Machine Learning Methods for Inverse Modeling

Geostatistics has become a preferred tool for the identification of lithofacies from sparse data, such as measurements of hydraulic conductivity and porosity. Recently we demonstrated that the support vector machine (SVM), a tool from machine learning, can be readily adapted for this task, and offers significant advantages. On the conceptual side, the SVM avoids the use of untestable assumptions, such as ergodicity, while on the practical side, the SVM out performs geostatistics at low sampling densities. In this study, we use the SVM within an inverse modeling framework to incorporate hydraulic head measurements into lithofacies delineation, and identify the directions of feuture research.

[1]  Alberto Guadagnini,et al.  Conditioning mean steady state flow on hydraulic head and conductivity through geostatistical inversion , 2003 .

[2]  S. P. Neuman,et al.  Estimation of Aquifer Parameters Under Transient and Steady State Conditions: 3. Application to Synthetic and Field Data , 1986 .

[3]  Daniel M. Tartakovsky,et al.  Delineation of geologic facies with statistical learning theory , 2004 .

[4]  Daniel M. Tartakovsky,et al.  Probabilistic reconstruction of geologic facies , 2004 .

[5]  S. P. Neuman,et al.  Estimation of aquifer parameters under transient and steady-state conditions: 2 , 1986 .

[6]  B. Wohlberg,et al.  Support Vector Machines for Delineation of Geologic Facies from Poorly Differentiated Data , 2006 .

[7]  M. Eppstein,et al.  SIMULTANEOUS ESTIMATION OF TRANSMISSIVITY VALUES AND ZONATION , 1996 .

[8]  S. P. Neuman,et al.  Inverse stochastic moment analysis of steady state flow in randomly heterogeneous media , 2006 .

[9]  Mikhail Kanevski,et al.  Analysis and modelling of spatial environment data , 2004 .

[10]  W. Yeh,et al.  Identification of Parameter Structure in Groundwater Inverse Problem , 1985 .

[11]  Frank T.-C. Tsai,et al.  Characterization and identification of aquifer heterogeneity with generalized parameterization and Bayesian estimation , 2004 .

[12]  Andres Alcolea,et al.  Pilot points method incorporating prior information for solving the groundwater flow inverse problem , 2006 .

[13]  Alexander J. Smola,et al.  Learning with kernels , 1998 .

[14]  Daniel M. Tartakovsky,et al.  Subsurface characterization with support vector machines , 2006, IEEE Transactions on Geoscience and Remote Sensing.