Trends and Developments in Biodiversity Informatics
Flora brasiliensis Revisited
Automated georeferencing web services for natural history collections
Reed Beaman, Natural History Museum and Biodiversity Research Center,
University of Kansas
|
An estimated 2.5 billion biological specimens are curated in biological
museums worldwide. Of the various classes of information linked to
biological specimens, geographic coordinates used for mapping species
distributions are among the most widely demanded by the scientific
community and the general public. Providing these coordinates
(georeferencing) has proved to be a significant challenge for data
custodians, floristicians, and monographers. Manual georeferencing of
collection localities from specimen label data, using maps and
gazetteers is typically a time consuming task.
Biodiversity informatics researchers are engaged in providing global
network access and developing tools to facilitate processing and
analysis of data associated with biological collections. Among these,
automated georeferencing web services may be leveraged to increase the
efficiency with which collections can be georeferenced. Our prototype
(http://georef.nhm.ku.edu) was able to georeference between 77 and 87
percent of locality records in three test data sets, with 12 to 26
percent of these showing some level of ambiguity through multiple
matches. Spatial results are comparable to those done by trained
specialists. Best use practices require that georeferences obtained by
automated methods should be reviewed, particularly to disambiguate
multiple matches and to calculate error values. Through planned
collaboration between CRIA, UCBMVZ, and KUNHM, we are developing
provisions for on-the-fly error calculation, georeferences as spatial
footprints, interoperable georeferencing services, and taxon-specific
validation through predictive species modeling.
webmaster@cria.org.br