EGU2020-1116, updated on 12 Jun 2020
https://doi.org/10.5194/egusphere-egu2020-1116
EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

A Hybrid PCG- Bat Algorithm for 2D Gravity Inversion: Applications for Ore Deposits Exploration and Interpretation of Sedimentary Basins

Mohamed Abdrabou1, Maha Abdelazeem2, and Mohamed Gobashy1
Mohamed Abdrabou et al.
  • 1Cairo University, Faculty of Science, Geophysics Department, Giza, Egypt (abdrabou.mohamed@yahoo.com)
  • 2National Research Institute of Astronomy and Geophysics (NRIAG), Helwan, Cairo, Egypt

Geophysical data such as gravity data can be inverted to get a subsurface image, which depicts the subsurface distribution of physical property. Consequently, inversion of geophysical data has an effective role for interpreting measured geophysical anomalies in hydrocarbons and mineral applications. Interest about ore deposits exploration and sedimentary basins interpretation is associated with their economic importance. The presence of sedimentary basins gives lower amplitude of gravity anomalies with negative signals, due to the negative density contrast as these sedimentary basins have lower density than that of the neighboring basement rocks. In prospecting ore deposits, studying the spatial distributions of densities in the subsurface is essential of significance.Two dimensional forward modelling strategy can be done via locating the rectangular cells with fixed size directly underneath the location of the observed data points using regular grid discretization. Density vector of the subsurface rectangular cells are obtained via solving the 2D gravity inverse problem by optimizing an objective function (i.e., the differences between observed and inverted residual gravity data sets). In this work, a hybrid algorithm merging a bat (BAT) algorithm with the preconditioned conjugate gradient (PCG) method is suggested as a mean for inverting surface gravity anomalies to obtain the density distribution in the subsurface. Like the hybrid, minimization algorithm has the capability to make use of the advantages of both two techniques. In this hybrid algorithm, the BAT algorithm was utilized to construct an initial solution for the PCG technique. The BAT optimizer acts as a rapid build-up of the model, whereas the second modifies the finer model approximated solution. This modern algorithm was firstly applied on a free-noise synthetic data and to a noisy data with three different levels of random noise, and good results obtained through the inversion. The validity and applicability of our algorithm are applied to real residual gravity anomalies across the San Jacinto graben in southern California, USA, and Sierra Mayor - Sierra Pinta graben, USA and prospecting of the Poshi Cu-Ni deposits, Xinjiang, northwest China. The obtained results are in excellent accordance with those produced by researchers in the published literature.

 

Keywords: Gravity data, 2D Inversion, BAT algorithm, Preconditioned Conjugate Gradient, Sedimentary Basins.

How to cite: Abdrabou, M., Abdelazeem, M., and Gobashy, M.: A Hybrid PCG- Bat Algorithm for 2D Gravity Inversion: Applications for Ore Deposits Exploration and Interpretation of Sedimentary Basins, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-1116, https://doi.org/10.5194/egusphere-egu2020-1116, 2019

Display materials

Display file

Comments on the display material

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

Display material version 1 – uploaded on 05 May 2020
  • CC1: Comment on EGU2020-1116, Peter Lelièvre, 05 May 2020

    Hello Mohamed. This is an interesting approach. The bat algorithm is a metaheuristic global optimization algorithm, right? I am unclear on why you feel a hybrid global-local algorithm is necessary for the gravity problem, for which the objective function is quadratic so local descent-based minimization should perform well. What does the bat algorithm do in the hybrid algorithm that local minimization fails to accomplish alone?

    • AC1: Reply to CC1, Mohamed Abdrabou, 05 May 2020

      Hi Dr.Peter Lelièvre,

       

      Thank you for your comment. In this hybrid algorithm, the BAT algorithm was utilized to
      construct an initial solution for the PCG technique. The BAT optimizer acts as a rapid build-up of
      the model, whereas the second modifies the finer model approximated solution.

       

      Utilizing a meta-heuristic algorithm to search for global minimum not for local minimum solution like local algorithms and I performed that to benefit from the hybrid technique (one global method and one local method) .

       

       

       

      • CC2: Reply to AC1, Peter Lelièvre, 05 May 2020

        Thank you for this additional information. However, I'm still unclear on how the global algorithm improves the situation. Multiple minima are not expected in the objective function, which is quadratic, so it seems a global optimization algorithm is not needed. If the BAT algorithm is able to generate an initial solution for the local minimization, then I suppose the solution time might be reduced. Is that your goal? If so, have you compared the speed of your hybrid algorithm against a local minimization working alone and starting from some random or zero-valued initial model?