Spatial Precision – What’s this all about?

8 December, 2011

Excerpt from the document, ‘Darwin Core Key Spatial Concepts’
Written by Simon Bennett, Atlas of Living Australia, 29 November 2011.

The full document, includes definitions and examples for each of the terms: http://tinyurl.com/6t6zyaa

Darwin Core Key Spatial Concepts

The following discussion provides a perspective on Darwin Core (TWDG 2009) based occurrences records as they relate to the fitness-for-use of occurrence data for spatial analyses such as species distribution mapping and spatial modelling.

Fitness-for-use often refers to a scale of spatial data quality that changes with varying data accuracy, precision and intended use (GBIF 2010). For some uses, such as continental scale distribution mapping relatively imprecisely located spatial can be used. While if the intended use is for fine-scale habitat mapping, then only reliable, precisely located spatial occurrences should be used.

The Location

In order to assess the fitness-for-use of a species occurrence record it is essential to know the:

  1. geographic coordinates;
  2. precision at which coordinates where collected;
  3. extent of the location where the occurrences is reported from; and
  4. geodetic datum (spatial reference system) under which the geographic coordinates where collected.
  5. Uncertainties associated with determining the coordinates

Terms (2), (3), (4) and (5) are added together to produce an overall measure of the uncertainty associated with a coordinate pair. See the following diagram as an example.

Spatial Concepts on Google Maps
Key spatial concepts link to Google Maps

Spatial Concepts text in Google Maps above

The location is a 2 ha survey plot. The coordinates are at the centre of the location. The geocode uncertainty is the error in recording the coordinates: typically 5m with a GPS. The location extent is the radius of a circle encompassing the actual location. If the datum is not recorded then there may be an additional 200m+ datum shift error. Coordinate precision relates to the number of decimal places the coordinates have (7 places or 0.0000001 with Google Maps).

The overall coordinateUncertaintyInMeters is geocoding uncertainty + location extent + potential datum shift (if datum unknown) + coordinate precision (<0 in this case) = 105m or 305m if datum is unknown. Note: Darwin Core does not contain terms for location extent or geocode uncertainty.

Data generalisation

Sometimes coordinates are generalised to protect sensitive species or to protect intellectual property. Common methods are to (i) round coordinates to fewer decimal places, (ii) grid the data or (iii) randomly move the location. If generalised then the method of generalization must be recorded in DwC:dataGeneralizations.

Where spatial generalisation does occur a new overall uncertainty should created by adding the uncertainty introduced by the generalisation process to the original coordinateUncertaintyInMeters. The original coordinateUncertaintyInMeters should be preserved as a ‘verbatim’ field.

Explicit verification of location

Where an occurrence record is on the edge of its range, or is a range extension, then it is useful to know whether the location coordinates have been verified (DwC:georeferenceVerificationStatus).

Determining Overall Spatial Uncertainty

The overall measure of spatial uncertainty is DwC:coordinateUncertaintyInMeters: ‘The horizontal distance (in meters) from the given decimalLatitude and decimalLongitude describing the smallest circle containing the whole of the Location’. Typically it is additive of distances/errors and associated spatial uncertainty: typically the extent of the location and any uncertainty related to the datum and coordinate precision. John Wieczorek (in lit.) notes that DwC:coordinateUncertaintyInMeters is exactly equivalent to Maximum Error Distance used in his Georeferencing Calculator.

A knowledge of the coordinates, their precision and geodetic datum, and the extent of location is essential to determine the overall spatial uncertainty of a location. Wieczorek (2007) demonstrates the calculation of the potential error in these terms to provide an overall measure of spatial uncertainty, the Maximum Uncertainty Distance = DwC:coordinateUncertaintyInMeters. The methods used in the Georeferencing Calculator are outlined in the MaNIS/HerpNet/ORNIS Georeferencing Guidelines, available at http://manisnet.org/GeorefGuide.html

The Georeferencing Calculator (http://manisnet.org/gc.html) is an interactive tool that determines the Maximum Uncertainty Distance (= DwC:coordinateUncertaintyInMeters) based on coordinate precision (DwC:coordinatePrecision in decimal degrees), the extent of the locality (locationExtent), and knowledge of the geodetic datum (DwC:geodeticDatum) and other uncertainties. Note that terms ‘Coordinate Precision’ and ‘Coordinate Uncertainty’ are sometimes confused and have been used interchangeably. So take care when using these terms.

Determining fitness-for-use for spatial analysis

GBIF (2010) notes that fitness-for-use refers to a scale of data quality that changes with the varying data accuracy, precision and intended use. For some applications, data quality can be relatively low and still be fit for use. In the context of geospatial data, we can split fitness-for-use into two broad categories: (i) Are the geospatial data correct? (ii) Are the geospatial data usable at the geographic scale of the question? For example, coarse scale geospatial data may only be usable for continental or global analyses but certainly not for local analyses. They also note that the minimum information needed to assess spatial fitness for use is a latitude-longitude-geographic uncertainty triplet.

Other important factors to consider in assessing overall fitness-for-use:

Future Developments

The focus of further work will be development of additional methods for validation of occurrence data, and development of explicit metrics and summaries to enable a user to make informed choices on spatial, taxonomic and temporal fitness-for-use.

Bibliography

ALA. 2011. Draft Darwin Core: Quality Model.

Chapman, A.D. 2005. Principles and Methods of Data Cleaning – Primary Species and Species-Occurrence Data.

Chapman, A.D. 2005. Uses of Primary Species-Occurrence Data. Version 1.0. Report for the Global Biodiversity Information Facility, Copenhagen.

GBIF. 2010. GBIF Position Paper on Future Directions and Recommendations for Enhancing Fitness-for-Use across the GBIF Network.

MaNIS/HerpNet/AORNIS. Undated. Georeferencing Guidelines. Available at http://manisnet.org/GeorefGuide.html

TDWG. 2009. Darwin Core Terms: A quick reference guide. Available at http://rs.tdwg.org/dwc/terms/

Wieczorek, J.R 2007. Georeferencing Calculator. Available at http://manisnet.org/gc.html.

Comments are closed.