EGU2020-5057
https://doi.org/10.5194/egusphere-egu2020-5057
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Satellite remote sensing of ice cliff migration

Bas Altena and Andreas Kääb
Bas Altena and Andreas Kääb
  • University of Oslo, Geosciences, Oslo, Norway (bas.altena@geo.uio.no)

Ablation patterns on debris-covered glaciers are highly complex and spatially variable, while accessibility is complicated due to steep topography and loose surface debris material. One of the main ablation components on debris-covered glaciers is ice melt on steep ice cliffs and associated cliff migration. When using measurement techniques that operate in absolute coordinates, a main challenge is to separate cliff retreat from the underlying ice movement. In-situ measurements are spatially limited, while giving highly detailed understanding of processes occurring on individual ice cliffs. Drones can extent such detailed measurements to a whole glacier tongue, but are still limited to a few glaciers and measurement times. Here we show how measurements of cliff migration rates towards a regional scale are possible with spaceborne optical instruments. For this study we focus on the Mt. Everest region, specifically the Khumbu Glacier and other glaciers in the surrounding. We use Venμs, a French-Israeli multi-spectral satellite, that provides images at high temporal resolution (a two day repeat), and at high spatial resolution (5m), at this spatial resolution it provides sufficient detail to investigate individual ice cliffs.

Migration of ice cliffs can have a dominant direction, but their shape evolves over time, complicating pattern matching. Similar challenges occur for velocity extraction of the underlying glacier ice, where the shadow casted by ice cliffs is a dominant feature on glacier imagery, thus instead of debris patterns, the velocity estimates have ice cliff migration patterns within. Hence, in order to reduce the interference between both processes we reduce the influence of shadow within the imagery and extract bulk glacier ice velocity. While specific ice cliff features are isolated and tracked. Thus different image tracking techniques are deployed, in order to distinguish one displacement from the other.

The ice-cliff migration can be separated from the general glacier velocity, which results in a regional estimate of ice cliff back wasting, and thus a proxy for clean ice mass-balance of debris-covered glaciers from space. Venμs is a demonstrator satellite, with a limited lifetime and acquisition strategy, but our automatic methodology is generic and can be transferred to, for example, the 10m imagery from Sentinel-2, making regional analysis feasible.

How to cite: Altena, B. and Kääb, A.: Satellite remote sensing of ice cliff migration, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-5057, https://doi.org/10.5194/egusphere-egu2020-5057, 2020

Comments on the presentation

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

Presentation version 1 – uploaded on 05 May 2020
  • CC1: Comment on EGU2020-5057, Amaury Dehecq, 06 May 2020

    Just copying my questions from the chat.

    First, congratulations for your fantastic display! I have a few questions for you:

    1. What is the (vertical) accuracy of your photogrammetric results with Venus?

    2. What is the magnitude of the ice cliffs velocity you estimate (your figures don't have a color scale...)

    3. Icecliff flow is a sum of two processes (ice flow + backwasting). Do you mask the ice cliffs when estimating the general ice flow? Does it improve the correlation?

     

    • AC1: Reply to CC1, Bas Altena, 06 May 2020

      Dear Amaury, thank you very much! Concerning your questions:

      1. What is the (vertical) accuracy of your photogrammetric results with Venus?

      The angle between the backward and forward looking pushbroom is very small (in the order of 3 degrees), this results in a Base-to-Height of 0.025. It is possible to generate DEM's from the raw imagery (see for example this work: https://figshare.com/articles/DEM_generation_from_native_stereo_Ven_s_acquisitions/11806989/1). But here we use the orthorectified imagery, and exploit the fact that high frequency terrain (like icecliffs) are not present in the DEM that is used for orthorectification. So the paralax signal is used, but not in an absolute manner, but just as a check/validation.

      2. What is the magnitude of the ice cliffs velocity you estimate (your figures don't have a color scale...)

      The colorbar goes from 0 to 5 meters, for the icecliffs, while the other figure with ice/debris speed is in the range of 0 to 100 meters.

      3. Icecliff flow is a sum of two processes (ice flow + backwasting). Do you mask the ice cliffs when estimating the general ice flow? Does it improve the correlation?

      We estimate two separate velocity fields, from two different images (and methods). So for the debris/ice velocity, we try to compress the illumination component as much as possible, to get only reflectivity/albedo. This is also why it does not seem to perform correctly in regions with blank snow or ice.

      • CC2: Reply to AC1, Amaury Dehecq, 06 May 2020

        Thanks for the reply. And thanks for the link, super interesting!