Rock physics properties from seismic attributes with global optimization methods
Qi Hu, Kristopher A. Innanen
The estimation of rock physics properties from seismic attributes is a nonlinear inverse problem. We investigate three global optimization methods: simulated annealing, genetic algorithm, and neighborhood algorithm for solving this problem. The input data are P- wave velocity, S- wave velocity, and density, and the rock physics properties to estimate are porosity, clay content, and water saturation. The two parameter sets are connected by an assumed rock physics model. Numerical examples are suggestive that the neighborhood algorithm is most efficient for improving data fit for the experiment we set up; porosity and clay content can be accurately estimated, whereas the water saturation estimate is prone to large errors. We explain this as a consequence of the low sensitivity of velocities and density to this property. However, simultaneous inversion for the whole set of the rock physics properties is problematic if the input data are erroneous. Consequently, we restrict the inversion to porosity and clay content only and assume a priori information of the exact water saturation. This makes the inversion stable with noisy data. Finally, we illustrate the application of the proposed global optimization method using the high-resolution results of elastic full waveform inversion (EFWI).