search

gbif / geocode

Geocode webservice and client
Apache License 2.0
4 stars 4 forks source link

Protected areas #21

Open MattBlissett opened 1 year ago

MattBlissett commented 1 year ago

I'm adding protected areas from https://www.protectedplanet.net/ for occurrence interpretation.

Note the user manual: https://wdpa.s3-eu-west-1.amazonaws.com/WDPA_Manual/English/WDPA_WDOECM_Manual_1_6.pdf

There are a few things to decide.

Interpretation:

  1. How should we handle point protected areas, where all we have for the area is a single point? See §5.5.2 "Point data" in the user manual. I've followed the advice there and calculated a geodesic buffer of the appropriate size around the points, although not all points are provided with an area. For those, I've also followed the advice and ignored them.

https://api.gbif-dev.org/v1/geocode/reverse?lat=29.86217&lng=-6.26089

  1. Distance. For GADM, we have assigned an administrative area to an occurrence only if the point is within the administrative boundary polygon, ignoring coordinate uncertainty. More careful analysis might want to do something with points near the protected area having a sufficiently large coordinate uncertainty, or those inside it where the uncertainty means they may be outside. However, putting too much in the area would make the filter less useful for finding occurrences outside these areas.

Data access:

  1. We can store the identifiers (integers) in interpreted occurrences. We'll need an API to map those identifiers to the names and other information on the protected areas. UNEP-WCMC have one, although they don't seem keen on people using it, and it excludes commercial users. If we make our own, we need to work out what information it should provide. Going the other way, we'll need an API to allow searching protected areas in order to use them in occurrence searches. Searching on name, ISO country, types and location (coordinates) seems appropriate.

  2. Do we also need tiles, so we can display the protected areas, for search and/or on occurrence maps?

  3. (Note providing GeoJSON would need UNEP-WCMC's permission.)

Updates:

  1. The dataset is published monthly, although the May version is not yet available. It's a particularly large dataset, so automating the updates will be more work than I anticipated.

Attribution:

  1. We need to work out how our use of this data should be attributed. Just https://www.gbif.org/acknowledgements or more?

CC @andrewrodrigues, possibly @timhirsch.

andrewrodrigues commented 1 year ago
  1. I think the approach of removal of points with null values for area is the only sensible approach to handling these data. I would suggest some accompanying information to explain to users that some points are missing.
  2. I would argue that we should be consistent in our approach to suing spatial filters to ease communication with users. Either we include uncertainties in all our spatial filters (GADM, PAs ect) or not. I would be comfortable to leave the filtering of data for uncertainties to the user post download.
  3. I am assuming the identifier is the WDPA ID? What do you mean by type? IUCN Category might be a useful as well as governance type. Other suggested searches seem appropriate.
  4. Visualisation of PAs would be great from a UI perspective
  5. The user agreement has expired and we need to extend this @timhirsch will follow up on this. The acknowledgements page is the place to put this.
andrewrodrigues commented 1 year ago

On attribution, visualization of the dataset should link to dataset terms and conditions of use - https://www.protectedplanet.net/en/legal and to the download of the dataset as per the original agreement