5 Discussion
5.1 Passive Seismic for Subsurface Layer Mapping
The MI surface model was shown to accurately represent the internal surface variations of the MI body and OT, demonstrating the efficacy of the passive seismic device and analysis method presented. The layer model also revealed variability not detectable from headwall exposures, their extrapolations or surface features, such as the rapid reduction in ice surface elevation and the bowl-like depression in the western section. Using the 2018 headwall exposures, a maximum difference of 1.4 m was found between the modelled and observed MI surface. When comparing the headwall exposures closest to the passive seismic measurement points, where the ice surface is not modelled below the headwall base, the layer boundaries were accurate to within 0.5 m. These validations demonstrate that the technique can be used to accurately detect subsurface layering depth and suggest that a greater density of measurements may be required to accurately capture the fine scale variations. Furthermore, knowledge of the MI variability and overburden thickness have been noted as critical to understanding the geomorphic response of Arctic landscapes to climate change and other anthropogenic disturbances (Pollard, 1990; Segal et al., 2016). On Banks Island, observations show that there has been a 60-fold increase in RTS numbers between 1984 and 2015 (Lewkowicz and Way, 2019). It has been suggested that a thin overburden layer, as little as 1 m in places (Lakeman & England, 2012), is a significant contributor to the observed landscape sensitivity on Banks Island. This thin overburden exacerbates the relatively slow warming trend by allowing the surface energy to more easily reach the near surface MI, thus triggering the formation of new RTSs (Rudy et al., 2017; Segal et al., 2016). This highlights the need for locally detailed maps of overburden and MI thickness variability in vulnerable areas, to aid in determining the likelihood of rapid geomorphic disturbances under continued warming in the near future. Finally, more accurate models of the relative proportions of ice and OT can allow for improved estimates of future soil losses, and thus better constrained predictions of carbon, nutrient and pollutant fluxes. They also have important implications for determining future levels of thaw subsidence, a significant contributor towards the cost of building and infrastructure maintenance across permafrost terrain (Clement, 2013; Couture et al., 2018; Couture and Pollard, 2017; Jones et al., 2008).
5.2 Factors Governing Short-Term HWR Rates
Severalprevious studies have focused on thaw indices and simple geometric models in an attempt to understand the drivers and controls on rates of HWR, but the results have been spatially and temporally inconsistent, with several authors citing a lack of locally accurate data on MI and OT as a significant constraint (Heginbottom, 1984; Jones et al., 2019; Lewkowicz, 1987a; Robinson, 2000, Zwieback et al., 2018). Here, we identify two dominant in-situ controls upon HWR, firstly a persistent exposure of MI inland of the current headwall position, regardless of its thickness, and secondly an average OT of under 4 m. Along transects where these criteria were met HWR rates were more than twice as fast as otherwise. It appears that on Peninsula Point the absolute thickness and the proportion of the headwall that consists of ice play little role in determining the HWR rates. Rather, the presence of ice allows for the initiation of HWR, with the resulting rate of retreat being largely dictated by the OT.
By using detailed observations of exposed headwall properties, it has been possible to predict the rate of HWR more accurately than simply extrapolating based on recent or historical averages. The accuracy of these predictions are further improved by incorporating the mapping of MI and OT inland of the headwall, with a reduction in the RMSE of 58% compared to predictions based on historical averages, and 36% compared to predictions based on the HWR rates recorded in the previous year. This is especially important in areas where HWR has not yet been initiated, but where the criteria necessary for HWR to occur can now be identified using passive seismic mapping. For example, in the western portion of Peninsula Point, 16 transects contained no visible exposures of MI in 2016 (Figure 6a). However, 11 of those transects experienced HWR > 10 m between 2016 and 2017, four of which exceeded 20 m. Had passive seismic monitoring been carried out in this region in 2016, it may have been possible to identify the region as having the requisite layering (MI and thin OT) to produce rapid rates of HWR once the MI was exposed. This provides the potential to better predict and therefore mitigate the dramatic geomorphic changes that occur in ice-cored terrain and highlights the need to detect and map MI variability, not only to improve forecasts of HWR rates, but to identify areas where new RTSs are likely to develop.
6 Conclusion
Passive seismic monitoring has been shown to be an effective tool for detecting OT and MI surface variability. The resulting models showed variations in MI not apparent from headwall exposures or surface topographic features, and the opposite of those inferred from simple extrapolations of headwall exposures.
OT and IT were found to exert a significant control over short term HWR rates from 2016 to 2018. Where the OT remained below 4 m and MI was present inland of the headwalls, HWR rates were more than twice as fast the headwalls with differing properties.
By conducting detailed surveys of headwall properties, it is possible to predict HWR rates more accurately than extrapolating based on past rates alone. Furthermore, by mapping and incorporating OT and MI variations inland of the current headwall position, the errors in HWR predictions can be more than halved compared to predictions based on historical averages, and reduced by more than a third compared to extrapolations from the previous years HWR rates
This research brings into focus the need for accurate data on MI and OT in order to understand the spatial and temporal variability in RTS activity. Better knowledge of MI variability can contribute to improved forecasts of coastal change, such as rates of shoreline retreat and volume loss. These findings have the potential to provide more accurate estimates of nearshore carbon and sediment fluxes and to improve assessments of the susceptibility of local landscapes to rapid geomorphic changes. We emphasise the need for widespread testing of these headwall metrics and passive seismic monitoring ensure their efficacy and robustness across different ice-cored permafrost terrains.