Geostatistics and Data Analysis
Objectives
The aim of this course is to provide students with tools for data analysis and processing, including univariate, bivariate, and multivariate analysis, geostatistics, spatial analysis, and kriging estimation, applied to the field of Earth Sciences. In terms of general skills, students will be able to integrate with teams that undertake studies involving data analysis and processing, such as analytical determinations to evaluate the quality of soils and groundwater, and geostatistical modeling studies, including variography and kriging estimation.
Some more specific competencies include:
- For a specific data set, being able to perform preliminary statistical analysis based on the available variables.
- Selecting the most appropriate data analysis methods for each situation (univariate, bivariate, or multivariate analysis).
- Analyzing the results obtained by one or more methods and relating them to each other in a descriptive perspective of the case study.
- Calculating variograms for a georeferenced data set, researching geometric and zonal anisotropies, and adjusting theoretical variogram models. Relating the results obtained with the representativeness of the sampling.
- Creating prediction maps for a specific variable, and criticizing the results and limitations. Interpreting estimation errors and result validity.
General characterization
Code
10666
Credits
6.0
Responsible teacher
José António de Almeida
Hours
Weekly - 4
Total - 68
Teaching language
Português
Prerequisites
Elementary knowledge of probability and statistics.
Bibliography
[1] Richard A. Johnson & Dean W. Wichern, Applied Multivariate Statistical Analysis, Prentice Hall, 2002, ISBN: 0-13-092553-5 (paperback).
[2] Amílcar Soares. Geoestatistica para as Ciências da Terra e do Ambiente. IST Press, 2014 (2ª edição), 232p.
[3] Edward H. Isaaks, R. Mohan Srivastava. Applied Geostatistics. Oxford University Press, 1989, ISBN: 0-195050134 (paperback).
[4] Pierre Goovaerts. Geostatistics for Natural Resources Evaluation. Oxford University Press, 1997. ISBN: 0-195115384 (hardcover).
Teaching method
Exposure with Powerpoint and board and practical classes where students solve problems devoted to each main topic: (1) univariate analysis; (2) bivariate analysis; (3) multivariate analysis; (4) variography; (5) kriging.
Evaluation method
The evaluation of the course consists of two components: theoretical and practical. In the theoretical component, the evaluation can be done through three tests or a final exam, which represent 60% of the final grade. Each test lasts about 2 hours and corresponds to 20% of the final grade. The tests are taken without consultation and, if the student cannot attend or chooses to opt out of the continuous evaluation model, they may choose to take the exam on a scheduled date, also without consultation. The minimum grade for the theoretical component is 8 points (average of the three tests or the exam grade).
As for the practical component, it consists of compiling the solution to problems proposed during computer practical classes, along with their respective comments and a brief theoretical introduction (about 2 pages) for each chapter. These problems must be submitted in a PDF document at the end of the semester, individually or in groups of two students, and represent 40% of the final grade. If the grade for the work is equal to or higher than 14 and the difference in grades between the practical and theoretical components is greater than 2 points, there will be an oral discussion to automatically adjust the grade of the practical component to 13 or to the difference of 2 points with the tests or exam.
Finally, to attend the course, it is necessary to be present in at least 2/3 of the classes.
Subject matter
Types of data and data analysis strategies. Categorical and continuous variables. Georeferenced information. Location map of samples. Parametric statistical analysis. Most commonly used univariate distribution laws in Earth Sciences variables. Distribution laws (normal, lognormal, uniform). Exploratory data analysis. Univariate analysis: summary measures and graphical representations. Bivariate analysis: correlation measures, contingency tables, and graphical representations. Multivariate analysis: principal component analysis (PCA) and hierarchical classification and k-means. Random variables. Theory of regionalized variables. Some characteristics of regionalized variables. Spatial covariance and variogram. Modeling of experimental variograms. Practice of variography. Kriging estimator. Properties. Deduction of the Kriging system. Kriging variance. Kriging practice: estimation of point and block grids.