There is no charge. Non-members of the CSPG are also welcome. Please bring your lunch. For details or to present a talk in the future, please contact Weishan Ren via email at

Geomodeling Committee

Weishan Ren | Chairperson         Tom Cox                      Hayley Silberg             
David Garner | Co-chair              Sasan Ghanbari           Darcy Novak

Olena Babak                               Damien Thenin             Eric Niven  

Upcoming Division Talk

Integration of Geophysical Data into Multivariate Geostatistical Modeling
Maryam Hadavand 
Center for Computational Geostatistics, University of Alberta

12:00 Noon 
Thursday, March 29, 2018 
Husky Conference Room A, 3rd Floor, +30 level, South Tower, 707 8th Ave SW, Calgary, Alberta

Integration of different source of geological information and developing improved numerical modeling techniques leads to more accurate and precise geostatistical models. Reservoir engineering and management tools use geostatistical models to make better decisions and optimize hydrocarbon recovery. Therefore, these developments and data integration will improve flow forecasting, hydrocarbon recovery and reservoir management. Among the different source of geological information, seismic data has major contribution in hydrocarbon exploration because it responds to multiscale subsurface features. The conventional stochastic seismic inversion techniques integrates different source of data to obtain high quality geostatistical models. These techniques initially provide acoustic impedance models that match both well and seismic data. Such acoustic impedance models are related to reservoir physical properties via rock physics models. These techniques aim to reproduce the data within the quality of original data. However, the results are sensitive to the relation between acoustic and physical reservoir properties to drive the final reservoir model. Fidelity with the original seismic data will be reduced due to an element of randomness at each step. To overcome this issue a new approach called "Multivariate Stochastic Seismic Inversion" is proposed.

This approach presents a research toward a fully coupled categorical - multivariate continuous reservoir modeling in stochastic inversion context with Petro-Elastic Model (PEM) and convolution. In this approach, the multiple categorical - continuous reservoir properties are simulated via the multivariate Gaussian simulation technique. This technique simulates multiple Gaussian variables to model facies and reservoir physical properties at the same time. The Gaussian variables related to the facies are simulated and turned to indicator vectors of facies by truncated (pluri) Gaussian approach. For each reservoir physical property, a Gaussian variable is simulated by sequential Gaussian simulation per facies categories. Based on the indicator value of each cell, the corresponded simulated continuous variables retained as simulated value of that cell. The new approach combines a trace by trace adaptive sampling algorithm with multivariate geostatistical techniques to pick the best physical properties of reservoir that match the actual seismic data. The adaptive sampling method uses the acceptance-rejection approaches to condition geostatistical models to both well and seismic data. This technique samples the data inside the size of space of uncertainty in presence of different parameters and rejects them if they are not in the target zone. This method helps generate reasonable number of realizations dynamically via the multivariate geostatistical algorithm at each location by quantifying the size of space of uncertainty.

The application of multivariate geostatistical techniques with a quantified space of uncertainty would improve conventional stochastic inversion approaches. Modeling multiple reservoir properties simultaneously through the close integration of seismic inversion and multivariate geostatistical techniques would lead to high resolution reservoir property models that are suitable for improved reservoir management. A synthetic case study compares the multivariate stochastic inversion with conventional stochastic inversion method.

Figure 1: Defining the dynamic number of local realization based on the size of space of uncertainty and acceptance - rejection criteria.

Figure 2: Comparison of multivariate and conventional stochastic inversion with actual seismic data.

Maryam currently is a postdoctoral fellow at University of Calgary working on EM modeling in SAGD projects. She has about 10 years of working experience in Oil & Gas industries with National Iranian Oil Company (NIOC) and Schlumberger as a geophysicist and geomodeler. Maryam holds a PhD in Geostatistics from University of Alberta, a MSc degree in geophysics from Technical University of Shahrood, Iran, and a BSC degree in applied physics from Iran University of Science and Technology (IUST). She is an active member of Society of Exploration Geophysics (SEG) and a member of Canadian Society of Petroleum Geology (CSPG). Her research interests are in the area of Geostatistics and Geophysics specifically focuses on Multivariate Geostatistical Modeling, Seismic Inversion, EM Modeling and Reservoir Simulation.

Division Profile
The mandate of the Geomodeling Division is to provide CSPG members with opportunities for education and information related to technical developments in the subject areas of geomathematics and computer technologies as they are used in the pursuit of petroleum exploration and development. As a main contribution of the division, technical luncheon presentations are held once a month, usually on the last Wednesday of the month. 

The subjects that are presented in these technical talks include, for example, The latest developments in geomathematical applications, Geological modeling technology, Geostatistical approaches to modeling and risk analysis, Geological case studies using computer technology and the benefits, Digital data organization - storage and retrieval. In addition, ad hoc forums may be organized where members can discuss geomathematical and geological computer issues with experts in the field.The Geomodeling Division does not endorse or promote the use of specific commercial software products, nor does it perform any testing or comparative studies of such products.We do encourage volunteers to present public talks on case histories that illustrate the use of technology and methods.

The success of the Division depends on volunteer participation. CSPG members are encouraged to attend the activities of the Geomodeling Division and to be involved in organizing these activities. Division meetings are held once a month over lunch. If you are interested in joining this committee or if you have suggestions for luncheon talks or other activities, please contact any members of the committee.