Data Analysis and Mapping with Geostatistics| Training course

At the end of this course, you will be able to use geostatistics to:

• better understand and value your data,
• produce better quality maps by integrating various types of data,
• quantify the uncertainties on the calculated maps,
• understand the underlying assumptions and choose the most suitable geostatistical technique for your data.

• This course is devoted to methodological presentations and practical exercises on real-life cases to deepen the understanding of concepts. The focus is on illustrations and practical contributions of the covered concepts.
• Computer exercises with Isatis.neo software and air quality data. Depending on the participants’ profiles, additional case studies from other sectors of activity will be shown.
• Course material provided.

This course is aimed at those who work with spatial data and wish to understand and put into practice the geostatistical methods for mapping: academics, agricultural engineers, air quality engineers, climatologists, environmental consultants and engineers, epidemiologists, foresters, geotechnical engineers, soil scientists, etc.

SESSION 1: ANALYSING DATA AND THEIR VARIABILITY IN SPACE

• Introduction:
  – Added-value of geostatistics compared to deterministic methods for spatial interpolation.
  – Presentation of usual deterministic interpolation methods (nearest neighbor, moving average, inverse distance, etc.) and their application limits.
• Exploratory data analysis and validation:
  – Use of statistics for data analysis and quality control, and identification of outliers: mean, variance, histogram, correlation coefficients, linear regression, etc.
  – Data visualization in 2D and 3D.
  – Spatial variability: concept and quantification through the calculation, interpretation, and modeling of the experimental variogram.
  – The various theoretical variogram models. Fitting of the experimental variogram.

SESSION 2: MAPPING

• Interpolation by kriging:
  – Kriging principles and properties. The smoothing effect of kriging.
  – Definition of the most appropriate neighborhood (single or moving, size, number of samples, etc.).
  – Analysis of the kriging weights according to the sample positions, the neighborhood, the presence or not of a nugget effect.
• Cross validation to validate the variogram model.
• The different variants of kriging (simple, ordinary, with variance in measurement error, etc.).

SESSION 3: REFINING THE MAPS

• Multivariate geostatistics: to reduce interpolation uncertainty
  – Analysis of the correlations between multidisciplinary data, quantitative and semi-quantitative (remote sensing data,  DEM, soil occupation, physicochemical models, lithology, etc.): calculation of scatter plots and coefficients of correlation.
  – Analysis of the spatial relationships between variables: multivariate variogram calculation and modeling.
  – Integration of one or several secondary variables in the interpolation: cokriging principles and practice. Collocated cokriging.
  – Analysis of the inputs of cokriging compared to kriging.
• Non-stationary geostatistics: when data show a drift or a trend.
• Introduction to simulations and risk analysis: added-value of simulations, examples.

SESSION 4: PRACTICAL EXERCISES

The course does not require prior knowledge of geostatistics. However, it is recommended to have a basic knowledge of statistics.