a location with a 200 m radius circle around it, and I know for certain I was within that area, then my record has high accuracy but somewhat low precision.

If I plot a location as a point, but I’m a couple of kilometers off from where I actually was, my location has high precision but low accuracy.

To me, and for iNat purposes, it’s more important to be more accurate than precise. I’ve had to plot some records where I wasn’t sure exactly where I took the photo but I know I was within a particular area that I can view on a map, so in some cases the circle around my record might have a radius of a few hundred meters. For most iNat purposes, and for many mobile species, that location data should be adequate.

7 Likes

Quite so, and no, I don’t think it has been raised upthread. I think some confusion comes from iNat calling the error radius value “accuracy” when it would more accurately be called “precision” or just “error.”

Either way, though, the lower the precision (larger error radius), the more I am going to question the accuracy of those center coordinates.

6 Likes

This is where a description of the location is a good thing. Many, maybe most, records lack a description of the location except the one generated by iNat based on the coordinates. In every other databasing system I’ve used, coordinates alone are not sufficient, there has to be a location description also which provides a check on possible error in the coordinates.

2 Likes

Well, when I started posting to iNat it was hard to change the radius to the smallest numbers for every observation, at least that’s what I thought, for long time I just had saved places with large 2-3 km diameter, and with that low precision it stil wass accurate and it’s sstill better for old observations, after 3 or more years you don’t remember where you took the photo, well, for some closeups I can’t say it after a week! And with forest it’s really hard to understand where you were, it all looks like trees from above.

5 Likes

https://support.esri.com/en/technical-article/000009982
“Google Maps and Microsoft Virtual Earth use a Mercator projection based on the World Geodetic System (WGS) 1984 geographic coordinate system (datum). This Mercator projection supports spheres only,”

The key word here is “based on”
google maps assumes the earth is a sphere, not an ellipsoid. A GPS reading is based on an ellipsoid. WGS84.

1 Like

I agree, although in the case of GPS fixes they tend to go together. The suggested changes would improve both accuracy and precision.

I think this would be useful, but it would have to be optional as I suspect few people would have the patience to type something in.

1 Like

right. that pseudo-Mercator projection (WGS 84 Aux Sphere) is EPSG 3857. but if everything shown on an iNat map – basemaps, observation markers, etc. – uses a EPSG 3857 projection, then i just don’t see where you run into a situation where “the Google map will be a little different”. (it’s only if you try to mix projections where you might run into problems, as far as i am aware.) anyway, i’m not going to argue this further, since it’s probably getting off topic at this point.

Sorry I didn’t mean to say that iNat was using a different projection, only that if you take a GPS reading at the base of a light pole that is visible on Google maps aerial view, and , when you enter those coordinates into Google Map aerial view then it will most likely be a little off, the amount will vary depending on where on earth you are.

1 Like

While this thread goes on and on about how to make records with automatically generated locations both more precise and more accurate (good goals), I keep thinking that they way I upload photos leads to accurate but not precise locations and I don’t (usually) plan to change that.

I go somewhere, take lots of photos with a camera that doesn’t automatically generate latilongs, and add locations when I get home. I do that by picking a central spot in the park / roadside / neighborhood where I was and giving it a large enough accuracy (what we have called “fuzz factor”) to include all the photos. I’ll use multiple latilongs to deal with changes in habitat or other natural breaks in the location, but it’s not at all unusual to have an accuracy of +/- 50 meters, 200 meters, sometimes even 500 meters or more. In a global perspective, that’s good enough. (The good news: Accuracy is often small because I’m both a decrepit old woman and a botanist, so I rarely get far unless I revisit a site multiple times.)

Bottom line: No matter what you do to improve accuracy of locations, people using iNaturalist data are going to have to deal with large accuracies in many observations.

7 Likes

Hi Barbara,

That’s a perfectly appropriate way to make observations, and thank you for doing so. The data will be very useful for many purposes. If you want to improve precision (for example, for a particularly interesting or unusual observation), one of the less painful ways to do so is to make an observation at the same time with the app, and then add your photos from your main camera to that observation later by dragging and dropping on the computer.

I agree that we’ll likely never get to a point where all observations have <10 m precision, but that’s fine. We can still try to find ways to make it easier for more people to make more accurate and precise observations.

6 Likes

:-) Thanks. However, you assume I carry something that supports apps more effectively than does the flip phone I’ve carried for years because it fits so well in my pocket. For me, the best way to get a good location for a photo is to pull out my little GPS unit and photo what it says. I’ve done that when I want to be very accurate (e.g. for rare plants), or when there aren’t good landmarks (e.g. for my apple survey earlier this fall), but most of the time, precision just doesn’t matter that much. The great majority of things I photo are common plants.

6 Likes

it all depends on what exactly you are using the data. As a spatial ecologist I can tell you that you can do a lot of things with the precise (20+ meter accuracy) observations that you just can’t with less precise ones, in terms of mapping, microclimate data, applied conservation, etc. If the choice is between an observation with 200m accuracy and no observation - by all means the 200 m accuracy observation is welcome and I’ve posted more than a few of those too. And if you just aren’t interested in the app or geotagging, etc, that’s fine! But for what it’s worth, in applied ecology and land management, the precise data is really valuable, really it’s a game-changer.

4 Likes