EGU2020-7489
https://doi.org/10.5194/egusphere-egu2020-7489
EGU General Assembly 2020
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Quantification of ground ice through petrophysical joint inversion of seismic and electrical data applied to alpine permafrost

Coline Mollaret1, Florian M. Wagner2, Christin Hilbich1, and Christian Hauck1
Coline Mollaret et al.
  • 1Department of Geosciences, University of Fribourg, Fribourg, Switzerland (coline.mollaret@unifr.ch)
  • 2Institute for Applied Geophysics and Geothermal Energy, RWTH Aachen University, Aachen, Germany

Quantification of ground ice is particularly crucial for understanding permafrost systems. The volumetric ice content is however rarely estimated in permafrost studies, as it is particularly difficult to retrieve. Geophysical methods have become more and more popular for permafrost investigations due to their capacity to distinguish between frozen and unfrozen regions and their complementarity to standard ground temperature data. Geophysical methods offer both a second (or third) spatial dimension and the possibility to gain insights on processes happening near the melting point (ground ice gain or loss at the melting point). Geophysical methods, however, may suffer from potential inversion imperfections and ambiguities (no unique solution). To reduce uncertainties and improve the interpretability, geophysical methods are standardly combined with ground truth data or other independent geophysical methods. We developed an approach of joint inversion to fully exploit the sensitivity of seismic and electrical methods to the phase change of water. We choose apparent resistivities and seismic travel times as input data of a petrophysical joint inversion to directly estimate the volumetric fractions of the pores (liquid water, ice and air) and the rock matrix. This approach was successfully validated with synthetic datasets (Wagner et al., 2019). This joint inversion scheme warrants physically-plausible solutions and provides a porosity estimation in addition to the ground ice estimation of interest. Different petrophysical models are applied to several alpine sites (ice-poor to ice-rich) and their advantages and limitations are discussed. The good correlation of the results with the available ground truth data (thaw depth and ice content data) demonstrates the high potential of the joint inversion approach for the typical landforms of alpine permafrost (Mollaret et al., 2020). The ice content is found to be 5 to 15 % at bedrock sites, 20 to 40 % at talus slopes, and up to 95 % at rock glaciers (in good agreement to the ground truth data from boreholes). Moreover, lateral variations of bedrock depth are correctly identified according to outcrops and borehole data (as the porosity is also an output of the petrophysical joint inversion). A time-lapse version of this petrophysical joint inversion may further reduce the uncertainties and will be beneficial for monitoring and modelling studies upon climate-induced degradation.

 

References:

Mollaret, C., Wagner, F. M. Hilbich, C., Scapozza, C., and Hauck, C. Petrophysical joint inversion of electrical resistivity and refraction seismic applied to alpine permafrost to image subsurface ice, water, air, and rock contents. Frontiers in Earth Science, 2020, submitted.

Wagner, F. M., Mollaret, C., Günther, T., Kemna, A., and Hauck, C. Quantitative imaging of water, ice, and air in permafrost systems through petrophysical joint inversion of seismic refraction and electrical resistivity data. Geophysical Journal International, 219 (3):1866–1875, 2019. doi:10.1093/gji/ggz402.

How to cite: Mollaret, C., Wagner, F. M., Hilbich, C., and Hauck, C.: Quantification of ground ice through petrophysical joint inversion of seismic and electrical data applied to alpine permafrost, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-7489, https://doi.org/10.5194/egusphere-egu2020-7489, 2020.

Display materials

Display file

Comments on the display material

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

Display material version 1 – uploaded on 04 May 2020
  • CC1: Comment on EGU2020-7489, Siobhan Killingbeck, 05 May 2020

    Hi Coline, 

    I was way too slow on the online chat to ask my question I must type faster!! I have a couple of questions about your joint inversion, which I find very interesting!

    How do you estimate/calculate porosity? And do you think this approach could be used subglaically to characterise water contents under ice, say if we have Vp and Res data? 

    Also have you thouhgt about using the surface waves to estimate Vs, then you can derive VpVs ratio and estimate porosity using the biot gassmann theory, but i supose if your matrix is a mixture of ice/rock it might be difficult. 

    Thanks

    Siobhan

    • AC1: Reply to Siobhan Killingbeck (CC1), Coline Mollaret, 05 May 2020

      Dear Siobhan,

      Thanks a lot for posting your question here! The chat was indeed really (too) quick. I repeat below your questions to try to make my answerd more understandable.

      • How do you estimate/calculate porosity?

