Main Reference to "Antarctic Seafloor AID": REF_JAN_2022.
This repository contains:
- The scoring guide showing samples of 53 VME indicator morpho-taxa, which was used to consistently annotate the imagery dataset, see `scoring_guide_vme_indicator_morpho_taxa_SO.pdf` file.
- The raw percentage cover data, for each grid cell and each morpho-taxon, see `percent_cover_vme_indicator_morpho_taxa_SO.csv` file.
- Image metadata (e.g., year, gear, image quality, sampling effort) can be found in `image_metadata.csv` file.
- Data exploration charts, showing differences across morpho-taxa in terms of prevalence, diversity, or abundance conditioned on detection, see folder `data_exploration`.
Data coordinates have been projected using the Antarctic Polar Stereographic system (EPSG: 3031, WGS: 84). The grid cell resolution is 500m x 500m.
### Percentage cover data
The data in `percent_cover_vme_indicator_morpho_taxa_SO.csv` is organised as follows:
- `cellID`: ID of the grid cell where data is from
- `proj_coord_x`, `proj_coord_y`: cell coordinates in the Antarctic Polar Stereographic system (EPSG: 3031, WGS: 84)
- the remaining columns are the VME indicator morpho-taxa and the values are the percentage cover estimated inside a given 500m x 500m grid cell
### Image metadata
The data in `image_metadata.csv` is organised as follows:
- `cellID`: ID of the grid cell where data is from
- `proj_coord_x`, `proj_coord_y`: cell coordinates in the Antarctic Polar Stereographic system (EPSG: 3031, WGS: 84)
- `age`, in years`: difference of years between the image acquisition and 2022.
- `image_quality_score`, relative score: averaged score across the images located within a given grid cell. The score is computed using the BRISQUE algorithm (Mittal et al., 2012) with the Python package image-quality (version 1.2.7).
- `sampled_portion`, in %: percentage of a grid cell area which has been imaged and scored
### Data exploration
This folder contains some statistics and graphics based on the raw data, as a result of data exploration analyses.
Note about the violin plots: The width of the violins is scaled by the number of observations in that bin. An observation is here the value of a metric (e.g. species diversity) inside a grid cell.
