A few years ago I had the pleasure of visiting with the staff of the Centre for Environment, Fisheries and Aquaculture Science (Cefas) in Lowestoft, U.K. During my visit we discussed best practices for analyzing various acoustic data sets with bathymetry for seafloor habitat mapping within Trimble eCognition.
I recently followed up with Alexander Callaway from Cefas and he informed me that “eCognition remains a constant in our mapping lives”. As such, Tyllianakis et al. recently published a paper entitled “The value of information: Realising the economic benefits of mapping seagrass meadows in the British Virgin Islands” in the journal Science of the Total Environment where eCognition is used to support a study demonstrating “how improved knowledge on the distribution of habitats can improve the accuracy of economic valuation exercises to provide local decision makers with better management capability”.
The study focuses on mapping seagrass meadows in the Sir Frances Drake Channel south of the island of Tortola. Seagrass plays an important role in “blue carbon” sequestration and storage – “blue carbon refers to carbon in coastal ecosystems”. In fact, the authors state that seagrass meadows, alone, account for “capturing 10% of the global organic carbon emissions per year”. Alarmingly, we have seen a significant decrease in the extent of such habitats – “seagrass extent in the world’s oceans has declined by approximately 29% since the beginning of the 20th century”.
Unfortunately, our knowledge of resources in marine environments is limited. To-date, “<10% of our ocean floors have been mapped in detail”. According to the authors, “coupling information on the extent of the seagrasses with an economic estimate of the organic carbon storage potential of the mapped seagrasses against the cost of detailed survey assesses the impact of such activities over a predetermined time period”.
The state of modern survey technologies has improved tremendously and, as Tyllianakis et al. point out, “enables the acquisition of data that can be used to more effectively quantify resource distribution”. Looking back on the past few years of our own eCognition software development, this is certainly true and it is wonderful to see our software used to realize such projects.
The authors used multibeam echosounder (MBES) bathymetry and backscatter data acquired in 2014 by the United Kingdom Hydrographics Office (UKHO) as input for eCognition for the generation of a seabed classification map. In total, 6 spatially continuous data layers were used: bathymetry (depth), backscatter (roughness), three different Bathymetric Position Indices (BPI) and slope.
A multiresolution segmentation approach was used to generate small image objects and form the building blocks for the subsequent analysis. These objects were then passed through a multiple object difference conditions-based fusion routine to “merge objects that had similar characteristic values across the layers of bathymetry, backscatter and BPI 5”. According to the authors, this combined segmentation approach worked well to differentiate between areas of sediment and reefs.
After establishing initial image objects, a random forest classification was applied, “an iterative process whereby multiple classification trees are created and the most statistically appropriate result is created from an aggregation of those runs”.
The final map was is compatible with the existing Coastal Resource Atlas established by Gardner et al., 2008 allowing for the authors to demonstrate the increase of information for known seagrass resources. 11 seabed classes were mapped across the study area and the results provided “finer, more detailed classifications than the previous Coastal resource atlas”.
The mapping results were used as input for a cost-benefit analysis (CBA) where the carbon sequestration and storage capacity of the seagrass habitats were calculated. The authors conclude that the study shows “following relatively modest investment, the projected benefits of seabed mapping is substantial”. The net present value of the carbon sequestration and storage will increase from £49,428 to £3,192,318 after 25 years and to £4,234,537 after 50 years.
It is great to see how information won through eCognition analysis tools flows into a project and it is particularly interesting to see, as the title of the paper states, “the value of information”.
The authors of the paper would particularly like to thank the Darwin Initiative for providing funding; the United Kingdom Hydrographic Office for data collection and partnership; the National Parks Trust of the Virgin Islands or data collection and partnership; and Marine-Bio Images for supplying the camera system and imagery.