Change Mapping of Obscured Observations to Improve Data Visibility and Reduce Confusion

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

URLs (aka web addresses) of any pages, if relevant:
www.inaturalist.org

Description of need:
Obscured observations are an important part of iNaturalist, both to protect sensitive species and to protect individual privacy. However, they can make the maps very difficult to read, hiding more fine-scale unobscured mapping.

Zooming in quite a lot one can see the obscured observations as circles instead of pins but it is still messy and on most maps they can not be filtered out

In addition the ‘wrong’ location of the circle in the obscured location leads to confusion and sometimes silliness as with cacti in the ocean, etc.

Feature request details:
I propose three ideas, in order of my preference:

-display all obscured observations as a grid, in the size of the obscured area. See below, but with just the rectangle not the circle. More observations would mean darker rectangle but always behind the unobscured ones.

image

-make it possible to turn off obscured observations in more maps.(no screen shot, the popup disappears when i try to clip it)

  • Verifiable Observations

*Obscured Observations

  • Observations w/o Media

  • Captive / Cultivated

  • GBIF Occurrences

This is harder because obscured observations may or may not have media and be cultivated.

-change the maps so the pins and circles display further out. This would not solve all teh problems but at least make it possible to see which observations are obscured without zooming way in. This may present a caching issue though, i am not sure.

I also like your first proposal the best. I think that it still communicates the key relevant information in a simpler, easier to read fashion: approximate number of observations and approximate location.

In terms of location, this proposal represents the true precision of the measurements (in the rectangle) better than the current dots (which can give a false impression of greater precision). A lot of beginners are confused by the dots (including me when I first started using iNat).

If the darkness of the rectangle communicates amount of observations, this may actually be a better visual indicator of observation density in some cases. When there are lots of obscured observations of species, it’s really impossible to assess/compare density visually - it’s just obvious that there’s a LOT in the area.

This proposed method would also have the advantage of making the underlying features of the map more visible. When there are lots of obscured observations, they tend to cover up other features of the map (such as coastlines) and make reading the map just generally more difficult.

Lastly, this might actually make maps load faster? Not exactly sure but it seems likee just loading the rectangle would be easier/quicker than all those individual dots.

I think the second option (unchecking obscured observations) would also help a lot, but I’d really like to see the first proposal in action!

5 Likes

I like idea 2: toggleable display, with the default being off for obscureds.

5 Likes

The only one of these which I think presents a useful improvement would be 2: the ability to toggle displaying obscured observations on the map. However the default I think should be that they do display as circles.

For the first one would we would be unable to open obscured observations from any maps. This is something I do very frequently for looking at photos of nearby observations and for adding disagreeing ID’s to out-of-range observations.

For the second, I don’t see why you would want to mix in cultivated and casual observations. If somebody wants to map those they can easily do that from the explore page. Most people do not want to see them and it would grossly distort range maps.

For the third. I don’t see how how it is an improvement either, if anything that exaggerates the problems you were mentioning at the beginning of the post.

2 Likes

I’d like to either see rectangle only or not see obscured observations on the map at all, it’s impossible to find out which spot has obs (unobscured) and which doesn’t when it’s a huge rectangle filled with circles, there’re some users nearby who hide all of their thousands of observations (done at public parks), I pray to not see those circles again.

5 Likes

The box would be clickable just like the circles.

I wasn’t proposing changing how filtering of captive and no media observations work. Just noting that there’s overlap between those and whether something is obscured.

1 Like

this all seems unnecessary to me. you’re never going to be able to have a perfect visualization that will handle every little quirk in the data. from my perspective, there are more cons if you go with any of your proposed methods vs just leaving things as they are.

if you really want to change something, i would propose that changing the density opacity scaling in the grid style tiles to go from, say, 50%-100% opacity to, say, 25%-100% opacity. that will help to deemphasize all the random one-off points in general, though that means making one-off points harder to locate on the map (which i think is a fair trade, since i’m usually more interested in deemphasizing one-offs than in locating them).

You feel that randomly placed points that appear to be more precise than they are is preferable to having a grid system?

8 Likes

We have a lot of plants confidently living in the South Atlantic Ocean around Cape Town. And I don’t mean seaweed / algae. I’d rather see them ‘somewhere in this rectangle’ on land.

PS this chafer is in my garden. iNat has pinned it firmly in the grounds of our local hospital. Not! Actually https://www.inaturalist.org/observations/102235187

4 Likes

I do think the display of obscured observations goes beyond a “quirk” in the data. There’s a fair amount of obscured observations on iNat, and looking at pretty much any map shows some obscured observations, so it is an experience that all users have.

I’d also be interested to hear what you think the cons of this change would be - some people might be able to suggest other improvements or alterations to the proposal that would satisfy all users.