      In the conventional 4PM (Hauck et al. 2011), the porosity had to be prescribed/given.
      The joint inversion approach now invert for the four phases (i.e. also invert for porosity, exactly the same way it estimates the ice, air and water content).
      We still give an initial porosity value, which has a non negligible influence when we use not very constraining petrophysical model (e.g. Archie’s law, which constrains only the water content and the porosity). In these cases, the porosity initial model has to be quite well guessed (otherwise the results are not reliable).
      But when using more constraining petrophysical models (such as the time-average equation and the resistivity geometric mean, both constraining the four phases), the initial porosity value does not influence the results (or at least much less).

      • And do you think this approach could be used subglaically to characterise water contents under ice, say if we have Vp and Res data? 

      Yes, it could be used subglacially using Vp and Res data. Do you have such sites in mind where Vp and Res were measured subglacially? That would be very interesting!

      • Also have you thouhgt about using the surface waves to estimate Vs, then you can derive VpVs ratio and estimate porosity using the biot gassmann theory, but i supose if your matrix is a mixture of ice/rock it might be difficult. 

      Yes I thought some years ago also using Vs measurements. But the aim of my (PhD) work focused to investigate inversion techniques to improve permafrost characterisation in order to (re-)analyse existing monitoring data sets (which are “only” electrical resistivity and Vp). This is one of the main reasons why not to use surface waves / Vs (because it was not historically measured). But your suggestion would definitely be worth to investigate, e.g. to add more constraint on the porosity estimation.

      Thanks

      Coline

      • CC2: Reply to AC1, Siobhan Killingbeck, 05 May 2020

        Thanks Coline for your detailed reply!

        So in your method you solve for porosity also, great, I am just reading through your 2020 frontiers paper now.

        We did look into using your approach with this data https://www.solid-earth.net/11/75/2020/ but we were mainly focused on combining surface waves (Vs) and resistivity and didn’t look into the P waves (yet). Also couldn’t think of a petrophysical way to directly combine Vs and res… yet, so it’s been parked for now. However, we are very interested in developing a joint inversion for investigating subglacial water content (under thicker ice). We plan on having Vp (from seismic velocities of a reflection profile (maybe not as accurate as a refraction survey but we are imaging much deeper)) and Res (from TEM data). If we incorporated your method into our analysis we should be able to output estimates of porosity, ice content, water content and rock content??  Can you see any problems using a thick ice scenario? And what if we had a subglacial lake where porosity and water content was near 100% do you think it could pick up on such a big change?

        (Sorry for all the questions!)

        Finally, with regards to deriving porosity from VpVs ratio I have an example here (https://meetingorganizer.copernicus.org/EGU2020/EGU2020-8820.html) just for your reference.

        Thanks!

        Siobhan

        • AC2: Reply to CC2, Christin Hilbich, 05 May 2020

          Dear Siobhan,

          may I add a small comment to yor conversation?

          The question regarding S-waves is a frequent question, but apparently in our data from mountain permafrost terrain, for some reason S-waves seem to be mostly absent or hard to identify. Therefore we didn't really follow up on this topic unti today...

          And another question: you mention subglacial lakes - I wonder if a water body below thick massive is detectable at all with P-waves? Is that possible? I would expect the lake forming a hidden layer because of velocity inversion...

          • CC3: Reply to AC2, Siobhan Killingbeck, 05 May 2020

            Hi Christin!

            Thanks for joining, yes your right surface wave dispersion is really difficult to pick on mountain permafrost sites mostly because the subsurface is so variable and the assumption of a homogeneous subsurface under the geophone spread cannot be met. I found the uncertainty associated to my S waves on frozen and unfrozen ground at the foreland of Midtdalsbreen really high! So maybe it is too ambiguous! But at least you have Vp and resistivity to further constrain them if you did ever mange to pick decent dispersion patterns.  

            With the subglacial sediments/lake under thicker ice, we would not be using refraction (where that would be a problem!) We would use seismic velocities from a reflection study, here we can pick them from semblance plots (during data processing) or/and look at the AVA (amplitude verse angle) response of the ice-sediment (or lake) interface and get an estimate of Vp for that layer and then sediment(lake) – bedrock interface and get an estimate of Vp for that layer. So we are hoping to have an estimate of Vp and Resistivity and combine them to get something useful out like porosity and/or water content?  

            Hope that makes sense!

            • AC3: Reply to CC3, Christin Hilbich, 05 May 2020

              Yes, thanks for clarifying. :-) Seems definitely worth to try!