Geotechnical centrifuge modelling of crevassing in glaciers and ice sheets

Omitted from the 2007 IPCC Fourth Assessment Report on Climate Change was the potential contribution from ice sheets to global sea level. This reflected the level of uncertainty with respect to the ice dynamics (motion) and mass balance (snow and ice accumulation vs. snow and ice loss) of the extant ice sheets in Greenland and Antarctica. One potential key control on ice dynamics is glacier crevassing which can facilitate the routing of surface melt water to the ice sheet bed leading to increased sliding velocities on outlet glaciers. Additionally, crevassing controls the production of icebergs at marine terminating margins, through which the Greenland Ice Sheet disposes of ~50% and the Antarctic Ice Sheet almost all of their respective annual ice loss. Iceberg production (calving) may be through a combination of both bottom-up and top-down crevassing but atmospheric warming, by increasing the availability of melt water to fill surface crevasses, is likely to be the main driver of change, in the short term at least. Only recently have advances been made in the development of physics-based crevassing/calving relationships with incorporation into predictive numerical models. These advances are vital for improving our predictions for the response of the big ice sheets to future warming. However, only one study to date has tested these physics-based crevassing relationships and then only for shallow water-free crevasses. Given the current research focus on glacier crevassing, there is an urgent need to test crevassing models. To do this in the field is however challenging, due to difficulties of working in crevasse zones of glaciers, measuring the depth of what ultimately ends in a hairline crack at depth and associating the crevasse with the instantaneous stress/strain field. Project Partner DB has a project in preparation to deploy instrumentation for continuous water level monitoring in crevasses on Kronebreen, Svalbard. Geophysical imaging is currently problematic for example signal attenuation on 'warm' temperate glaciers, signal interference from adjacent crevasses in crevasse fields and obtaining the resolution to image the crevasse (crack) tip. Likewise controlling water-depth to force crevasse penetration would present significant challenges for example, the volume of water needed for filling a crevasse or connection with the englacial drainage system leading to water loss etc. Field monitoring of glacier crevassing is thus in its infancy.

A modelling approach therefore represents an ideal way forward. However, lab-floor models are useless because the stresses relevant to crevasse propagation increase as a function of both the self-weight stress (gravity x ice density x ice thickness) and crack length i.e. the crevasse depth. The geotechnical centrifuge is a unique modelling tool which allows the magnitude self weight stresses to be reproduced, with stress equivalence between the prototype (real world) and the model by scaling down the dimensions in the model but 'enhancing' gravity. This is achieved by 'flying' (spinning) the model in the centrifuge such that an Nth scale model flown at N times gravity generates the same self-weight stress as the prototype. Scaling relationships are already established for all the parameters relevant to this study so no scaling issues are anticipated, but the standard modelling of models centrifuge technique will be employed to confirm this. Then a series of models will be run, replicating the stress levels experienced in a prototype glacier section ~50x80x50 m. Pre-cast crevasses will be filled with water to facilitate step-wise full-depth crevasse penetration allowing the current state of the art physics-based models to be tested. This project will provide a proof of concept which will facilitate further grant applications where more complex models (e.g. bottom-up and top-down) can be built and used to test and develop physical models.

The direct beneficiaries of this research are the glaciological and wider academic communities concerned with understanding climate change, through hindcast (calibrated through reconstructions of landforms, climate etc) and forecast modelling studies. While, as noted in the academic beneficiaries section, this work will contribute to large scale coupled climate models these links are not direct. Therefore, while it will ultimately contribute to research which will come under the notice of policy-makers at national and international levels it would be facetious to suggest a direct contribution.

The academic community will gain access to the results from this project through the normal academic routes: publications in peer review journals and presentations at relevant workshops and conferences. Direct connection with the numerical modelling community comes from the involvement of Project Partner DB (see LoS_1).

The general public undoubtedly has a level of interest in the world's large ice sheets and are fascinated (as are glaciologists) by the calving of large icebergs at marine and lacustrine outlet glaciers. Therefore this project has the opportunity for significant public engagement. However, as we are not doing any field-work per se, the scope is somewhat reduced, none-the-less some opportunities remain. The wider community includes schools (especially those teaching geology or geography at Higher, Advanced Higher and A-level) and the general public. Aberdeen and Dundee both run popular Café Scientifique programmes providing opportunities to enter into dialogue with the general public and where possible advantage will be made of these. Aberdeen's community cafe programme in particular, represents one of the largest initiatives in the UK, featuring eight parallel cafe programmes located in rural areas as well as the city. Aberdeen has an established science festival and open days to which we will offer engagement activities. Resources are limited but we plan to offer a poster and possibly electronic display and simple interactive hands-on experiments i.e. the principles of centrifuge modelling can be demonstrated using salad spinners.