Most of the field of macroecology (the study of the global distribution patterns of species) relies on analyses at coarse scales, such as 1-degree or 20 × 20 km grid cells. There are hundreds if not thousands of papers on macroecology using these sorts of data, including for plants. Accuracy and precision are important for many purposes, and I strive for both in my observations, but in many of the less well-known parts of the world, we risk losing very valuable data if we impose overly-stringent criteria.
I agree. Old specimens with just a month and year, for example, are still very useful, especially if they have accurate locations. I believe GBIF accepts anything with a month and year. Obviously, as with location data, any new users adding data should strive to make it as accurate and precise as possible.
Anyhow for what it’s worth, there’s nearly zero chance the inat admins will decide now to do this, since they specifically reversed it before. So you all are safe with your huge circle research grade obs :(
What was proposed was not sending it to GBIF, under which scenario it would be essentially invisible to much of the research community who access data through that site. I might be wrong but I think GBIF is still the route through which most researchers analysing large datasets obtain iNaturalist data.
I’d prefer to see customising the maps made easier, and also ideas as to how we nudge new users to map sightings as accurately as possible - perhaps messages that could appear if the accuracy circle is very large, or if people plot a location by simply searching for a county, national park or other large area…
The best observations I’ve seen regarding photo quality and rarity of Flora are from observation more than 100metres. The rare or uncommonly observed organisms are found in rugged remote places where someone is unwilling to use their phone for GPS precision. The best photos are taken from cameras that usually don’t have GPS
If iNat uses restrictions for inaccuracy then more than 20% of unique flora species in my state (such as iNat 1st and only) will never be able to be researched by other websites, it’ll become pointless for hikers or skilled photographers with proper cameras to use iNat. Not many people are willing to exhaust batteries in phones for the sake of an accuracy of 100metres. Many remote National Parks stretch on for 50+ kilometres, if a flora species is found 100metres away then it’ll also likely be seen 2kilometres away. I am a plant guy, I see no loss of quality for inaccuracy of <100metres unless a rapid change of geological feature such as a steep gully, wetland, creeks etc
I think a researcher needs to adapt in their research techniques to accept changes of geological features, whereas layering plant observations on a reasonably flat map of altitude varying of 200metres may have tolerance of 2km, the edge of a river tolerance of 50metres, along a coast a tolerance of 100metres etc. This will ensure the correct Ecological vegetation Class.
Prohibiting Research Grade for inaccuracies such as this may consequently lead to poaching of rare organisms such as threatened orchids. It’ll make documenting organisms a chore, it may reduce quality of observations by the means of poor photographs + less photographs + less identifiable feature photographed leading to inaccurate ID’s.
We also must understand that the only flora species that will reach Research Grade on iNat are usually the common species, or the distant silhouette of a rare tree. People are usually more willing to identify an observation is when they can correct someone’s mistakes, or they can’t be re-corrected themselves. Observations of Flora that show identifiable features rarely make it to research grade anyway. – But this is irreverent to the topic…
I suggest the use of filters should be the preferred method.
Interesting. I don’t seem to have any problem documenting rare species in remote areas within 5-10 meters accuracy. I use a DSLR camera without GPS, with time synched to a separate GPS unit, and later geotag the images from the GPS data. A little more work back in the office, granted. I guess it’s all about what is most important to the individual observer.
According to Google, the average human walking speed is recorded at 84metres per minute. My photographs only display minutes not seconds. Terrain does vary so it’s likely that a majority of photos will be over 100 metres uncertainty if using that method.
that’s a dramatic stretch, sorry. i can’t see it happening.
Me neither. We do it all the time for work. If people think it’s a huge burden to get a GPS reading, we just need to make sure people are able to use their existing equipment properly. It’s super fast and easy.
huh? why?
i’ve gotten good at making observations while walking, but still, i don’t know anyone who goes the average human walking speed while doing iNaturalist, probably it’s rare for anyone to move even half that while poking around making observations.
