Application of genetic algorithms to constrain shallow elastic parameters using in situ ground inclination measurements

Abstract

Among the class of global optimization techniques, which includes Monte Carlo and simulated annealing methods, the Genetic Algorithms constitute a new class of methods to solve highly non-linear optimization problems. The issue has generated considerable interest in the field of artificial intelligence, and recently, in some multi-parameter optimization geophysical problems. In this study, we explore the applicability of genetic algorithms to the inversion of high resolution ground inclination measurements produced by known loads placed at known distances. Our objective is to find a model for dynamic properties of the subsoil such as shear and compressional wave velocities and depth distributions of the uppermost strata, which are related to elastic moduli. Three parameters are needed for describing elastic isotropic horizontally homogeneous media: mass density ρ and Lame constants λ and μ or mass density and P-wave and S-wave velocities. In general, the choice of parameters is not always a simple matter. In fact, although theoretically equivalent, if they are not adequately chosen, the numerical algorithms in the inversion can be inefficient. Ground inclination surveys were performed at the virgin Texcoco Lake, near Mexico City, and at the European Synchrotron radiation facility (ESRF) of Grenoble, France. From both sets of data we study the feasibility of applying genetic algorithms to rapidly and effectively explore the model space to find an optimal model for the shallow structure under study. Forward solution of vertical and radial displacements in a layered medium, under static loads, is calculated by means of the stiffness matrix approach (Kausel, E. & Roesett, J. M. Stiffness matrices for layered soil. Bulletin of the Seismological Society of America, 1981, 71(6), 1743–1761.) Comparisons with results from common established techniques such as seismic wave refraction profiles, cone resistance values and inversion of surface wave dispersion curves were used for validation purposes. Our procedure represents a fast and accurate alternative method to infer the shallow elastic parameters in surficial soils.