Multispectral airborne imagery and associated classifications, training data and validation data, for mapping nectar-rich floral resources for pollinators, Northamptonshire, UK 2020
Data presented here include imagery with ground-sampling distances of 3 cm and 7 cm for March 2019, May 2019 and July 2019. Also included are the corresponding ground-truth training and verification data presented as shapefiles, as well as the classification output and other data relevant to the project such as the width of floral units.
The imagery was acquired by Spectrum Aviation using A6D-100c (50mm) Hasselblad cameras with bayer filters, mounted on a Sky Arrow 650 manned aircraft. Ground-truth data for training maximum likelihood classifications and for verifying the accuracy of classifications were gathered within eight days of imagery acquisition. Ground-truth data were acquired from sown field margins and hedgerow surrounding one study field.
This dataset was acquired from March to July 2019 at a farm in Northamptonshire, UK. Data were acquired as part of a NERC funded iCASE PhD studentship (NERC grant NE/N014472/1) based at the University of East Anglia and in collaboration with Hutchinsons Ltd. The aim of the research was to map the floral units of five nectar-rich flowering plant species using very high resolution multispectral imagery. Each species constitutes an important food resource for pollinators. The plant species in question were Prunus spinosa, Crataegus monogyna, Silene dioica, Centaurea nigra and Rubus fruticosus.
The imagery was acquired by Spectrum Aviation using A6D-100c (50mm) Hasselblad cameras with bayer filters, mounted on a Sky Arrow 650 manned aircraft. Ground-truth data for training maximum likelihood classifications and for verifying the accuracy of classifications were gathered within eight days of imagery acquisition. Ground-truth data were acquired from sown field margins and hedgerow surrounding one study field.
This dataset was acquired from March to July 2019 at a farm in Northamptonshire, UK. Data were acquired as part of a NERC funded iCASE PhD studentship (NERC grant NE/N014472/1) based at the University of East Anglia and in collaboration with Hutchinsons Ltd. The aim of the research was to map the floral units of five nectar-rich flowering plant species using very high resolution multispectral imagery. Each species constitutes an important food resource for pollinators. The plant species in question were Prunus spinosa, Crataegus monogyna, Silene dioica, Centaurea nigra and Rubus fruticosus.
Publication date: 2021-07-26
Where/When
- Study area
-
- Temporal extent
-
2019-05-15 to 2020-07-08
Provenance & quality
Multispectral data were acquired across red, green, blue and near-infrared bands on 28 March, 15 May and 4 July. Prior to imagery acquisition, 40X60cm boards that were visible within the imagery were established within the margins surrounding one field at our study farm. These were used as ground-control points to locate clusters of floral units within the margins. Ground-truth data (i.e. the locations of floral units) were gathered within 8 days of imagery acquisition.
The company that acquired the multispectral data (Spectrum Aviation) carried out orthorectification and stitched individual images together into an orthomosaic. Data values were kept in a raw digital number format.
Between July 2019 and February 2021, further image processing, e.g. clipping of the image and removal of irrelevant spectral bands was carried out in QGIS. During the same timeframe, ground-truth data were divided into data to be used for training the classifications and for carrying out independent accuracy assessments. The maximum likelihood classifications were applied to the imagery with different iterations applied each time, i.e. the training sets were tweaked to increase classification accuracies.
The company that acquired the multispectral data (Spectrum Aviation) carried out orthorectification and stitched individual images together into an orthomosaic. Data values were kept in a raw digital number format.
Between July 2019 and February 2021, further image processing, e.g. clipping of the image and removal of irrelevant spectral bands was carried out in QGIS. During the same timeframe, ground-truth data were divided into data to be used for training the classifications and for carrying out independent accuracy assessments. The maximum likelihood classifications were applied to the imagery with different iterations applied each time, i.e. the training sets were tweaked to increase classification accuracies.
Citations
Barnsley, S. L., Lovett, A. A., & Dicks, L. V. (2022). Mapping nectar-rich pollinator floral resources using airborne multispectral imagery. In Journal of Environmental Management (Vol. 313, p. 114942). Elsevier BV.
https://doi.org/10.1016/j.jenvman.2022.114942
Correspondence/contact details
Other contacts
- Custodian
-
NERC EDS Environmental Information Data Centreinfo@eidc.ac.uk
- Publisher
-
NERC EDS Environmental Information Data Centreinfo@eidc.ac.uk
- Rights Holder
-
University of East Anglias.barnsley@uea.ac.uk
- Rights Holder
-
University of Cambridgelvd22@cam.ac.uk
Additional metadata
- Topic categories
- Biota , Environment
- Keywords
- blackthorn, bramble, Centaurea nigra, common knapweed, Crataegus monogyna, DeWALT laser beam measure, Environmental survey, hardhead, hawthorn, hedgerow, Land cover, Pollinators, Prunus spinosa, QGIS version 3.4.15, red campion, Rubus fruticosus, SCP plugin, semi-automatic classification, Silene dioica
- INSPIRE Theme
-
Habitats and Biotopes
Land Cover - Funding
- Natural Environment Research Council Award: NE/N014472/1
- Spatial representation types
-
Raster
Vector - Spatial reference system
- OSGB 1936 / British National Grid
- Last updated
- 12 July 2022 09:11