Vorlesung/Übung: Advanced Geostatistical Methods (07426 / 07427)
Do.: 14:00-16:00, K 4 (NW II) video conference; Do.: 18:00-20:00, CIP-Pool S 24a (GEO), and; Fr.: 08:00-12:00, CIP-Pool S 24a (GEO), irregularly, for dates see details; Augsburg
Christoph Beck, Jucundus Jacobeit
V, 1 ECTS, per video conference (bi-weekly)
Ü, 2 ECTS, as block course
14.05.2009 Prof. Jacobeit in Bayreuth
Dates for the Exercise:
24.07. (14-20) and 25.07. (9-15) final session in Augsburg
This course consists of two parts: lecture and exercise. The latter will be conducted by Christoph Beck (see separate comments from him), the former by Jucundus Jacobeit with the following contents:
Advanced methods in Geostatistics comprise multivariate methods (in contrast to univariate and bivariate ones as basic methods), and they will be discussed with special reference to spatial analyses being important for climate research. An example would be the statistical downscaling mentioned in the context of regional climate modelling during the winter term lecture. Generally, a basic introduction into the following methods will be performed:
- Principal Component Analysis (PCA), Common Factor Analysis CFA) and Empirical Orthogonal Function (EOF) Analysis
- Multiple Correlation and Regression Analysis
- Canonical Correlation Analysis (CCA) and Redundancy Analysis
- Cluster Analysis
- Discriminant Analysis (multiple-groups-multiple-variables)
Corresponding applications will be performed during the exercise part.
The exercise part of Advanced Geostatistical Methods will focus on selected methods introduced in the Lecture conducted by Jucundus Jacobeit.
The R software environment for statistical computing will be used to apply these statistical analyses to climatological data sets.
Exercises will comprise (besides an introduction to the R system) the application of Principal Component Analysis and Cluster Analysis for determining large-scale atmospheric circulation variations (Major teleconnection patterns, Circulation types).
In a subsequent step statistical models linking large-scale circulation dynamics and regional-scale climate variations will be developed using Multiple Regression Analyses and Canonical Correlation Analysis .
Statistical methods will be applied to circulation and climate data from the North Atlantic European domain in order to investigate the relationship between large-scale circulation and Central European drought variability during the 2nd half of the 20th century.
Zu dieser Veranstaltung stehen weitere Unterlagen zum Download bereit...
<< Zurück zu Semester-Übersicht