EGU2020-5561, updated on 11 Jan 2021
https://doi.org/10.5194/egusphere-egu2020-5561
EGU General Assembly 2020
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Comparison Between Surface Melt Days Estimation from a Regional Climate Model and Near-Daily Synthetic Aperture Radar Backscattering

Quentin Glaude1 and Christoph Kittel2
Quentin Glaude and Christoph Kittel
  • 1Laboratoire de Glaciologie, Université Libre de Bruxelles, Brussels, Belgium (qglaude@ulb.ac.be)
  • 2Laboratoire de Climatologie, Université de Liège, Liège, Belgium

 

Remote sensing has long been used as a powerful tool for the observation in cryospheric sciences. With the advances brought by the ESA Copernicus program, Earth observation goes a step further in its ability to get acquisitions at very high temporal rate. This is even amplified in polar regions due to heliosynchronism of satellites’ orbits. Earth observation shifts from sporadic observations to Earth monitoring.

Observations are a critical aspect for the assessment of geophysical models. The ability of a model to replicate observations is crucial as a benchmark. It also allows to refine our comprehension of Earth systems, such as in cryospheric sciences.

In this work, we are using the regional climate model MAR to compute the surface melt on a domain focusing on the Roi Baudouin Ice Shelf, Queen Maud Land, East Antarctica. From the results, we extract the number of days with surface melt in a region. In parallel, we employ remote sensing to obtain comparison data. Synthetic aperture radar appears as a solution of choice thanks to its day-and-night (critical in polar regions) and atmospheric-free capabilities. Radar backscattering anomalies between different dates are witnesses of substantial increase of soil moisture. Using Sentinel-1 in its wide-swath modes (namely Interferometric Wide Swath and Extra Wide Swath modes) and multiple satellite paths, near-daily acquisitions can be obtained. By comparing the two independent results, we better constraint model’s outputs while also better interpret SAR acquisitions. 

How to cite: Glaude, Q. and Kittel, C.: Comparison Between Surface Melt Days Estimation from a Regional Climate Model and Near-Daily Synthetic Aperture Radar Backscattering, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5561, https://doi.org/10.5194/egusphere-egu2020-5561, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 01 May 2020
  • CC1: Comment on EGU2020-5561, William D. Harcourt, 05 May 2020

    Thanks for the interesting display, its nice to see the MAR output and radar data agree to some degree. Have you considered any methods for classifying this data using the MAR model into different surface facies? 

    Alos, are these calibrated RCS values or simply the output from a SAR processing?

    Thanks,

    Will

    • AC1: Reply to CC1, Quentin Glaude, 05 May 2020

      Thank you for your comment.

      By different surface facies, I assume you talk about snow, blue ice, rocks, and so forth. This is the the objective of a future work, and especially the understanding on where MAR and SAR observations differ. This is crucial to improve the model.

      The values from SAR include radiometric corrections. This correction is performed using calibration vector from Sentinel-1 metadata. Nevertheless, the use of differents orbits and incidence angles brings a bias related to different volume scattering mechanisms. In the time series, it introduces a small oscillatory pattern between the different orbits.

      Best regards,

      Quentin Glaude

      • CC3: Reply to AC1, William D. Harcourt, 06 May 2020

        Thank-you for your reply, I look forward to seeing the results of the future work!

  • CC2: Comment on EGU2020-5561, Olaf Eisen, 05 May 2020

    Hi Quentin,

    regarding my question during the chat, a bit more specific now: could there be cases where MAR predicts no surface melt but SAR still detects liquid water not at the surface, but a bit down, e.g. residuals from previous warmer days?

    Thanks, Olaf

    • AC2: Reply to CC2, Quentin Glaude, 05 May 2020

      Hi Olaf,

      Yes, totally ! This is the reason why we did not only compute melt from MAR but also water content as well as other variables (rainfall, runofff, snowfall, surface temperature, and so forth). We observed cases where water content is present without melt nor rainfall, but in the same time radar backscaterring is very low (which confirms the presence of water).

      Best regards,

      Quentin

      • AC3: Reply to AC2, Christoph Kittel, 05 May 2020

        Dear Olaf,

        in addition to what was said by Quentin, this phenomenon can be observed in our time series. If we compare the summer of 2017 (slide12), we can see that MAR simulates a lot of melting at the beginning of the season, less afterwards, while SAR keeps a strongly affected signal until the end of the season. 
        For a better comparison, we have also used the liquid water content in MAR (slide14), and the agrement seems better because some of the melted water remains in the 1st meter of snowpack. 

        This double comparison SAR/melt and SAR/water content could help explain where the water seen by the SAR comes from. Such cases could also help to assess the model's ability to correctly simulate the refreezing or the percolation of water in the mantle, both processes reducing the liquid water content.

        Best
        Chris (co-author)

        • CC4: Reply to AC3, William D. Harcourt, 06 May 2020

          I think this ties into my question above a little - it would be very interesting to see whether zones on the ice shelf can be determined from the SAR image i.e. refrozen meltwater, snow, bare ice etc. Might be difficult, although there was some really interesting discussions in the Machine Learning session (CR2.4) about this.