Like, i get it, some of you don’t care that much about spatial accuracy and we can’t make you care. And I have some coarser mapped observations too, but i am aware that the data is less valuable for a lot of conservation and ecology uses. Sometimes that’s all you can get, but unless it’s a large tree or some other very large and obvious plant, no ecologist will ever relocate something with a >100 meter accuracy, or at least, it would take tons of time. I add 50+ observations on a given hike without slowing down much and my accuracy is usually 5-10m, that’s really easy to do. But if you don’t care, you don’t care, and there’s no real way to make people care. It’s not mandatory.
For what it’s worth, i think there’s a near zero chance the admins (re)institute a policy removing RG from coarsely mapped observations, so you all can stop worrying about it.
Depends on a variety of factors, you can’t assume everyone objective in the wilderness environment has a sole objective to document the occurrence of species. Runners, Hikers, Photographers, Mountain Bike riders, Ecologists, Hunters, Tourists are just a few more types of people whom also use iNaturalist, many of which stumble across fascinating organisms. The GPS method mentioned suggests any person traveling at a distance of 100m per minute should not have their observations make it to research grade due to uncertainty of coordinates, this is also assuming their GPS tracker is synced to the microsecond with a fast refreshment rate of a second or less. Why would I photograph a plant species 1000 times within such a short distance? Instead photograph the species only once then move on is the more likely approach for myself, if the same species becomes repetitive then I will increase my journey speed as I have already observed that species in that area. If I see an unfamiliar daisy it’ll sometimes take less than 5seconds to pull my camera from my side camera bag & photograph it, then move on.
Research Grade doesn’t make an observation to be identifiable with 100% certainty, I’m sure more than 25% of Research Grade observations in plants I have seen May or May Not be identified correctly, but due to lack of decent evidence I give benefit of the doubt, therefore it remains at research grade. So using Flora maps as a layer will be misleading regardless of your approach. In my opinion it’s best to create an Ecological Vegetation Class using species that have been personally confirmed or by a known work colleague or botanist as to be the basis of ecological projects. If you want to increase certainty, then use filters to include observations from sources that are more likely to show identifiable plants.
It’s not that I don’t care about accuracy, it’s just as I see very little value of layering maps that have such minor inaccuracies, I see it pointless, irrelevant and unnecessary. Whereas I prefer to only include Observations that have identifiable features associated with the observation such as fruit or buds due to poorly speciated plants in my state. It seems that people care more about a misidentified tree that’s 50metres off location then the actual ID of the species, which doesn’t make sense to me.
There has also been orchids stolen from various nature reserves on many occasions such as ‘Langwarrin Flora & Fauna Reserve’ where people are concerned about Plant mapping applications that make it quick to steal flora from reserves.
Most of my Plant species remain at ‘Need ID’, there are not many plant ID’ers in Victoria. I think about 90-95% of my rare/uncommon flora species will never make it to research grade anyway, but still there is always hope.
If you are using a GPS tracker, it shows the GPS location on the device. Take a photo of the bug, then take a photo of the GPS tracker reading. When uploading photos drop them all into the uploader in the order they are on camera, and for each GPS tracker photo add the coords to the photo(s) before it. Then delete the GPS tracker photos before committing the batch…
One problem I have with a criteria like this is that it leads to am almost requirement to do multiple or duplicate entries. For example,one of the parks I visit most often is about 400 meters long. When I visit, I typically put a circle encompassing the park.
Then for example if I see 6 Black-capped Chickadees in the park, I will record one record with that circle, and note in the observation fields that 6 individuals were observed.
If I needed to record (or even try and photograph) all 6 separate individuals as separate records, because otherwise they will be downgraded to casual, I’m not sure I’m going to do that.
Yesterday, I was taking photos along a semi-wooded trail near a university district in Seoul. My DSLR doesn’t have a built-in GPS unit so I was taking photos of the camera screen with my cellphone to get coordinates and combining the photos together when uploading observations to iNat. Most things showed up in the area I expected them to be placed but the photo for one wasp was (a) plotted more than 1000m away from my other observations and (b) given an accuracy range of 2000m.
