Novel approach to large‐scale monitoring of submerged aquatic vegetation: A nationwide example from Sweden

Abstract

This article is part of the special series "The Future of Marine Environmental Monitoring and Planning." The series will take a sneak peek into the future of marine monitoring where integration of new monitoring technologies with advanced ecosystem modeling will make it possible to estimate real-time ecosystem status, improve model precision, and provide a robust basis for marine environmental assessments.

Submerged aquatic vegetation: Overview of monitoring techniques used for the identification and determination of spatial distribution in European coastal waters

“…Most studies using passive optical satellite imagery for seagrass mapping have focused on local to regional scales, typically using open and freely available imagery from the Copernicus Sentinel‐2 and NASA Landsat 8 missions (e.g., Dierssen et al, 2019; Fritz et al, 2019; Hogrefe et al, 2014; Xu et al, 2021; Yadav et al, 2017), which collect data at moderate spatial resolution (10–30 m). Only recently, with the advent of the Sentinel‐2 fleet, combined with technical developments in cloud‐computing and machine learning, large‐scale mapping became feasible (e.g., Huber et al, 2022; Traganos et al, 2018). In addition to the free imagery, several studies have successfully mapped coastal habitats with commercial satellite imagery, providing spatial resolutions down to the sub‐meter scale (Table 1; e.g., Fritz et al, 2017; Marcello et al, 2018; McLaren et al, 2019; Pu & Bell, 2017; Roelfsema et al, 2014; Traganos & Reinartz, 2018).…”

“…Other simpler methods, such as empirical log-ratio methods (Lyzenga, 1978), can also be used to derive a bathymetry map for the analyzed area using satellite data. Additional variables, such as depth-invariant indices, simple convolutions of the spectral bands, and additional independent layers, such as maps of exposure estimates, can further be included as input features to the classification algorithm (Huber et al, 2022).…”

Global climate change and the Baltic Sea ecosystem: direct and indirect effects on species, communities and ecosystem functioning

“…Research on effects of climate change would also benefit from methodological diversity. For example, more extensive use of biochemical and genetic methods, such as biomarkers (Turja et al, 2014(Turja et al, , 2015Villnäs et al, 2019), stable isotopes (Voss et al, 2000;Gorokhova et al, 2005;Morkune et al, 2016;Lienart et al, 2021), compound-specific isotope analyses (Ek et al, 2018;Weber et al, 2021) or metabarcoding (Leray and Knowlton, 2015;Bucklin et al, 2016;Klunder et al, 2022), as well as development of remote sensing methods (Huber et al, 2021), could yield novel information on stress levels experienced by organisms and environmental niches preferred by species. Such information would allow validation of the biogeochemical models under different environmental and climate scenarios.…”

Introduction to the special series, “The future of marine environmental monitoring and assessment”

“…The second paper (Huber et al, 2022), continues along the lines suggested by Lønborg et al (2022). It describes an SAV monitoring platform combining Sentinel-2 satellite imagery and ML techniques for advanced mapping of SAV in Swedish nationwide marine waters.…”

“…Submerged aquatic vegetation is in many locations sampled by video transects with kilometers between the transects, but combining, for example, video transect data with satellite data provides a method for extrapolating and providing spatial SAV coverage. Here, Huber et al (2022) describe the feasibility and how new monitoring technologies and models (ML) can be applied in a cloud solution with the web interface, allowing nonmodelers to classify SAV based on, for example, traditionally collected diver and/or video transect data. The cloud solutions then use the classification to continuously train the ML model and eventually optimize the precision of the monitoring platform.…”