Methodology:

This dataset uses as input the following datasets: 'ERA5 monthly averaged data on single levels from 1940 to present' sub-variables, 'Total cloud cover' and '10m wind speed' (see https://doi.org/10.24381/cds.f17050d7)

'Total cloud cover' (0 and 1 or 0% and 100%) is the portion of a grid cell covered by cloud, and integrates all cloud information from the Earth's surface to the top of atmosphere. '10m wind speed' (ms-1) represents horizontal wind speed in meters per second, at 10m above the Earth's surface. For details on precisely how to download ERA5 data, see supplemental_ material_ 2 in the associated manuscript or view downloading_era5_data.pdf in the associated Github repository: https://github.com/PennyJClarke/feasibility-mapping.

The bounding box coordinates of the three study areas were selected to encompass the entirety of the Exclusive Economic Zone (EEZ) of the study area, and used to delineate and extract wind and cloud data from the ECMWF reanalysis ERA5 data.

The assessment of all data was conducted using open source QGIS 3.28 pyQGIS console. An open-source pipeline to identify areas that meet a threshold of environmental variables in extracted ERA5 data, is available at: https://github.com/PennyJClarke/feasibility-mapping and a description follows.

All years of data, (1940-2023), were downloaded and later subset to extract the most recent five years (2018-2022) of data for environmental variables 'Total cloud cover' and '10m wind speed'. To perform a threshold analysis, for each month, first we calculated the five-year monthly median for 'Total cloud cover' and '10m wind speed' from 2018-2022 (.tif files). Median was selected for its greater representation of central tendency. The monthly median outputs were then masked to retain only those areas that met or were below the predefined threshold. The masked areas of both variables were converted into a vector (polygon shapefile).

For live cetaceans, both vectors were used to perform an intersection (extraction of overlapping features from the two input layers) of the matching monthly 'Total cloud cover' and '10m wind speed'. The resulting final 'suitable area(s)', where both 5 year monthly average environmental variables met the predefined threshold (vector polygon shapefiles), was mapped to the study area to develop feasibility maps. For stranded cetaceans, the outputs for 'Total cloud cover' alone, masked by the coastline (to include 2km buffer either side of the coastline) form the resulting final 'suitable area(s)' to develop feasibility maps. Finally, for projects interested in exploring stranded and floating dead/live cetaceans ('float_and_strand'), the 'suitable area(s)' for both live and stranded cetaceans were merged and boundaries dissolved, to develop feasibility maps. In addition, to understand the (interannual) variability in the dataset, the standard deviation was extracted and mapped (.tif format).

To evaluate monthly average 'Total cloud cover', we devised thresholds to define the likelihood of achieving a suitable image collection (see table 1 below and detailed further details in the associated paper). The term suitable image is used to describe an image that is cloud free or contains thin or spatially restricted cloud, which does not render an image unusable and for live cetaceans only, where the sea state is below Beaufort 4 (equal 8 ms-1, for studies monitoring small cetacean species, it may be sensible to be conservative and reduce the wind speed threshold to Beaufort 3, equal to 5 ms-1).

Table 1 - ERA5 'Total Cloud Cover (TCC)' monthly average thresholds (%) and associated likelihood of achieving a suitable image collection. 'TCC' is given as a value between 0 and 1, as probability is a value between 0 and 1 or 0% and 100% inclusive, therefore, to assess the likelihood (percentage likelihood) of achieving a suit...(12)

Data collection:

The assessment of all data was conducted using open source QGIS 3.28 pyQGIS console. An open-source pipeline to identify areas that meet a threshold of environmental variables in extracted ERA5 data, is available at: (https://github.com/PennyJClarke/feasibility-mapping)

Data quality:

The datasets were extracted from the European Centre for Medium-range Weather Forecasts (ECMWF) reanalysis ERA5 data. The ERA5 data used in the feasibility mapping exercise is a synthetic version of Earth's atmospheric conditions, and its performance is dependent upon the quality of observed and modelled data. As observation records are spatially and temporally biased and inconsistent in their standards, modelled data will be less accurate in regions with fewer observations. We used monthly averages to map feasibility; more refined daily products available through ERA5 could provide higher resolution information about satellite imaging feasibility than our exploration. Regardless of the timescale; daily; hourly; monthly; or yearly, analysing averages can conceal anomalous periods.

ERA5 data is a global dataset, which is advantageous for the replicability and comparability of these feasibility assessments on a global scale. However, in the future, regional reanalysis datasets or locally derived observation datasets, may provide a higher resolution alternative to ERA5, which, when combined with local knowledge, could achieve more accurate local-level assessments. The spatial resolution of ERA5 data is much coarser than the VHR optical satellite imagery we are interpreting feasibility for. The data are also much coarser than the conditions which they are representing, and which conservation managers require to make informed decisions.

ERA5 is a gridded dataset with a native horizontal resolution of approximately 31km (0.28125 degrees), which, when downloaded, is reprojected to a 0.25 by 0.25 degrees latitude/longitude (the resolution of the data in kilometres is consistent across latitude, however, longitude decreases poleward due to convergence).