Karst network geometry and topology: automatic classification of network geometries.

Philippe Renard and David Bernasconi and Andrea Borghi and Pauline Collon and Cecile Vuilleumier. ( 2013 )
in: 15th IAMG Conference, 2‐5 September 2013, Madrid, Spain

Abstract

Over the last years, different techniques have been proposed to model the geometry of karstic conduit networks. Some are based on purely statistical techniques, other are based on more process based methods. One of the key questions arising then is whether the resulting models are representing properly the main features of real karstic systems. In this paper, we propose to use a set of statistical indicators to characterize the geometry and topology of karstic networks. These indicators are computed on a relatively large set of real karst networks. We then use these measures as an objective basis for the classification of other karst networks into homogeneous classes. The classification obtained is then compared with the widely-used classification of Palmer. Furthermore, we extent our method to synthetic karstic networks. We use a supervised learning algorithm to automatically classify any new simulated network based on its statistical properties. We show that the method is able to estimate objectively how distant the simulated network is from the real karst networks. This distance can therefore be used as a quality measure of the simulation and may indicate the direction to take in order to improve the future simulations.

Download / Links

    BibTeX Reference

    @INPROCEEDINGS{,
        author = { Renard, Philippe and Bernasconi, David and Borghi, Andrea and Collon, Pauline and Vuilleumier, Cecile },
         title = { Karst network geometry and topology: automatic classification of network geometries. },
         month = { "sep" },
     booktitle = { 15th IAMG Conference, 2‐5 September 2013, Madrid, Spain },
          year = { 2013 },
      abstract = { Over the last years, different techniques have been proposed to model the geometry of karstic conduit networks. Some are based on purely statistical techniques, other are based on more process based methods. One of the key questions arising then is whether the resulting models are representing properly the main features of real karstic systems.
    In this paper, we propose to use a set of statistical indicators to characterize the geometry and topology of karstic networks. These indicators are computed on a relatively large set of real karst networks. We then use these measures as an objective basis for the classification of other karst networks into homogeneous classes. The classification obtained is then compared with the widely-used classification of Palmer. 
    Furthermore, we extent our method to synthetic karstic networks. We use a supervised learning algorithm to automatically classify any new simulated network based on its statistical properties. We show that the method is able to estimate objectively how distant the simulated network is from the real karst networks. This distance can therefore be used as a quality measure of the simulation and may indicate the direction to take in order to improve the future simulations. }
    }