[API-users] birdhouse meets GBIF

Tim Robertson trobertson at gbif.org
Thu Jun 2 14:45:21 CEST 2016

Hi Nils

We don’t have any OGC services, but there is an asynchronous download API which can deliver CSVs.
Off the top of my head, the only way you can automate this at the moment would be to do periodically issue a download (e.g. Daily) process as you see fit, and cache the result for your app.

In the download API you can get any size from 1 - > 660 million records, which is why it is asynchronous.  It’s used a lot by various applications and communities.

I hope this helps,

From: API-users <api-users-bounces at lists.gbif.org<mailto:api-users-bounces at lists.gbif.org>> on behalf of Nils Hempelmann <info at nilshempelmann.de<mailto:info at nilshempelmann.de>>
Date: Thursday 2 June 2016 at 14:25
To: "api-users at lists.gbif.org<mailto:api-users at lists.gbif.org>" <api-users at lists.gbif.org<mailto:api-users at lists.gbif.org>>, "wps-dev at lists.dkrz.de<mailto:wps-dev at lists.dkrz.de>" <wps-dev at lists.dkrz.de<mailto:wps-dev at lists.dkrz.de>>
Subject: [API-users] birdhouse meets GBIF

Dear all

Here is an current problem to solve :-) for the birdhouse WPS and GBIF developers.
There is a species distribution process in the birdhouse WPS, processing Climate data based on occurrence coordinates from GBIF.

Currently the GBIF data are given to the process via an url of the GBIF csv (which had to be generated manually before).
I was implementing a direct occurrence search from birdhouse to GBIF with pygbif. Worked fine BUT: there is a limit of 300 records which is far to less to train the Species distribution model.

So here the question: Is there a way to request the species occurrence coordinates somehow directly (e.g. like a WPS request)?

(The previous conversation is following)

And some link for the background informations:


Species distribution process:

Birdhouse architecture:


-------- Forwarded Message -------
Subject:        Re: pygbif for occurrence coordinates
Date:   Thu, 2 Jun 2016 09:00:49 -0300
From:   Mauro Cavalcanti <maurobio at gmail.com><mailto:maurobio at gmail.com>
To:     Nils Hempelmann <info at nilshempelmann.de><mailto:info at nilshempelmann.de>
CC:     wps-dev at lists.dkrz.de<mailto:wps-dev at lists.dkrz.de>, Carsten Ehbrecht <ehbrecht at dkrz.de><mailto:ehbrecht at dkrz.de>, Wolfgang.Falk at lwf.bayern.de<mailto:Wolfgang.Falk at lwf.bayern.de>, Scott Chamberlain <myrmecocystus at gmail.com><mailto:myrmecocystus at gmail.com>


No, there is no other way to get at once a larger number of records using the GBIF API (although one could be able to achieve this by sequentially querying the server using "batches" of 200 or 300 records each). As I said, there are good operational reasons for the GBIF developers to have imposed such limit.

As of your other question, it should better be put to the GBIF developers themselves (in the discussion list, so that we can all benefit from the answers! :-))

With warmest regards,

Dr. Mauro J. Cavalcanti
E-mail: maurobio at gmail.com<mailto:maurobio at gmail.com>
Web: http://sites.google.com/site/maurobio

Em 02/06/2016 08:49, "Nils Hempelmann" <info at nilshempelmann.de<mailto:info at nilshempelmann.de>> escreveu:
Hi Mauro

Oh ...
No way to set it unlimited?
The 'manual via browser' option and parsing the returning csv is the current status.
or any alternative to pygbif?

If I understood it correctly, GBIF database is organized as Web Server, so I gues there should be a way to connect to the birdhouse WPS, am I right?

(put the web-dev list in copy)


On 02/06/2016 12:40, Mauro Cavalcanti wrote:


That's a limit imposed (for good operational reasons) by the GBIF API. If you want a larger number of records, you'll have to download them "manually" (that is, via browser) and then parse locally the csv file returned from the GBIF server.

Hope this helps.

Best regards,

Dr. Mauro J. Cavalcanti
E-mail: maurobio at gmail.com<mailto:maurobio at gmail.com>
Web: http://sites.google.com/site/maurobio

Em 02/06/2016 04:48, "Nils Hempelmann" <info at nilshempelmann.de<mailto:info at nilshempelmann.de>> escreveu:
Hi Scott

works fine. Thanks a lot. Just have a question for the search limits:

The maximal records seems to be limited to 300. Is that on purpose?
And requesting more than 200000 gives a 'bad request'


In [68]: len( occurrences.search(taxonKey=key, limit=100)['results'])
Out[68]: 100

In [69]: len( occurrences.search(taxonKey=key, limit=300)['results'])
Out[69]: 300

