About this Research Topic
In recent years, data science has attracted great attention from researchers and practitioners from both academia and industry, for its huge potential in lowering costs, improving efficiency, increasing safety, reducing emissions, and unlocking new business opportunities. There have been substantial efforts dedicated to the research and applications of data science in various disciplines, including those in earth science, e.g., atmosphere, ocean, hydrology, petroleum, to name a few.
On the other hand, inverse and optimization theories also find wide applications in earth science. Examples in this regard include (but not limited to): state and/or parameter estimation of earth models, improved hydrocarbon recovery for subsurface reservoirs, CO2 sequestration, weather forecast, and so on.
In comparison to the recently emerged data science, inverse and optimization theories have a relatively longer history within earth science. While various inverse and optimization methods have been well established and successfully applied to real-world problems, there is still room to further improve and strengthen their performance and applicability to real field case studies, in terms of, e.g., estimation accuracy, computational efficiency, and user convenience.
As such, one of the goals of the proposed research topic is to facilitate innovation and new knowledge generation within the areas of inverse and optimization theories, through multidisciplinary approaches. Of our particular interest are ensemble-based algorithms, such as various forms of ensemble Kalman filter (EnKF), iterative ensemble smoother (IES), ensemble optimization (EnOpt), which, in the past two decades, have been widely adopted to tackle numerous inversion or optimization problems in various disciplines of earth science, due to their simplicity in implementation, their derivative-free and non-intrusive nature, and their capability of providing uncertainty quantification for the inversion or optimization results. On the other hand, we also welcome contributions focusing on other types of more conventional inversion or optimization algorithms, such as 3D- or 4D-variational data assimilation (3D-Var or 4D-Var) algorithms, local gradient or global search based optimization algorithms (e.g., particle swarm optimization) etc.
Given the recent advances in data science (including those in artificial intelligence, data analytics, machine/deep learning etc.) and their successful applications to various real-world problems, we believe that incorporating data science into the inverse and optimization theories, including but not limited to aforementioned ensemble-based inversion and optimization algorithms, would help achieve this goal. In addition, with the similarities among certain areas (e.g., machine/deep learning) of data science and inverse and optimization problems, we expect that the newly developed multidisciplinary approaches may in turn enrich the knowledge base of data science.
The scope of the Research Topic is mainly on the development of new inversion/optimization workflows by combining methods and algorithms from both data science and inverse/optimization theories. More specifically, we expect that the following aspects will be covered within the Research Topic:
• Representation learning to discover lower-dimensional representation of big earth models and/or big field datasets.
• Model reduction or creation through data-driven approaches, for applications that lack physical earth models, or for management of complex earth systems.
• New inversion/optimization algorithms developed for inverse/optimization/learning problems.
• Other aspects in inverse/optimization problems of earth science that involve the use of data science.
• Applications to field case studies.
While all article types can be accepted as part of this Reserch Topic, we expect that the manuscripts would mainly be of Type A, on original research, systematic review etc.
---
Disclosure: Topic Editor Dr. Alexandre Anozé Emerick is employed by Petrobras.
On the other hand, inverse and optimization theories also find wide applications in earth science. Examples in this regard include (but not limited to): state and/or parameter estimation of earth models, improved hydrocarbon recovery for subsurface reservoirs, CO2 sequestration, weather forecast, and so on.
In comparison to the recently emerged data science, inverse and optimization theories have a relatively longer history within earth science. While various inverse and optimization methods have been well established and successfully applied to real-world problems, there is still room to further improve and strengthen their performance and applicability to real field case studies, in terms of, e.g., estimation accuracy, computational efficiency, and user convenience.
As such, one of the goals of the proposed research topic is to facilitate innovation and new knowledge generation within the areas of inverse and optimization theories, through multidisciplinary approaches. Of our particular interest are ensemble-based algorithms, such as various forms of ensemble Kalman filter (EnKF), iterative ensemble smoother (IES), ensemble optimization (EnOpt), which, in the past two decades, have been widely adopted to tackle numerous inversion or optimization problems in various disciplines of earth science, due to their simplicity in implementation, their derivative-free and non-intrusive nature, and their capability of providing uncertainty quantification for the inversion or optimization results. On the other hand, we also welcome contributions focusing on other types of more conventional inversion or optimization algorithms, such as 3D- or 4D-variational data assimilation (3D-Var or 4D-Var) algorithms, local gradient or global search based optimization algorithms (e.g., particle swarm optimization) etc.
Given the recent advances in data science (including those in artificial intelligence, data analytics, machine/deep learning etc.) and their successful applications to various real-world problems, we believe that incorporating data science into the inverse and optimization theories, including but not limited to aforementioned ensemble-based inversion and optimization algorithms, would help achieve this goal. In addition, with the similarities among certain areas (e.g., machine/deep learning) of data science and inverse and optimization problems, we expect that the newly developed multidisciplinary approaches may in turn enrich the knowledge base of data science.
The scope of the Research Topic is mainly on the development of new inversion/optimization workflows by combining methods and algorithms from both data science and inverse/optimization theories. More specifically, we expect that the following aspects will be covered within the Research Topic:
• Representation learning to discover lower-dimensional representation of big earth models and/or big field datasets.
• Model reduction or creation through data-driven approaches, for applications that lack physical earth models, or for management of complex earth systems.
• New inversion/optimization algorithms developed for inverse/optimization/learning problems.
• Other aspects in inverse/optimization problems of earth science that involve the use of data science.
• Applications to field case studies.
While all article types can be accepted as part of this Reserch Topic, we expect that the manuscripts would mainly be of Type A, on original research, systematic review etc.
---
Disclosure: Topic Editor Dr. Alexandre Anozé Emerick is employed by Petrobras.
Keywords: Data Science, Artificial Intelligence, Inverse Problem, Machine Learning, Deep Learning, Optimization
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.
