Day 4

Detailed paper information

Back to list

Paper title Mapping the grounding line of Antarctica in SAR interferograms with machine learning techniques
Authors
  1. Sindhu Ramanath Tarekere Speaker
  2. Lukas Krieger German Aerospace Center (DLR), Remote Sensing Technology Institute
  3. Konrad Heidler German Aerospace Center (DLR)
  4. Dana Floricioiu German Aerospace Center (DLR), Remote Sensing Technology Institute
Form of presentation Poster
Topics
  • A9. Polar Science and Cryosphere
    • A9.04 Mass Balance of the Cryosphere
Abstract text The grounding line marks the transition between ice grounded at the bedrock and the floating ice shelf. Its location is required for estimating ice sheet mass balance [Rignot & Thomas, 2002], modelling of ice sheet dynamics and glaciers [Schoof 2007], [Vieli & Payne, 2005] and evaluating ice shelf stability [Thomas et al., 2004], which merits its long-term monitoring. The line migrates both due to short term influences such as ocean tides and atmospheric pressure, and long-term effects such as changes of ice thickness, slope of bedrock and variations in sea level [Adhikari et al., 2014].

The grounding line is one of four parameters characterizing the Antarctic Ice Sheet (AIS) ECV project within ESA’s Climate Change Initiative (CCI) programme. The grounding line location (GLL) geophysical product was designed within AIS_CCI and has been derived through the double difference InSAR technique from ERS-1/2 SAR, TerraSAR-X and Sentinel-1 data over major ice streams and outlet glaciers around Antarctica. In the current stage of the CCI project, we have interferometrically processed dense time series throughout the year from the Sentinel-1 A/B constellation aiming at monitoring the short-term migration of the DInSAR fringe belt with respect to different tidal and atmospheric conditions. Whereas the processing chain runs automatically from data download to interferogram generation, the grounding line is manually digitized on the double difference interferograms. Inconsistencies are introduced due to varying interpretation among operators and the task becomes more challenging when using low coherence interferograms. On a large scale this final stage of processing is time consuming, hence urging the need for automation.

An attempt in this direction was made in the study of [Mohajerani et al., 2021], where a fully convolutional neural network (FCN) was used to delineate grounding lines on Sentinel-1 interferograms. In a similar vein, the performance of deep learning paradigms for glacier calving front detection [Cheng et al., 2021], [Baumhoer et al., 2019], showcase the strengths of using machine learning for such tasks. However, unlike grounding lines, calving fronts are visible both in optical and SAR imagery. This makes available a greater amount of training data. The visibility of the calving front also enables the use of classical image processing techniques [Krieger & Floricioiu, 2017]. Additionally, the complexity of InSAR processing and wrapped phases is absent.

This study further investigates the feasibility of automating the grounding line digitization process using machine learning. The training data consists of double difference interferograms and corresponding manually delineated AIS_CCI GLL’s derived from SAR acquisitions between 1996 - 2020 over Antarctica. In addition to these, features such as ice velocity, elevation information, tidal displacement, noise estimates from phase and atmospheric pressure are analyzed as potential inputs to the machine learning network. The delineation is modelled both as a semantic segmentation problem, as well as a boundary detection problem, exploring popular existing architectures such as U-Net [Ronneberger et al., 2015], SegNet [Badrinarayanan et al., 2017] and Holistically-nested Edge Detection [Xie & Tu, 2015]. The resulting grounding line predictions will be examined with respect to their usability in the detection of short-term variations of the grounding line as well as the potential separation of a signal of long-term migration. The detection accuracy will be compared to the one achieved by human interpreters.



References

Adhikari, S., Ivins, E. R., Larour, E., Seroussi, H., Morlighem, M., and Nowicki, S. (2014). Future Antarctic bed topography and its implications for ice sheet dynamics, Solid Earth, 5, 569–584

Baumhoer, C. A., Dietz, A. J., Kneisel, C., & Kuenzer, C. (2019). Automated extraction of antarctic glacier and ice shelf fronts from sentinel-1 imagery using deep learning. Remote Sensing, 11(21), 2529

Badrinarayanan, V., Kendall, A., Cipolla, R., (2017). Segnet: A deep convolutional encoder-decoder architecture for scene segmentation. IEEE transactions on pattern analysis and machine intelligence.

Cheng, D., Hayes, W., Larour, E., Mohajerani, Y., Wood, M., Velicogna, I., & Rignot, E. (2021). Calving Front Machine (CALFIN): glacial termini dataset and automated deep learning extraction method for Greenland, 1972–2019. The Cryosphere, 15(3), 1663-1675

Krieger, L., & Floricioiu, D. (2017). Automatic calving front delienation on TerraSAR-X and Sentinel-1 SAR imagery. In 2017 IEEE International Geoscience and Remote Sensing Symposium (IGARSS)

Mohajerani, Y., Jeong, S., Scheuchl, B., Velicogna, I., Rignot, E., & Milillo, P. (2021). Automatic delineation of glacier grounding lines in differential interferometric synthetic-aperture radar data using deep learning. Scientific reports, 11(1), 1-10.

Rignot, E., & Thomas, R. H. (2002). Mass balance of polar ice sheets. Science, 297(5586), 1502-1506.

Ronneberger, O.; Fischer, P.; Brox, T. U-Net: Convolutional Networks for Biomedical Image Segmentation. In Medical Image Computing and Computer-Assisted Intervention—MICCAI 2015; Navab, N., Hornegger, J., Wells, W.M., Frangi, A.F., Eds.; Springer International Publishing: Cham, Switzerland, 2015; Volume 9351, pp. 234–241. ISBN 978-3-319-24573-7

Schoof, C. (2007). Ice sheet grounding line dynamics: Steady states, stability, and hysteresis, J. Geophys. Res., 112, F03S28, doi:10.1029/2006JF000664.

Xie, S., Tu, Z., 2015. Holistically-nested edge detection. In: Proceedings of the IEEE International Conference on Computer Vision. pp. 1395-1403

Thomas, R., Rignot, E., Casassa, G., Kanagaratnam, P., Acuña, C., Akins, Brecher, H., Frederick, E., Gogineni, P., Krabill, W., Manizade, S., Ramamoorthy, H., Rivera, A., Russell, R., Sonntag, J., Swift, R., Yungel, J., & Zwally, J., (2004). Accelerated sea-level rise from West Antarctica. Science, 306(5694), 255-258.

Vieli, A., & Payne, A. J. (2005). Assessing the ability of numerical ice sheet models to simulate grounding line migration, J. Geophys. Res., 110, F01003, doi:10.1029/2004JF000202