|
The technical forum of the NBN Trust for use by the Network. It includes discussions and announcements on the NBN Atlas, Recorder 6 and Indicia |
You are not logged in. Please login or register.
Where can I find the "public" resolution of a dataset? I've looked at lots of dataset metadata under https://registry.nbnatlas.org/datasetsand have been unable to find any information about whether the Atlas data are at "capture" resolution or have been blurred for public display.
Keith
Hi Keith
Thanks for your message. All species data on the NBN Atlas are visible at a 'public' resolution. For sensitive species that means all records submitted have been screened and blurred to the criteria on the sensitive species list that runs in the database and these records are visible at that resolution. All other (ie non-sensitive) records are visible at the submitted resolution which is chosen by the data partner, so other than for sensitive species there are no blurred records or multiple resolutions of the same datasets.
If you need to know if the records in a particular dataset are provided at capture resolution, please request this information directly from the data partner, whose contact details are available on their data partner pages.
There is more information about the handling of sensitive species in the Help section: https://nbnatlas.org/help/sensitive-species/
Christine
Christine Johnston
NBN Scottish Liaison Officer
I know that in the absence of proper access controls on the Atlas it is only capable of displaying one resolution, the one chosen by the data supplier. What I'd like to know easily, without having to write to the data provider, is what that is. It was on the metadata forms when the data were supplied so that information must be easily available.
I note that in all the metadata for a dataset (e.g. Higher-level groups; Watsonian Vice-counties; Data quality assertions; By-monthly phenology, Useage Stats, etc.) there doesn't appear to be a breakdown of the grid ref resolutions. e.g. how many records at each of 10km, 2km, 1km, 100m, 10m & 1m there are within the dataset? If this was available it would give a very good clue about the public viewing level chosen by the supplier and whether it would be worth contacting them to ask about better resolution access.
Keith
Hi Keith
The species datasets in the NBN Atlas do not currently carry all the metadata that is supplied via the NBN Exchange format metadata forms but we are working on the format at the moment.
However to answer your second paragraph it is possible to filter a dataset by coordinate uncertainity and that will give you a breakdown of the record resolutions within that dataset. To do this, search for the dataset and go to view its records (link is found near the bottom of the page). Go to the 'Customise filters' button (top left), clear the filters and choose 'Coordinate uncertainty' under the Location heading. You will then see a breakdown of the resolutions (in metres) of records in that dataset.
I hope this is helpful.
Christine
NBN Scottish Liaison Officer
Hi Christine, thanks for your reply.
I tried your recipe for viewing the grid ref resolutions on this dataset https://records.nbnatlas.org/occurrence … chartsView.
I got charts for "By Species", "By higher-level groups", "Watsonian Vice-counties", "Data quality assertions", "By Month" and "By Year" but none for grid reference resolution. I was expecting a chart as this seems to be ideal chart material.
I then re-read your recipe more carefully and noted that I had to click on the "Narrow Your Results" Location heading to make the info appear, and bingo, there it was - all 123,599 records are 10km resolution only and have therefore been blurred before presenting to the public, and no means now to be granted enhanced access via the Atlas, which was only a few clicks away on the Gateway.
A bit painful to find but the info is indeed there if you know the recipe to get at it. Thank you.
Regards, Keith
Ladybirds? All of them? Only at 10km resolution? I shall email Helen and see if that was the intentional. It seems very surprising to me.
Jim Bacon.
Hi Keith. 10km was not the intention. Helen will follow it up and get it fixed though I don't know how long that will take. It is good that the matter was aired on the forum rather than by email otherwise the connections might not have been made.
Jim.
Browsing this I found that one record that we supplied to the NBN gateway had public visibility set at 10km square resolution (NY36) - during migration this now appears to (incorrectly) have a co-ordinate uncertainty of 1000 metres - see here. Rest of the records in that dataset show co-ordinate uncertainty at 10000m correctly. Very odd!
MARK
I would hazard a guess that this is because it is the one record of a sensitive species in the data resource so the coordinate resolution has been tinkered with during data ingestion.
Jim Bacon
Interesting (and a bit worrying). So would that mean that records might have been "blurred" to a more precise level than was intended? That could have quite significant implications for many data users so this should be clarified quickly.