I was able to drag the accuracy circle back to the same general area as my other observations but without those other nearby observations it would have been a lot trickier to place the wasp in an accurate location and I would have used a larger/more imprecise accuracy range were that the case.
It’s may also be worth pointing out that Google Maps in South Korea does not offer the same level of precision that it does in some other countries – I was blown away at how detailed maps in the USA are when zoomed in compared to what I often see here. However, that’s an issue between Google and the South Korean government rather than anything iNat has control over.
Dang, I removed the accuracy circle from a lot of my recent observations because I was able to pinpoint their precise locations on the satellite map. Guess I’ll be going back over them and adding accuracy data tonight
I don’t see iNat supporting different location accuracy thresholds for different taxa, and it’s unlikely we’ll support a location accuracy threshold for Research Grade status in general, but I can certainly bring it up with the team.
And one should only vote yes for the “Location is accurate” part of the Data Quality Assessment if you think the observer put the wrong location for the observation (eg a wild porpoise observation in the middle of the Atacama). It is not related to positional accuracy.
Not to go too off-topic here, but in my experience this happens when you open your phone’s camera app and immediately snap a photo. The phone doesn’t have time to connect with satellites and other signals to get a precise location, so I usually try to wait 5 seconds or so before snapping a photo, especially if I’m in an area without cell coverage. But if you’re using a DSLR I suggest geotagging your photos on a computer if possible.
That’s the method I most commonly use of photographing the phone’s GPS coordinates which requires taking the phone out of my rucksack (between 20 seconds to 1 minute time consumed), another method mentioned is synchronise your camera time with your phone, record the entire trip so with a constantly running GPS tracker. An example would be a subject photographed at 2:38pm you can go through the GPS tracker log to see where you were at 2:38pm and sync the observations to that of the photo. This will mean that an average inaccuracy of 84 metres because we are not synchronising to seconds, people synchronise to the minute therefore the average walking distance of 5km/h. If you move faster than accuracy decreases, so if you move faster than 6km/h downhill then your inaccuracy will breach the threshold of 100 metres and unable to reach research grade (proposed). If you use the method you suggested and take 300 photos costing 20 seconds per photograph of the phone that accumulates to 1 hour 40 minutes, in an 8 hour sunlight Winter day than that a massive loss of time people will arrive at camp in the dark.
If the cost to you is in time, then make half the number of observations?
Alternatively, take 2 photos… one with your “good” camera and one with your phone. Upload all, merge each pair such that the phone photo provides the location data, and then afterward delete that photo
Or… work with the 200m to 500m accuracy range… vague data is better than no data!
If speed through the environment is your primary objective, then a camera with GPS would surely help?
Do you know why this is? It seemed like a good solution to the problem, rather than allowing super coarse plant observations to be RG because it is valuable for certain wide ranging animals? This is what always bugged me about this, it seemed a good middle ground and was rejected without an explanation. Is it too hard to code?
i see that 100m is maybe too tight a cutoff but to some extent, if you aren’t willing to do something well, ALL of the hour (instead of hour and 40 minutes) is wasted time! You could also just use the app, some people don’t like it but it works great.
Funny that, I have the same opinion for people whom don’t add supporting photos to plants such as fruit, form, leaves etc… What the point. So much lost data, so many misidentified species with May or May not be which are the most common to make it to research grade. I only wish I had means to filter the observation that don’t contain additional images of plant features that way I have certainty that the plant has been correctly identified. Especially where I’m from where the type of hairs may define what species it is, or the operculum, the juvenile leaves. Etc.
I think we all have different expectations of quality, whereas my primary concern is of correct ID, yours may be the accuracy of location.
When I document observation, I hope to train AI such as ‘Seek’, future software or Human by allowing it to acquire knowledge of these identifiable features (although it would be nice if I could tell the software that the photo is of fruit) but I see the value of perfect accuracy as low due to the regular occurrence of those species in the area, whereas you see it as high. So I think what each person is doing is contributing to what they think is valuable, different perspectives from different environments from different people with different objectives.
Perhaps high human density cityscape vs remote wilderness?
