Total number of participants: 17
Number of different nationalities represented: 6
Total number of speakers: 11
Total number of talks: 12

Background

In 2007, the Computational Geosciences project initiated a pilot study of a stochastic approach to simulation of sedimentation and erosion in a marine environment. It is generally accepted that the understanding of these processes is vital to the correct assessment of hydrocarbon potential in a sedimentary basin. However, the use of sedimentation models in the industry is sparse, mainly due to the facts that (1) several model parameters are subject to a large degree of uncertainty; and that (2) it is difficult to assess the geological relevance of the computed results. The introduction of mathematical techniques for quantifying and controlling the error in the model's predicition has a great potential for industrial impact. In addition to the stochastic approach, the CG project is also looking at the possibility of calibrating model parameters by measurements through use of an inverse problem approach.

As the next step in the direction of stochastic modelling, the CG project is currently establishing a research collaboration with Dr. Zenon Medina-Cetina, who is a highly qualified specialist in the field. In the fall of 2008 he will start buliding a research group for stochastic mechanics at the Texas A&M University. Through Simula's funding from StatoilHydro, this collaboration will include positions for two PhD students who will focus on stochastic methods applied to some of the geoscientific applications in the CG project's portfolio of activities. The intention is that over time this collaboration will be further strengthened, also involving other types of applications that are relevant to the SC department.

The purpose of this workshop is to provide an overview of the ongoing applications in Computational Geosciences, as well as trying to get as many as possible of the SC staff aware of the power that lies in stochastic simulation. The geoscience presentations will largely be aimed at computational issues and should be relevant to all types of SC personnel.

Program

09:30 Are Magnus Bruaset: Computational Geosciences at Simula
09:45 Omar al-Khayat: A coupled Lattice Boltzmann model for a turbulent sand-laden fluid flow
10:00 Joachim Berdal Haga: Simulation of deformation and heat flow in sedimentary basins
10:15 Xing Cai: A geostatistical inverse approach to tomographic surveys for hydrology

10:30 BREAK 

10:45 Achim Schroll: Parameter identification in dual lithology sedimentation - an inverse problem approach
11:00 Stuart Clark: Handling uncertainty in numerical models of sedimentation - a stochastic approach
11:15 Øyvind Hjelle: Surface reconstruction from trace data
11:30 Øyvind Hjelle: Compound modelling and fast marching level set

11:45 LUNCH 

12:30 Trond Vidar Stensby: Multi-resolution visualization of time-dependent horizons on the globe
12:50 Stuart Clark: Subduction dynamics, back-arc basin spreading and mantle convection in the Western Pacific
13:05 Are Magnus Bruaset: New activities 2008-2009
13:15 Zenon Medina-Cetina: Uncertainty-based predictions in geosciences, part I

14:00 FRIDAY CAKE BREAK 

14:30 Zenon Medina-Cetina: Uncertainty-based predictions in geosciences, part II

Abstract

The probabilistic solution to inverse problems is becoming a standard in the scientific community for improving inferences on the modelling of Geo-processes. This is because it allows for identifying, characterizing and quantifying the uncertainty contained in observations (data) and predictions (theoretical models) in a transparent, logical and elegant manner. Furthermore, approaches such as the Bayesian permits to introduce expert's opinion into the probabilistic formulation of the solution, taking advantage of subjective sources of information. Due to technological advances in Geo-Sciences such as emerging data acquisition technologies and advanced numerical modelling, assessing optimal representations of model predictions is not enough. Assessing associations and causal dependencies crossing the physical, model (parameters) and probability spaces, significantly increases the inferences that can be retrieved from Geo-models predictions, showing full joint probability representations of data and theoretical models, but most importantly measures of the relationships between them.

This presentation is divided in two parts. In the first part elements of probabilistic solution to inverse problems are presented and being illustrated by elemental applications. In the second part, space-time simulation techniques are introduced for applications ranging between stationary Gaussian to non-stationary non-Gaussian, as a way to represent unknown causal sources such as loadings, material properties, structure geometries, etc., and for potentially simulating massive computational predictions with high efficiency.

About the Speaker

Dr. Zenon Medina-Cetina holds currently a dual appointment at the International Centre for Geohazards and the Computational Geomechanics Division at the Norwegian Geotechnical Institute. He obtained a Master in Science in Soil Mechanics (Geostatistics) from the Universidad Nacional Autonoma de Mexico, and a Master in Science and PhD (Stochastic Mechanics) from the Johns Hopkins University. Recently he has been appointed Assistant Professor at the Civil Engineering Department of Texas A&M University.