For instance, @kevinfaccenda 's point that change 1 would make it more difficult to locate specific obscured observations is a point I hadn’t thought of.

I think that making option 1 from the proposal above the default while also having the availability of a check box as in option 2 that “turns on” the dots that we currently see could alleviate this issue. If that check box is sticky across sessions, it could lead to very little hassle for users who like having the current setup with dots.

4 Likes

yes. in your first 2 proposals, you’re asking for some extra layer(s) of complexity whose benefits i just can’t see outweighing the costs. just educating folks that, at some level, obscured observations exist and that they may show up in nonintuitive ways on the map is a much simpler and effective solution, though no solution is going to be perfect.

of your proposals, i think the least objectionable is the last one:

… but i think a better version of this request would be to allow the user to select which map style – grid, pins, heatmap – they prefer to see, regardless of zoom level.

the cons vary depending on the exactly how you implement any particular change. i could write many, many paragraphs to explain exactly what all the problems are for each approach, but there are too many potential implementations, and for simplicity, i’m just going to say that it mostly boils down to unnecessary complexity. you’re effectively asking for the grid version of the map to gain extra options that other versions of the map – like the pins – don’t. you’re asking potentially for a new type of map visualization to be created, too, and to be mixed together with the grid version of the map. so that’s extra processing power and maintenance, too.

someone asked for something similar to charlie’s first proposal a while back: https://forum.inaturalist.org/t/option-to-display-obscuration-rectangles-grid-on-explore-map/7646. the reasoning for the change in that thread was a little different than the reasoning in this thread, and in that thread, i describe mostly why it’s a bad approach from the perspective of privacy, since that was the main con in in the context of that thread, while unnecessary complexity is the main con in the context of this thread (and in general).

Well dont vote for it then, I guess. I think it increases complexity to have those intentionally wrong circular markers all over the place, myself. :man_shrugging: As for the relative processing power and caching and such I think only the devs could tell us if it matters and which is better.

My main issue is at some zoom levels the maps on observations are made nearly worthless due to not being able to distinguish the obscured observations or see what the actual range is using un obscured data

2 Likes

We’re aware of that, and they’re not helping exploring the map, but instead playing the opposite role of disturbance, knowing what they are can’t help with that problem.

2 Likes

Edit the location

1 Like

Yeah that observation isn’t obscured

1 Like

I totally agree that the circles for the obscured locations are confusing. They are often difficult to differentiate from those that are accurate, difficult to see on the map, and the large numbers of them in some areas makes clicking on them individually annoying/impossible, especially when they are difficult to see. Here’s an example of where there are often difficult to see:
image

I really like the rectangle area idea as that gets rid of the ambiguous nature of all the random points and makes it clear that there are points there when they are difficult to see. I think if you click on that rectangle, it should just give you a query showing all of the obscured observations it represents. That’s far easier than trying to find them all and click on them individually. I’m a little concerned about how dark/opaque the rectangles would be. They should not be so dark as to make it difficult to see the map underneath. There’s already a huge issue with not being able to see the underlying map if there are a lot of observations shown.

I like the option of being able to turn off obscured observations, but still would much prefer the rectangles over that if it were one or the other. A toggle between rectangle and circles could also be a cool option. I really see no use in the circles beyond visualizing how many observations are in an area, but that is legitimately useful.

As far as the zoom-level display goes, currently, if you zoom out, everything turns into an orange square whether it is an animal, plant, or whatever. Like this:
image
Changing the color of squares that only contain obscured observations could be useful. It’s already confusing that the colors change with the zoom level though.

Zoomed in a little more is this:
image
At this zoom, you can clearly see where rectangles for the obscured observations would really help. It also raises the question of how you would color it if you did a query like the above where multiple kingdoms are shown. Seems like there would need to be a color for multiple kingdoms as the differently colored rectangles would cover each other.

Seems like it would also be nice, if you zoom in beyond the level of the rectangle, to make it disappear as it would at least partially obscure the background imagery. This could also just be done with a toggle. Note here how it is difficult to see the background where covered by the rectangle:
image
It would be really annoying if you couldn’t turn off the rectangles if you wanted to see the detail of the background underneath.

