Fine-scale habitat maps for five study areas in the Falkland Islands
Data and Resources
Additional Info
| Field | Value |
|---|---|
| Last Updated | December 21, 2020, 10:10 (PST) |
| Created | January 30, 2020, 05:26 (PST) |
| Region | Falkland Islands |
| Language | eng |
| Topic Category | Environment; environmental resources, protection and conservation |
| Temporal Extent Start | 2016-09-15 |
| Temporal Extent End | 2019-12-31 |
| Dataset Reference Date | 2019 |
| Lineage | The habitat models developed to create these maps were based on two methodologies; an object-based approach and a pixel-based appoach. The former approach created shapefiles through dissolving segmented polygons classified via a Random Forest classifier at five study sites around the Falklands. The latter approach created raster outputs classified via a Random Forest classifier at the same five sites: (1) Stanley Common and Cape Pembroke , (2) Steeple Jason Island, (3) Cochon Island, Kidney Island and the Murrell Peninsula, (4) Minefield 7 and (5) Port Sussex. Habitat modelling for sites 1 to 3 was based primarily on WorldView 2 & 3 commercial satellite imagery (kindly provided through a Digital Globe Foundation grant). Habitat modelling for sites 4 & 5 were based on orthomosaics created from flying drone mapping surveys of the study site. Specifically, the last site (5) also utilised multispectral drone imagery (Near Infra-Red and Red Edge) in addition to Red, Green and Blue bands, in order to model the invasive weed calafate at Port Sussex. Further information on the habitat mapping workflow can be found in the following document: FalklandIslandsFineScale_CoastalHabitatMapping_Narrative.pdf available with the data. Drone imagery was assessed, compiled, and segmented using Open Source software (QGIS 3.4 for preprocessing, dzetsaka plugin in QGIS 3.4 for classification, SAGA 7.2.0 for segmentation). Groundtruthing was split into training and validation sets and applied to an object-based Random Forest classification. Small scale habitat classes nest into the broad scale habitat map. A confusion matrix pdf file is provided with the data to describe the confidence level of each mapped class. Orthogeorectified drone imageries were used as the baseline for the classifier in the minefield "area" 7 and Port Sussex sites, while Worldview 2 and 3 commercial satellite imagery was the primary imagery source for the remaining sites. Drone-surveyed NIR imagery was also incorporated at the Port Sussex site for the purpose of better isolating the invasive weed Calafate. |
| West Longitude | 187477.456 |
| South Latitude | 4133693.234 |
| East Longitude | 452531.384 |
| North Latitude | 4341392.195 |
| Spatial Reference System | UTM 21S WGS84 |
| Responsible Organisation Name | Hidden (personal data protection) |
| Contact Mail Address | Hidden (personal data protection) |
| Responsible Party Role | Hidden (personal data protection) |
| Access Limitations | Open access |
| Use Constraints | Open, but copyright and/or Intellectual Property Rights apply |
| Resource Reference | Cite data source as: Golding. N, & Black. B., 2020. Final Report from the DPLUS065 Mapping Falklands and South Georgia coastal margins for spatial planning project. SAERI. |
| Data Format | shape |
| Update Frequency | one off |
| Accuracy | Please refer to the confusion matrix |
| Resource Type | Dataset |
| Original Title | Hidden (internal use only) |
| Metadata Date | 2020-01-23 |
| Metadata Point of Contact | datamanager@saeri.ac.fk |
| Contact Consent | Contact details hidden |
| Unique Resource ID | FK-SAERI-584 |