Open-source satellite image pre-processing and annotation workflows: Stranded whale and dolphin case study

Very high-resolution (VHR) satellite imagery can increase the temporal frequency and spatial scale of wildlife studies and complement conventional monitoring methods, to address data gaps globally. However, most gaps in wildlife monitoring, particularly the study of stranded cetaceans, coincide with less-resourced countries, and commercially operated VHR satellite imagery are costly. Increasing access to imagery, expertise, resources, and open access tools is therefore, essential to achieving a globally scalable solution for wildlife monitoring using VHR satellite imagery. This study provides open access workflows to pre-process and annotate satellite imagery by publishing:
• a step-by-step VHR satellite image pre-processing and annotation workflow for detecting features of interest (geolocated point, bounding box and image chip outputs), developed using QGIS 3.28.
• a replicable step-by-step VHR satellite image pre-processing and annotation workflow in Python 3.10.
• a data standard for the ‘satellites to study stranded cetaceans’ community, including a template for recording annotation metadata and accompanying training document, to enable data sharing internationally.
In the long term, greater equity of access to satellite imagery and workflows like those shared here, will innovate applications and methodological developments to improve environmental management outcomes and empower local communities to lead in this space.