Soft Computing in GeoScience

Many problems in geoscience exhibit fundamental difficulties, like:

  • high complexity of the underlying physical processes, and thus unfeasibly large computational demands of the corresponding models
  • large amounts of data
  • uncertainity and/or imprecise and sparse availability of measurement data
  • still insufficient understanding of some of the underlying physical processes
which make them intractable by conventional computational means. In order to tackle these problems we develop methods, which combine computational techniques from machine learning and optimization with the knowledge base of geoscience. In this way we want to provide suitable solutions with justifiable computational effort and to identify new, previously unconsidered degrees of freedom. Currently we work in cooperation with geoscientists from the ETH Zurich, the University of Tübingen and the TU Dresden projects like:
  • Evolutionary Algorithms for the optimization and adjustment of large geothermal systems, which feature higher efficiency, lower enviromental impacts and material savings compared to standard practice.
  • Evolutionary Algorithms for the optimization of irrigation schedules.
  • Neural networks for the approximation of water transport processes in the vadose zone.
  • Neural networks for the approximation of tsunami inundation.

Temperature distribution of a non-optimized and an optimized borehole field after 30 years of simulation featuring the same energy extraction (MWh) per year.

Overview of the FIM model-system used for the simulation of crop-growth in deficit irrigation scenarios.

Publications

[1] Markus Beck, Peter Bayer, Michael de Paly, Jozsef Hecht-Mendez, and Andreas Zell. Geometric arrangement and operation mode adjustment in low enthalpy geothermal borehole fields for heating. Energy, 49:434--443, 2013. [ details ]
[2] Jozsef Hecht-Mendez, Michael de Paly, Markus Beck, and Peter Bayer. Optimization of energy extraction for vertical closed-loop geothermal systems considering groundwater flow. Energy Conversion and Management, 66:1--10, 2013. [ details ]
[3] Markus Beck, Michael de Paly, Jozsef Hecht-Mendez, Peter Bayer, and Andreas Zell. Evaluation of the Performance of Evolutionary Algorithms for Optimization of Low-Enthalpy Geothermal Heating Plants. In Genetic and Evolutionary Computation Conference, GECCO-2012, pages 1047--1054, Philadelphia, USA, July 2012. [ details ]
[4] Jozsef Hecht-Mendez, Michael de Paly, Markus Beck, Phillip Blum, and Peter Bayer. Strategic optimization of large-scale vertical closed-loop shallow geothermal systems. In Geophysical Research Abstracts, EGU12079, volume 14, Vienna, apr 2012. EGU General Assembly 2012.
[5] Michael de Paly, Jozsef Hecht-Mendez, Markus Beck, Philipp Blum, Andreas Zell, and Peter Bayer. Optimization of energy extraction for closed shallow geothermal systems using linear programming. Geothermics, 43:57--65, 2012. [ DOI | details ]
[6] Markus Beck, Jozsef Hecht-Mendez, Michael de Paly, Peter Bayer, Philipp Blum, and Andreas Zell. Optimization of the energy extraction of a shallow geothermal system. In Proceedings of the IEEE Congress on Evolutionary Computation (CEC), pages 3622--3628, Barcelona, Spain, July 2010. [ DOI | details ]
[7] Michael de Paly, Niels Schuetze, and Andreas Zell. Determining crop-production functions using multi-objective evolutionary algorithms. In Proceedings of the IEEE Congress on Evolutionary Computation (CEC), pages 1870 -- 1877, Barcelona, Spain, July 2010. [ DOI | details ]
[8] Peter Bayer, Michael de Paly, and Claudius N. Bürger. Optimization of high-reliability-based hydrological design problems by robust automatic sampling of critical model realizations. Water Resources Research, 46(5):W05504, May 2010. [ DOI | details | link ]
[9] Niels Schuetze, Michael de Paly, and Uri Shamir. Novel simulation-based algorithms for optimal open-loop and closed-loop. Journal of Hydroinformatics, 2010. [ DOI | link ]
[10] Michael de Paly and Andreas Zell. Optimal irrigation scheduling with evolutionary algorithms. In Lecture Notes in Computer Science (EvoWorkshops 2009), volume 5484, pages 142--151, Tübingen, Germany, 2009. Springer-Verlag Berlin Heidelberg. [ link ]
[11] Volker Gundelach, Michael de Paly, and Dieter Eisenburger. Recognition of patterns from geological structures in radar signals with the neuronal network simulator jnns. In Ultra-Wideband, 2008. ICUWB 2008. IEEE International Conference on, volume 3, pages 167 --170, September 2008. [ DOI ]
[12] Tim Häring, Michael de Paly, Carsten Henneges, and Volker Hochschild. Modelling tsunami vulnerability. the development of a tsunami inundation model with machine learning tools. In Digital Earth Summit on Geoinformatics 2008: Tools for Global Change Research, pages 182--187, Heidelberg, Germany, 2008. Wichmann.
[13] Gerd H. Schmitz, Thomas Wöhling, Michael de Paly, and Niels Schütze. Gain-p: A new strategy to increase furrow irrigation efficiency. Arabian Journal for Science and Engineering, 32(1):103--116, 2007.
[14] Niels Schütze, Thomas Wöhling, Michael de Paly, and Gerd Schmitz. Global optimization of deficit irrigation systems using evolutionary algorithms. In Proceedings of the XVI International Conference on Computational Methods in Water Resources, Copenhagen, Denmark, 2006.
[15] Niels Schütze, Michael de Paly, Thomas Wöhling, and Gerd Schmitz. Global optimisation of deficit irrigation systems using evolutionary algorithms and neural networks. In Proceedings of the ICID 21st European Regional Conference 2005, Integrated Land and Water Resources Management Towards Sustainable Rural Development, Frankfurt (Oder), Germany and Slubice, Poland, 2005.
[16] Niels Schütze, Thomas Wöhling, Michael de Paly, and Gerd Schmitz. Meeting challenges of the blue revolution: increasing irrigation efficiency with soft-computing optimisation methods. In Workshop on Integrated Water Research and Water Management, pages 92--95, Altmorschen (Kassel), Germany, 2004.

Contact

Markus Beck, Tel.: (07071) 29-78979, m.beck at uni-tuebingen.de

Michael de Paly, Tel.: (07071) 29-78979, michael.depaly at uni-tuebingen.de