Interpretation of self-potential anomaly over a 2D inclined structure using very fast simulated-annealing global optimization; an insight about ambiguity
Interpretation of self-potential anomaly over a 2D inclined structure using very fast simulated-annealing global optimization; an insight about ambiguity (in Assessing uncertainty in geophysical problems, Aime Fournier (prefacer), Klaus Mosegaard (prefacer), Henning Omre (prefacer), Malcolm Sambridge (prefacer) and Luis Tenorio (prefacer))
Geophysics (May 2013) 78 (3): WB3-WB15
- Bavaria Germany
- Bavarian Forest
- Central Europe
- copper ores
- data acquisition
- data processing
- electrical anomalies
- electrical methods
- Europe
- geophysical methods
- geophysical surveys
- Germany
- graphite deposits
- interpretation
- metal ores
- optimization
- ore bodies
- self-potential methods
- surveys
- three-dimensional models
- two-dimensional models
- uncertainty
- Rakha Mines
- Surda India
- fast simulated-annealing
A very fast simulated-annealing (VFSA) global optimization procedure is developed for the interpretation of self-potential (SP) anomaly measured over a 2D inclined sheet-type structure. Model parameters such as electric current dipole density (k), horizontal and vertical locations of the center of the causative body (x (sub o) and h), half-width (a), and polarization/inclination angle (a) of the sheet are optimized. VFSA optimization yields a large number of well-fitting solutions in a vast model space. Even though the assumed model space (minimum and maximum limits for each model parameter) is appropriate, it has been observed that models obtained by the VFSA process in the predefined model space could also be geologically erroneous. This offers new insight into the interpretation of self-potential data. Our optimization results indicate that there exist at least two sets of solutions that can fit the observed data equally well. The first set of solutions represents a local optimum and is geologically inappropriate. The second set of solutions represents the actual subsurface structure. The mean model estimated from the latter models represents the global solution. The efficacy of the developed approach has been demonstrated using various synthetic examples. Field data from the Surda area of Rakha Mines, India and the Bavarian woods, Germany are also interpreted. The computation time for finding this versatile solution is very short (52 s on a simple PC) and the proposed approach is found to be more advantageous than other approaches.