| |
|
QG services |
|  |
courses > non-linear geostatistics
Previous Master Class course with Jacques Deraisme in 2005
The course was lead by Dr Jacques Deraisme, one of the worlds
leading researchers and teachers in the field of mining geostatistics,
with assistance from three QG personnel.
Jacques Deraisme
Jacques Deraisme is the Technical Manager in charge of the consulting
and training at Geovariances in Fontainebleau, France. After
graduating as a Mining Engineer (Ecole des Mines de Nancy, 1969)
he worked from 1972 to 1986 at the Centre de Géostatistique
de l'Ecole des Mines de Paris where his research activities
focused on stochastic and mining simulations, and from where,
in 1978, he obtained his PhD. Since 1986 has worked for Geovariances
(he was a co-founder) and conducted many geostatistical studies
for mining and oil companies. Jacques has led numerous training
courses in applied geostatistics and we look forward to co-presenting
this important course with him.
Attendees
Specialist Resource Geologists, Geostatisticians and Geology
Managers from the following sites/companies attended:
- Anglocoal
- BHP Billiton Iron
- BHP Billiton Mitsubishi Coal
- Cadia Valley Operations
- Cannington Mine
- Cowal Mine
- Leinster Nickel
- Lihir Gold Mine
- Mount Keith Nickel
- Olympic Dam Mine
- St Barbara Mines
- St Ives Gold Mine
- Sunrise Dam Gold Mine
Click image for large course group photo.
Course Outline
Day one provided a crash introduction to refresh the main concepts
prior to the subsequent, more technical three days. In addition,
on day one, two detailed case studies comparing Uniform Conditioning
and Indicator methods were presented giving participants a practical
exposure to Non-Linear techniques. These courses were based
on QG consulting experience using non linear estimation methods
at the Lihir and Cowal mines.
The course wraped up with a half day session on simulation and
a forum on future directions of non-linear geostatistics in
mining, with emphasis on multivariate methods, including an
innovative proposal for multivariate Uniform Conditioning.
The material covered was illustrated throughout using Isatis
examples and participants could follow the studies on their
own computer (for those who did not have Isatis, a fully functioning
trial version was arranged by Geovariances).
The course topics were:
Day 1 Tuesday : Introduction
and Reminders
- Introduction and refresher in Linear geostats
- Why non-linear estimation?
- Overview of main methods ('high level', without technical
details)
- Two case studies (UC and IK)
- Crash course in using Isatis
Day 2 Wednesday : Global Recoverable
Resources and Local Recoverable Resources I
- Dispersions and estimation variance
- Linear combinations
- Global recoverable resources
- Discrete Gaussian Model and Hermite polynomials
- Indicator kriging
- Median Indicator kriging
- Multiple Indicator kriging
- Disjunctive kriging
- Post-processing
Day 3 Thursday : Local Recoverable
Resources II
- Service Variables
- Conditional Expectation
- Gaussian Disjunctive Kriging
- Uniform Conditioning by Panel Grade
- Incorporation of Information to Local Estimation
- Step by Step Implementation of UC in Isatis
PM - course dinner
Day 4 Friday : Simulation and
Wrap Up/Forum
- Introduction to Simulations
- Risk Mapping
- Turning Bands
- SGS and SIS
- Boolean Models
- Building CI's
- Alternative for CI's using DGM
- Forum on expansion of Non Linear Methods in particular UC
to multivariate Setting
- Wrap up
QG can organise this course, or others for an in house group
on request. See also QG Courses for
more details. |
|
Scott Jackson (left), Scott Dunham (centre), and John Vann (right) are the Directors of Quantitative Group
|
|
