Proposed topics of GEOSTAT2018 Workshop
Wroclaw University of Science and Technology, Wroclaw, 22.01.18-25.01.18
Professor Barbara Namysłowska-Wilczyńska (Wroclaw University of Science and Technology, Poland)
Case Studies (“Geostatistical methods for mineral deposits applications” and “Assessment of copper ore deposit variability”)
Zechstein sedimentary Cu ore deposits:
- Exploratory analysis: Evaluation of basic statistics; histograms analysis; Q – Q plots; scatter – diagram plots; variogram; cross-variogram: isotropic (omnidirectional) variogram; directional variogram; cross-variogram.
- Variogram (semivariogram) modelling.
- Interpolation:Quick interpolation techniques: inverse distances; Kriging with linear model; nearest neighbours
- Estimation: Ordinary Kriging, cokriging; filtering model components; Kriging efficiency variables; simple Kriging; bundle indicator Kriging.
- Visualization: Raster, isolines, indicator maps: (projection, in perspective), block-diagrams of co – kriging estimates.
Simulation turning bands method
- Exploratory analysis: Evaluation of basic statistics; histograms analysis; Gaussian anamorphosis modelling; Gaussian variogram; Gaussian cross-semivariogram.
- Gaussian variogram (semivariogram) modelling
- Conditional turning bands simulation: Simulated macro-simulation variable; post-processing: Statistical maps of realizations, iso-frequency maps, iso-cutoff maps, risk curves.
- Visualization: Raster, isoline, threshold maps: (projection, in perspective), block-diagrams of simulated values (realizations).
Dr Emmanouil Varouchakis (Technical University of Crete)
“Space-time geostatistics for environmental applications”
- Spatial Geostatistics principles (Deterministic & Stochastic methods)
- Data normalization methodologies (Box-Cox, Modified Box-Cox, Trans-Gaussian kriging, Gaussian anamorphosis)
- Spatial & spatio-temporal variography
- Spatio-temporal Kriging methods
- Simulations, uncertainty estimation (Bayesian, conditional and unconditional simulation methodologies)
- Applications to environmental case studies
Professor Dionissios Hristopulos (Technical University of Crete)
“Spartan spatial random fields and stochastic local interaction (SLI) models”
- Introduction to Spartan spatial random fields (SSRFs): motivation and properties
- New SSRF covariance functions: properties and examples of applications to real data.
- Discretized SSRF models for data on regular grids and connection with Gauss-Markov random fields.
- Parameter estimation and interpolation, uncertainty estimation.
- Geometric anisotropy: an introduction and anisotropic SSRF models on regular grids.
- Introduction to the Stochastic Local Interaction (SLI) model for irregularly spaced data: kernel functions, adaptive selection of the kernel bandwidth, sparse precision matrices and applications to big spatial data.
- Applications to synthetic and real data, extensions to space-time models, topics for further research.
Prof. Vasily Demyanov (Heriot-Watt University, Edinburgh, United Kingdom)
“Uncertainty quantification in spatial geoscience prediction problems: geostatistics, machine learning, Bayesian inference”
- Introduction: Sources of uncertainty and it’s perception in prediction modelling
- How uncertainty is modelled in geostatistics: geological and environmental case studies.
- Uncertainty prediction with calibrated models: inverse problems and Bayesian inference for subsurface reservoir flow modelling.
- Machine learning applications for geological case studies:
– use of intelligent prior information based on modern river analogues
– modelling porous property complex multiscale fluvial deposits and flow prediction
– machine learning facies classification based on core interpretation and wireline logs: real deposit and modern river case studies.
Professor Mikhail Kanevski (University of Lausanne, Switzerland)
“Machine learning in geosciences”: Fundamentals of machine learning
- Generic methodology of learning from data.
- Intelligent exploratory data analysis.
- Machine learning algorithms (basics).
- Advanced topics and new challenges.
- Simulated and real data case studies.
- Conclusions and future research.
AIMS OF GEOSTAT2018 WORKSHOP
The dissemination of methods of applied (spatial) statistics, i.e., geostatistical methods, to representatives of various scientific disciplines (geology, mining, environmental engineering, climatology, renewable energy resources, etc.). The attendees are expected to have knowledge and expertise in diverse areas of significant interest for the national economy.
The creation of opportunities to acquire both basic and applied knowledge, founded on a firm theoretical background, of geostatistical methods.
The development of practical skills for the use of geostatistics and specialized software during the Workshop sessions.
An original and innovative (modern) research methodology which involves the use of geostatistical methods, and it is applied to the spatial analysis of the variability of regionalized variables (i.e. variables whose position in space is described by two (plane) or three coordinate values and possibly time as well for dynamic phenomena.
The scope of applications of the geostatistical methodology is increasing following the creation and availability of spatial databases in various scientific areas, and due to the increasing capacity (content) of existing databases. This increasing availability of spatial data generates new possibilities for many fields of research, and the geostatistical methodology provides a practical suite of tools that can be used to process and extract information from spatial data.
Scope of the Workshop
- Concepts of geostatistics.
- Research methodology based on the geostatistical approach.
- Empirical measures of spatial variability and respective theoretical models.
- Variogram estimation and modeling.
- Cross-validation and model selection.
- Kriging estimators (kriging, cokriging).
- Quality and effectiveness of kriging techniques.
- Conditional simulations and risk mapping.
- Methods of fast interpolation (e.g., kriging linear model, inverse distances, nearest neighbours).
- Case studies and best practices.
- Introduction to environmental data mining using machine (computer) learning algorithms.
- Sparse precision matrices, stochastic local interactions and quick spatial interpolation.