The awarded paper presents a new approach based on principal component analysis (PCA) for the representation of complex geological models in terms of a small number of parameters. Unlike standard PCA-based methods, in which the high-dimensional model is constructed from a small set of parameters by simply performing a multiplication using the basis matrix, in the O-PCA approach, the mapping is formulated as an optimization problem. This enables the inclusion of bound constraints and regularization, which are shown to be useful for enhancing highly connected geological features and non-Gaussian property distributions. Such a parameterization is particularly useful within a data assimilation (history matching) framework because it can reduce computational cost, mitigate the ill-posedness of the inverse problem, and retain geological realism in history-matched models. Furthermore, the differentiable nature of the parameterization enables its use for gradient-based minimization procedures, which can improve further the efficiency of the history matching process. The MATLAB code for the O-PCA model construction (including examples) is provided as online Supplementary Material. This is the first time that source code has been published together with a paper in Mathematical Geosciences.
Last but not least
The Best Paper Award is, as always, a major recognition of the effort of the authors to reach excellence. Congratulations to the 2014 winners and a most sincere thanks for their efforts, as well as their contribution to Mathematical Geosciences and the profession.