It has been a busy few months at University with exams and deadlines, including my third-year (BSc) independent research project, hence my lack of blogging. This latest post will discuss this project, which explored the use of Autonomous Underwater Vehicle (AUV) habitat mapping as an approach for detecting change in the North Sea benthos, in relation to carbon capture and storage.
Introduction to Carbon Capture and Storage
Carbon capture and storage (CCS) is the process of separating carbon dioxide (CO2) from industrial or energy-related sources, and transporting it to a secure storage location for long-term isolation from the atmosphere (Metz et al., 2005). It has been presented as a viable solution for the mitigation of greenhouse gas emission and has the potential to reduce future world emissions from energy by 20% (Haszeldine, 2009). Thus, CCS could provide an alternative to a reduction in consumption of fossil fuels, which would ultimately cause a radical disruption to contemporary life (Chadwick et al., 2004; Monastersky, 2013). For more information, please see the Global CCS Institute.
Despite this, little is known about the environmental implications associated with CCS (Noble et al., 2012); thus, the aim of my research project was to assess the use of benthic imagery as a biological monitoring method for evidence of potential CO2 seepage impact at the Statoil Sleipner CCS field in the North Sea.
Data Acquisition and Analysis
Benthic images generated by an AUV (Autosub 6000 mission 66, deployed during the 77th voyage of RRS James Cook) were analysed for variability in shell cover. Recent studies have demonstrated that high concentrations of CO2 elicit a stress-induced surfacing response in benthic infauna, such as echinoderms and molluscs (Schade et al., 2016). Therefore for the purposes of this project, shell debris was considered an indicator of infaunal stress to assist in the location of CCS reservoir seepage. Benthic image analysis was automated via Matlab (R2016b), and a contour plot map of shell cover was produced, using coloured pixels as a proxy. Automated image analysis can be an effective tool for identifying change in the environment (Schoening et al., 2016), since it is a non-invasive technique that removes the concern of time-consuming manual quantification and thus, human error (Schoening et al., 2012).
Figure 1: Examples of the benthic images taken by Autosub6000 in the North Sea. Figure 1a shows an area of high shell debris, and Figure 1b shows an area of low shell debris.
However, the use of coloured pixel as a proxy for shell cover was discovered to be inconsistent and unreliable. Comparison of the coloured pixel results produced in Matlab and visual analysis of shell debris (Figure 1) demonstrated that there were large inconsistencies in the method, with a number of false-positive results. For example, the coloured pixel results indicated that Figure 1a was an area of low shell debris, and that Figure 1b was an area of high shell debris, despite visual analysis indicating otherwise. The produced contour plot map was therefore likely to have incorrectly portrayed the shell cover of the benthos at the survey site.
Similar results have been observed in other studies aiming to simplify the time-consuming task of manual image analysis through autonomous methods. Kannappan et al. (2015) concluded that although automated analysis provides a more attractive option to manual processing, available automated detection of organism populations (scallops in this case) do not work well with noisy low-resolution images since they are likely to produce a significant number of false-positive results.
An alternative proxy discussed for the detection of shell cover variation was the quantification of seabed roughness through the use of a high resolution sidescan sonar-equipped AUV, where high roughness would be expected to indicate high shell cover (Jaramillo and Pawlak, 2011). This method may have proved more effective in identifying Figure 1a as an area of high shell cover, and Figure 1b as an area of low shell cover, and therefore should be investigated further during future studies.
The Efficiency of Biological Monitoring
Biological monitoring of CCS sites through benthic imagery does have the potential to be effective (Noble et al., 2012), and is necessary since public concern is largely focussed on the potential environmental impacts of CCS (Widdicombe, 2015). Despite this, biological indicators are far more difficult to quantify than chemical or physical ones. It is widely agreed that exposure to acidified seawater or sediment environments (due to an increase in CO2 concentrations), significantly alters the macrofaunal community structure, partially through surfacing behaviour of infaunal species (Thistle et al., 2005; Schade et al., 2016). However, variability in inter- and intra-species tolerances to CO2 means that the likelihood of species loss is determined by both phenology and ecology (Widdicombe, 2015), thus adding an extra dimension to the monitoring. Also, whilst surfacing behaviour is widely considered to be an indicator of infaunal stress, this is not necessarily limited to high sedimentary CO2 levels indicative of a CCS leakage (ECO2, 2015).
The best candidates for biological indicators of CCS leakage are those that can be mapped on a large scale by benthic imagery, for example, increased shell cover (as discussed in this project) or microbial mats on the sediment surface. Areas with a high indication of leakage could then be investigated further to determine the cause of this environmental variability, though care should be taken with the use of these indicators of CCS leakage since they are transient signals. However, it is also argued that biological responses are more ideally suited to monitoring the progress of a leakage for ecosystem recovery once it has been detected, rather than the locating of potential leak sites (Widdicombe, 2015).
The project concluded that a multidisciplinary approach to CCS monitoring, integrating biological, chemical and physical analyses, is likely to be the most effective method, and it is recommended that this is explored further in future studies (Hicks et al., 2015). Nonetheless, it is essential for future research directives to address determining a biological baseline for these indicators, to quantify the natural fluctuations so that unnatural variability can be differentiated and detected.
With increasing pressures on both oil and gas companies and governmental bodies for a reduction in fossil fuel emissions, the need for CCS development, and thus monitoring, is greater than ever. Biological monitoring through assessment of surficial shell cover provides a reasonable solution for the monitoring of CCS leakage sites. Nonetheless, caution is required with the use of biological indicators, AUV imaging technology, and the use of proxies for indicator parameterisation during automated image analysis. Discrepancies in the use of pixel colour as a proxy for shell cover and AUV survey programming do not, however, suggest that the use of benthic imagery is an ineffective method of monitoring potential CCS sites for evidence of CO2 seepage impact. These method limitations and their associated problems could be mitigated with future development and considerations.