occurrence search returns observation outside specified polygon provided in geometry parameter

[aubreymoore]

When I search for Oryctes rhinoceros within a bounding box for Guam, I get an occurrence record for the Philippines. Using python 2.7/pygbif 0.2.0.

from pygbif import occurrences as occ
occ.search(scientificName='Oryctes rhinoceros',
geometry='POLYGON((144.61 13.22, 144.61 13.66, 144.96 13.66, 144.96 13.22, 144.61 13.22))',
limit=1)

Returns:

{u'count': 3,
u'endOfRecords': False,
u'facets': [],
u'limit': 1,
u'offset': 0,
u'results': [{u'acceptedScientificName': u'Oryctes rhinoceros (Linnaeus, 1758)',
u'acceptedTaxonKey': 4995642,
u'basisOfRecord': u'HUMAN_OBSERVATION',
u'catalogNumber': u'4864787',
u'class': u'Insecta',
u'classKey': 216,
u'collectionCode': u'Observations',
u'coordinateUncertaintyInMeters': 88056.0,
u'country': u'Philippines',
u'countryCode': u'PH',
u'crawlId': 141,
u'datasetKey': u'50c9509d-22c7-4a22-a47d-8c48425ef4a7',
u'datasetName': u'iNaturalist research-grade observations',
u'dateIdentified': u'2016-12-30T15:01:30',
u'day': 23,
u'decimalLatitude': 13.4756,
u'decimalLongitude': 120.851388,
u'eventDate': u'2016-12-23T04:00:00',
...

_

[sckott]

thanks for the report @aubreymoore having a look

[sckott]

looksl like your WKT has the wrong winding direction.

GBIF docs have for the geometry parameter https://www.gbif.org/developer/occurrence#p_geometry

I'll update the docs on this.

If you reverse the winding order of your WKT it works as expected I think:

geom_rev="POLYGON((144.61 13.22,144.96 13.22,144.96 13.66,144.61 13.66,144.61 13.22))"
res2 = occ.search(scientificName='Oryctes rhinoceros', geometry=geom_rev, limit=10)
[ z['country'] for z in res2['results'] ]

['Guam',
 'Guam',
 'Guam',
 'Guam',
 'Guam',
 'Guam',
 'Guam',
 'Guam',
 'Guam',
 'Guam']

[ [z['decimalLatitude'], z['decimalLongitude']] for z in res2['results'] ]

[[13.491572, 144.845642],
 [13.430427, 144.801009],
 [13.474637, 144.783201],
 [13.430502, 144.800539],
 [13.511599, 144.803081],
 [13.427608, 144.788581],
 [13.521747, 144.879896],
 [13.588958, 144.867862],
 [13.49438, 144.776794],
 [13.493889, 144.77417]]

I haven't found an easy Python lib for reversing winding order, but in R, you could

install.packages("wicket")
x = 'POLYGON((144.61 13.22, 144.61 13.66, 144.96 13.66, 144.96 13.22, 144.61 13.22))'
wicket::wkt_reverse(x)

[aubreymoore]

Thanks for the explanation. And thanks for maintaining this useful package.

[sckott]

okay, this should do it:

from geojson_rewind import rewind
from geomet import wkt
import decimal
import statistics

# from https://stackoverflow.com/a/12472564/1091766
def flatten(S):
    if S == []:
        return S
    if isinstance(S[0], list):
        return flatten(S[0]) + flatten(S[1:])
    return S[:1] + flatten(S[1:])

def wkt_rewind(x):
    '''
    change WKT winding order

    :param x: WKT string

    Usage::

        x = 'POLYGON((144.61 13.22, 144.61 13.66, 144.96 13.66, 144.96 13.22, 144.61 13.22))'
        wkt_rewind(x)
    '''
    z = wkt.loads(x)
    coords = z['coordinates']
    nums = flatten(coords)
    dec_n = [ decimal.Decimal(str(w)).as_tuple().exponent for w in nums ]
    digits = abs(statistics.mean(dec_n))
    wounded = rewind(z)
    back_to_wkt = wkt.dumps(wounded, decimals = digits)
    return back_to_wkt

[sckott]

the function is now in the library if you reinstall