In [70]: len( occurrences.search(taxonKey=key, limit=3000)['results'])
Out[70]: 300

In [71]: len( occurrences.search(taxonKey=key, limit=200000)['results'])
Out[71]: 300

In [72]: len( occurrences.search(taxonKey=key, limit=200001)['results'])
HTTPError                                 Traceback (most recent call last)
<ipython-input-72-2f7d7b4ccba0> in <module>()
----> 1 len( occurrences.search(taxonKey=key, limit=200001)['results'])

/home/nils/.conda/envs/birdhouse/lib/python2.7/site-packages/pygbif/occurrences/search.pyc in search(taxonKey, scientificName, country, publishingCountry, hasCoordinate, typeStatus, recordNumber, lastInterpreted, continent, geometry, recordedBy, basisOfRecord, datasetKey, eventDate, catalogNumber, year, month, decimalLatitude, decimalLongitude, elevation, depth, institutionCode, collectionCode, hasGeospatialIssue, issue, q, mediatype, limit, offset, **kwargs)
    251         'collectionCode': collectionCode, 'hasGeospatialIssue': hasGeospatialIssue,
    252         'issue': issue, 'q': q, 'mediatype': mediatype, 'limit': limit,
--> 253         'offset': offset}, **kwargs)
    254     return out

/home/nils/.conda/envs/birdhouse/lib/python2.7/site-packages/pygbif/gbifutils.pyc in gbif_GET(url, args, **kwargs)
     17 def gbif_GET(url, args, **kwargs):
     18   out = requests.get(url, params=args, headers=make_ua(), **kwargs)
---> 19   out.raise_for_status()
     20   stopifnot(out.headers['content-type'])
     21   return out.json()

/home/nils/.conda/envs/birdhouse/lib/python2.7/site-packages/requests/models.pyc in raise_for_status(self)
    843         if http_error_msg:
--> 844             raise HTTPError(http_error_msg, response=self)
    846     def close(self):

HTTPError: 400 Client Error: Bad Request for url: <http://api.gbif.org/v1/occurrence/search?taxonKey=2882316&limit=200001&offset=0> http://api.gbif.org/v1/occurrence/search?taxonKey=2882316&limit=200001&offset=0

In [73]:

On 02/06/2016 02:22, Scott Chamberlain wrote:
Fixes to docs and fix to download_get are up on Github now. Will push new version to pypi soon. Let me know if some thing still don't work for you after reinstalling from github


On Wed, Jun 1, 2016 at 4:51 PM Nils Hempelmann <info at nilshempelmann.de<mailto:info at nilshempelmann.de>> wrote:
Hi Scott and Mauro

Mauro sended me a snipped of code which worked fine:

I found the example here:

and here:

Thanks a lot very great. That enables a lot ;-)
I ll keep you posted


On 02/06/2016 01:45, Scott Chamberlain wrote:
And where are those example from exactly? I don't see those examples searching the repo (which includes all docs).

`pygbif.name_suggest` wouldn't work because `name_suggest` is a method in the `species` module, so you'd have to do pygbif.species.name_suggest, or `from pygbif import species`, then `species.name_suggest`

Looks like occ.get(taxonKey = 252408386)  is a documentation bug, that should be `key` instead of `taxonKey`, a copy paste error. will fix that.

The `occ.download_get` call has a small bug, will fix that

All other calls work for me


On Wed, Jun 1, 2016 at 4:34 PM Scott Chamberlain <<mailto:myrmecocystus at gmail.com>myrmecocystus at gmail.com<mailto:myrmecocystus at gmail.com>> wrote:

What version of pygbif are you on? And what version of Python>

Best, Scott

On Wed, Jun 1, 2016 at 4:02 PM Nils Hempelmann <<mailto:info at nilshempelmann.de>info at nilshempelmann.de<mailto:info at nilshempelmann.de>> wrote:
Hi Mauro and Scott

Was checking out pygbif. seems to be a very useful tool.

Can you help me with the syntax (or forward me to the appropriate person
;-) )
The given snippets of code are outdated.

I am basically just looking for the occurrence coordinates:
here is my first try :

import pygbif
occ = pygbif.occurrences.search(scientificName='Fagus sylvatica')

... and further? ;-)

the examples in the docu are throwing errors:

key=  pygbif.name_suggest(q='Helianthus annuus',rank='species')['key']

from pygbif import occurrences as occ occ.search(taxonKey = 3329049)
occ.get(taxonKey = 252408386) occ.count(isGeoreferenced = True)
occ.download_list(user  =  "sckott",  limit  =  5)
occ.download_meta(key  =  "0000099-140929101555934")


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.gbif.org/pipermail/api-users/attachments/20160602/f5d71ba6/attachment-0001.html>

More information about the API-users mailing list