Gillian Rowan

McGill University
M.Sc. candidate

Supervisor: Margaret Kalacska
Start: 2018-09-04


Aquatic vegetation is a critical component of freshwater ecosystems, providing habitat for small fish and invertebrates, stabilizing sediments, altering flow regimes, and improving water quality. Vegetation communities are, however, facing new and escalating pressures due to climate change and increased human disturbances. Monitoring and managing these ecosystems are therefore vital if the services aquatic plants afford are to be maintained. Remote sensing has been suggested as a preferred method to monitor these habitats, but the technology hasn’t yet seen extensive implementation particularly to submerged aquatic vegetation (SAV) in the freshwater context. The goal of this thesis is to facilitate the application of remote sensing techniques to SAV monitoring through both information transfer and filling in foundational knowledge gaps in the field. A systematic literature review was conducted of previous work in SAV monitoring using remote sensing, which was synthesized with relevant general principles of remote sensing, to create a resource for ecosystem managers and ecological researchers unfamiliar with the discipline. This resource provides an overview of all aspects of a typical optical remote sensing workflow, concentrating on applications to SAV monitoring, to instruct non-specialists in whether and how to adopt remote sensing as a research method. The majority of previous work focused on coastal systems and primarily on determining community extent. While these applications did produce moderate to good results, the narrow scope of the data precludes many critical ecosystem management and research questions from being answered. For this reason, original research was undertaken to implement optical remote sensing techniques to a larger range of targets in a non-ideal (i.e., freshwater) environment. This work assessed the spectral separability of targets under laboratory conditions at various grouping levels (i.e., species to kingdom, vegetation/non-vegetation), under multiple sampling conditions, and modelled across spectral resolutions. In situ imagery was additionally analyzed to compare the expected modelled accuracy to what is possible under field conditions. Samples from thirteen species of SAV were collected across two seasons and were found to be spectrally separable (leave-one-out nearest neighbour criterion of 0.8 to 1 depending on grouping) during the peak of the growing season. Spectral separability depended directly on the spectral resolution and number of bands of the sensor chosen. Airborne hyperspectral imagery was effective with individual class recall of up to 100%, and overall recalls of 88% and 94% when detecting vegetation types and between vegetation or non-vegetation, respectively. Image analysis was limited to targets of canopy-forming or carpeting vegetation and large unvegetated patches due to the ~ 1 m spatial resolution. The targets present in this freshwater ecosystem would be conducive to mapping and monitoring using remote sensing techniques, however, imagery of the spatial and spectral resolution required for such applications is expensive and not widely available. Development of targeted and sufficiently high-spatial resolution sensors aboard space-borne platforms should therefore be prioritized if aquatic vegetation is to be effectively monitored at the regional to global scales.


freshwater, spectral separability, multispectral, remotely piloted aerial system, airborne imagery, spatial resolution, hyperspectral, St. Lawrence River, multi-scale


1- A Review of Remote Sensing of Submerged Aquatic Vegetation for Non-Specialists
Rowan, Gillian S. L., Margaret Kalacska
2021 Remote Sensing