6 Likes

if you’re going to potentially display obscured observations with non-obscured observations, you’re always going to have some sort of noise in the visualization to deal with, unless you display a density grid with a 0.2 deg long x 0.2 deg lat resolution that incorporates both obscured and non-obscured observations. the problem with that kind of approach is that the resolution of such a grid is so low that it’ll be much less useful for displaying non-obscured observations.

well then, some folks might say… why not just show obscured and non-obscured observations as separate layers? it could be a 0.2 x 0.2 deg grid for obscured observations, plus the existing grid or pins for non-obscured observations, right? as much as you think the obscured observations clutter up the map view now, they’re going to cause that much more clutter/noise if you stack 2 layers on top of each other. any time you have an obscured observation, you’re going to have a giant 0.2 x 0.2 deg box in the background that represents an obscured observation. how will the opacity scaling work for the 0.2 x 0.2 deg boxes? will the scaling for the obscured observations be independent of the scaling of the non-obscured observations? will the opacity scaling for the giant box go from 50%-100% or some other range? will the color of the giant box be different form the non-obscured observation markers? you could easily end up in a situation where you have giant boxes between the basemap and non-obscured observation markers that totally block out the basemap and/or blend in with the non-obscured observation markers. (in other words, this kind of implementation could potentially be way noisier than the existing existing maps.)

or suppose you’re looking at a specific taxon, and those observations are all obscured. now imagine such a map with 0.2 x 0.2 deg boxes. does that really improve upon the current situation? folks still want to be able to click on the giant box to bring up a particular observation. but what if that box is composed of many obscured observations? right now, you could display many clickable obscured points within a 0.2 x 0.2 deg box, even if they are not displayed at their true locations. but if you have just a single 0.2 x 0.2 deg box, when you click on that box, what do you expect to happen? do you just throw up the single most recent obscured observation in that 0.2 x 0.2 deg box? do you only allow the box to be clicked when there actually is an obscured observation with obscured coordinates at that particular spot (in which it could be confusing why the rest of the box can’t be clicked)? do you display a list of observations (in which case that’s going to involve a significant change to how things are displayed, akin to the discussion here, which is on the backburner: https://forum.inaturalist.org/t/indicate-overlapping-observations-for-map-pins/18406).

yeah, obviously. but i’m trying to help you all understand why this probably isn’t a viable request – at least not in this current incarnation. i don’t disagree with general ideas like “improved data visibility”, etc., just like i don’t disagree with the idea of world peace, ending world hunger, etc. but the devil is in the details, and you haven’t provided any details here that seem to lead to an improved situation across the board. as i said, your proposals may come with some benefits, but they also come with many cons, which as far as i can tell still end up net negative.

just as a reminder, i did indicate that – besides education – there are a couple of things that i could support:

you can see an example of 0%-100% opacity scaling here: https://jumear.github.io/stirfry/iNat_UTFgrid_based_density_map_for_Leaflet.html. 0% is probably too low for the minimum opacity, but it’ll give you an idea of how a different range will help to deemphasize extraneous stuff.

you can see an example of this here: https://jumear.github.io/stirfry/iNat_map.html?view=elevation&taxon_id=42414&place_id=1. by default, this page gives you a grid style map up to zoom level 10 and then pins at more granular levels, similar to what iNat’s maps do, but you can also use the layer selector to choose grid, pins, or heatmap, regardless of zoom level.

… and with that, i think that’s all i have to say here. either you’ll see what i’m talking about at this point, or you won’t, and it’s unlikely anything else i say will change that.

I’m okay with displaying them on separate layer that could be turned off. I really don’t care much how it’s going to be implemented, I just want to not see them as much as they’re visible over other observations now.
Seeing random circles when all taxon is obscured isn’t any more useful that would be rectangles, I checked it a lot with Oystercatcher where many observations done close to each other on map looked like chaotic mess, how it helped studying? Seeing one shape of all rectangles summarized would at least work as mapping/atlas feature. Also it can be just done differently on explore pages and taxon pages if seeing circles is needed for some reason.

2 Likes

i realize this, but it isn’t convincing me at all, and it’s coming off as condescending and overly aggressive (i am saying this as a forum user, not a moderator, you aren’t breaking any rules here to be clear). I am hearing that you prefer the system as it is, I am saying I find it very hard to use as more obscured observations appear, especially when zoomed out to a mid-level. Given it has 16 votes others seem to agree with me. (of course, there are no negative votes tracked so it could be more people disagree than agree, i have no idea). As for whether it is viable, i think speculation here isn’t too helpful other than if it is from the devs. I’ve seen this sort of thing on lots of other websites, so hardly think it makes sense to say it isn’t possible or reasonable to do.

2 Likes

as a person who (in part) makes maps for a living i find the circles to be really frustrating because they are intentionally wrong spatial data. It even gets exported to GBIF that way. Obscuring location is absolutely important and valid, and the ‘fake location’ system made sense when iNat was small and there weren’t that many of them. But right now it’s untenable, confusing, and maybe most importantly, it’s creating wrong data that isn’t clearly wrong to people who aren’t iNat experts. We had a similar issue where I work when things like this accidentally lead to actual applied conservation and development policy being made based on a misinterpretation of map symbology. It does matter, and it does make sense to change it.

9 Likes