Isatis 2014 allows better estimation control

Offering kriging and simulations with bayesian drift

In presence of trends, the new kriging with bayesian drift bridges the gap between the traditional kriging with external drift and a simple kriging of the residuals, allowing a better trend control. The methodology is of particular interest when the amount of data is scarce.

Indeed, in case of non stationarity, kriging methods are based on the dichotomy of the variable which can be written as the sum of a trend and a stationary residual.

With bayesian kriging, you provide the prior knowledge you have gained from similar fields regarding the trend coefficients. These trend coefficients are represented by a random variable characterized by a known prior distribution. Correlation between coefficients can also be controlled. When data are numerous, you can also control the local posterior behaviour of the trend coefficients.

The methodology is particularly suited to oil reservoir structure modelling for which input data derive only from a few wells and a certain amount of seismic data. It is also meaningful for air quality mapping where maps are obtained from a limited number of air pollution measurements supplemented with auxiliary information such as the output of physic-chemical models.

Avoiding negative weights

Rescaled Ordinary Cokriging (RCK) has been implemented to reduce the risk of getting negative estimates because of the negative weights which are assigned to the secondary variables

Supporting several auxiliary variables

Collocated Cokriging may now use several auxiliary variables to improve the estimation precision. This feature has also been added to the turning bands simulations.

Checking grade-tonnage variable consistency

New QTM Validation checks the consistency of (Q,T,M) grade-tonnage variables computed from Uniform Conditioning, Localized Uniform Conditioning or Grade Reblocking or imported from other software packages and applies automatic corrections if necessary.

Providing new control parameter

Block kriging now lets you compute Kriging Efficiency, giving you an additional way to assess the kriging estimate quality.

Kriging Efficiency measures the expected error for each block grade. A high value indicates that the estimate tends to correctly reproduce the block grade.

Isatis 2014 enhances facies modeling

Providing dedicated simulated facies post-processing

New Facies Simulation Post-processing lets you compute the local probability for each facies and store the locally most and least probable facies.

The application may also apply a global proportion correction to ensure that the original facies proportions are honoured, thus avoiding that scarce facies disappear from the field.

The volume distribution of each facies is also calculated.

Improving model conditioning to data

Flumy, the process-based simulation algorithm dedicated to meandering channelized systems, has been improved to ensure perfect conditioning to wells data.

Improving facies proportion control

Multiple-Point Simulations have been updated with version 2.1.0 of the Impala library provided by Ephesia, ensuring a better reproduction of the input proportions.

Isatis 2014 facilitates seismic data processing

• New Velocity Calculator delivers average velocity and Dix velocity from RMS velocity. The calculations can be applied in time or in depth.

    These velocities are particularly useful for time-to-depth conversion or incidence angle analysis.

• Import and export of SEG-Y data have been optimized. A better trace visualization and accurate statistics help adjusting import parameters at best. Because most of the industry software packages require that the SEGY cube starts at - or at least near - 0, when exporting data in SEG-Y format, it is now possible to introduce on the fly additional layers when the origin of the Isatis 3D grid is not close enough to 0.

Isatis 2014 improves data knowledge

• Exploratory Data Analysis now delivers Normal Probability Plots for testing distribution normality. The chart compares the normal distribution displayed as a probability against any variable displayed in quantiles. The closer the distribution to a normal distribution, the closer the points to a line.
• Box-plots or swath-plots can now be computed in any direction of the space. It allows for instance comparing trends on boreholes and estimates.

Isatis 2014 increases performances

• Quick Interpolate runs up to five times faster thanks to the multithreading implementation of the following interpolation algorithms: inverse distance, moving average, moving median, nearest neighbourhood or linear kriging.
• Selection from Wireframes has been optimized so that its performance has been drastically improved from 10 to 100 times.