I don't think you need to be concerned. The record was supplied as a 10km square so no more accurate location can be obtained even if it appears so.
Jim
I don't think you need to be concerned. The record was supplied as a 10km square so no more accurate location can be obtained even if it appears so.
Jim
Oh yes I see, maybe not so immediately critical then. But still, useful to know how the issue arose.
Cheers
I am, of course, just making deductions based upon the data we can see and what I might do as a developer. The last word can only come from the NBN. My guess would be that the Atlas was designed in the expectation of being supplied with accurate location data and ensures that sensitive records are blurred. The process may not take in to account that it has been supplied with data that is already blurred. It would then set a suitably blurred accuracy (e.g. 1km) on a record that was actually supplied at a still coarser resolution (e.g. 10km)
You can find 3.4 million sensitive records of which 836 thousand have 1km uncertainty, 497 thousand of which are from BTO but a quick scan of the records shows grid references of 1km, 2km and 10km accuracy. See https://records.nbnatlas.org/occurrences/search?q=*:*&fq=sensitive:"generalised"&fq=coordinate_uncertainty:"1000.0"&fq=data_resource_uid:"dr528"
Jim Bacon.
Interesting. If you more closely you see these BTO records are of just 12 species. Without spending ages on this, there are perhaps a number of underlying reasons for the inconsistencies although I might have a more careful dig around when I'm back in the office.
I provided the BTO data pre-blurred to appropriate levels, depending on species, season and survey protocol. So I'm reasonably relaxed that we shouldn't have too many issues with things being shown at too high a resolution. It's more of a shame that lots of records have to be blurred unnecessarily, because the NBN Atlas can't deal with differential sensitivity in different seasons. As a result, for a species like Purple Sandpiper which is a very rare breeder in the Scottish Highlands which would be prone to disturbance (by photographers) or egg-theft, the NBN Atlas has to blur the other 99.99% of records of wintering birds around the coast. This is largely a bird-specific issue, so I can see why it's perhaps a lower priority (
Additionally, there are a number of differences of opinion over the appropriate blurring levels between the NBN/statutory lists and ours(/RSPB/Rare Breeding Bird Panel etc). In some cases we're stricter, in other cases more lenient. These differences are resolvable with time, but will just take some discussion with a number of parties and that will be easier to undertake now the Atlas is live. I'm wondering if the 12 species here might just be ones where BTO has been more lenient than the NBN system is implementing? Or stricter. I'll have to check in more detail though with my rule-sets...
Thanks again Jim
Hi all,
Data that is on the sensitive species list but provided at a lower resolution (e.g. provided at 10km when list only states 1km) should remain displayed at 10km.
The sensitive species lists are not determined by the NBN. At the moment they have been provided by the statutory agencies (SNH, NRW, NE) but we are planning a workshop around this.
We are also looking at making the resolution of the records visible on the NBN Atlas and will let you know of progress.
Regards,
Mandy
Hi Mandy,
If you go to the example Mark gave, https://records.nbnatlas.org/occurrence … ca08dfa998, and zoom in on the map showing the location of the record you will find a circle of radius 1km is displayed.
Jim Bacon.
Hi All
I have passed Mark's comment about coordinate uncertainty not aligning with coordinate precision to our developer - it is a curiosity and may indicate a bug. Will post again when we get an answer.
Thank you all for your comments on this issue.
Christine
Christine Johnston
NBN Scotland Liaison Officer
If you view the example record on a grid map then it is shown as a 10km square. (Give it time to load and automatically zoom in. It zooms in too far so the calculation of bounds might be based on the 1km size.)
The Atlas is a bit schizophrenic about whether to show records as points or squares.
What I have just noticed is that there are 640 records of adder in this dataset and only one has been marked sensitive. Checking the list of sensitive species and the record details I see this has happened as it has been interpreted as being in England and the species is only sensitive in England, not Scotland (NY36 is about 14% Dumfriesshire and 86% Cumberland). That all makes sense once you know although I'm not sure how Mark will feel about one of his records being considered English!
Jim Bacon
Powered by PunBB 1.4.6, supported by Informer Technologies, Inc.