Drone-Based Optical and Thermal Remote Sensing for Salmonid Habitat and Hydrological Applications in Eastern Canada

Abstract

Remote sensing has long been a valuable tool in environmental monitoring, with applications spanning hydrology, aquatic ecosystem management, and watershed research. Habitat degradation, including river warming, poses a significant threat to cold-water anadromous fish populations. Mapping of river habitats with satellite imagery is often limited by coarse spatial resolutions especially in the thermal band. Advancements in high-resolution drone-based optical and thermal surveying presents an efficient approach for mapping and monitoring salmonid habitat, particularly cold-water refugia in thermally stressed rivers, as well as groundwater-surface water interactions. This study surveyed two rivers in Prince Edward Island using photogrammetry and GIS workflows for image processing and analysis. By integrating optical and thermal datasets, the developed workflows provide a comprehensive approach to modelling riverine environments. Applications include watershed monitoring, habitat conservation, and telemetry research, such as analyzing habitat selection and fish behavior in thermally stressed rivers. Surveys were conducted in fall 2024 using a DJI Mavic 3T Enterprise drone equipped with RGB (12 MP and 48 MP) and thermal infrared (640 × 612 px) sensors. Photogrammetric ortho-mosaics, digital surface models, and GIS-derived data layers were generated to assess in-stream and riparian habitat. Data processing involved RStudio and ArcGIS Pro for spatial analysis (e.g., supervised classification, point sampling, zonal statistics), while initial photogrammetric reconstructions were performed using Agisoft Metashape. Results demonstrated how drone-based thermal imaging effectively captures fine-scale thermal heterogeneity, supporting habitat assessment and hydrological modeling. Generalized additive models were used to analyze temperature variations in relation to key environmental variables. Discussions include best practices for reporting uncertainty, mitigating methodological pitfalls, and integrating drone-based remote sensing into ecological and hydrological research. These findings highlight the potential of drone-based approaches to support movement ecology, watershed management, and conservation planning.