GBIFsiteChecker

Checks and validates location data from the global biodiversity facility (GBIF)

Getting Started

Install devtool package in R install.package("devtools") get GBIFsiteChecker from github devtools::install_github("GBIFsiteChecker/GBIFsiteChecker")

Prerequisites

R Version >= 3.2.3

Example

Decide on the correction level of the GBIFsiteChecker ("strict=1", "correction=2", "flagging=3"). Strict removes false data, "correction" shifts/corrects wrong data and flags them and "flagging" keeps the data as they are downloaded but flags that they are probably false. Enter the startup function with the desired species name, the size definition of the chunks and how many occurences should be looked at.

startup(species_name = "Balaenoptera musculus", records_per_chunk = 200, number_of_records = 4000, correction_level = 2)

Description of the example

The tests validate the GBIF loaction data by comparing the data to external databases: 1) the world register of marine species (WORMS) and a natural earth data shapefile. If, the requested species is present in the WORMS list, and does not match any polygon from the shapefile, the occurrence is flagged as correct. If the species is not found in the WORMS list, it is assumed that the species is terrestrial and therefore is found within a polygon. Thus a point from latitude and longitude (EPSG:4326) from the GBIF data was created, which was tested to be located within a polygon of the Natural Earth shapefile. The data point was plotted in a polygon and returned the ISO2 code of the polygon which it was located in. Further, the returned ISO2 code was intersected with the ISO2 code provided by the GBIF data. In case of a matching code, the data was flagged as valid. If the data was invalid, the lat and or long or lat/long coordinates were swapped and the newly created point was rechecked.

Existing Problems

• Accumulation of occurrences in a cental position of countries (originates from recorded occurrences without coordinates that are artifically relocated in the center of the respecitve country)

• Lacking ISO2 code in some presumably correct occurrences

• Division of marine and terrestrial species neglects aerial species, such as doves that could be detected both on land and ocean

• Other validation checks could be implemented to make the GBIFsiteChecker even more precise, such as a match of elevation of occurrences to the shapefiles, increasing the check precision from two to three dimensions