Modellierung dynamischer und räumlicher Prozesse
Winter Semester 2008/2009
- Two marks will be given: one for the lecture part, one for the exercises
- The lecture mark will be the result of the final test; the final test will
be held at 3.2.09.
- The exercise mark will be averaged from hand-in exercises (50%), and
a final assignment/mini-project (50%)
- Not handing in a final assignment means not passing the exercises
- Maximally two exercises may be missed during the semester
- Details about the assignment/mini project are found below.
Lecture slides are found here; the file is updated
shortly before each lecture.
Details (when/where) will follow.
The exercises use the open source statistical environment R (an implementation of the S
language for data analysis). You can find introductory/tutorial material
through its respective web sites.
Exercise and model input are found here
If needed, these two books can be borrowed from me.
- C. Chatfield, The analysis of time series: an introduction. Chapman
and Hall: chapters 1, 2 and 3
Applied Spatial Data Analysis with R, by R. Bivand, E. Pebesma and V. Gomez-Rubio
Springer, New York:
- Ch 1, 2, 3
- Ch 4, 5, 6 (whatever is convenient from it)
- Ch 8 (geostatistics)
Data and scripts
- As an introduction to R, you could go through the first 6 chapters (up to "lists
and data frames") of An Introduction to R: go to the R home page, click Manuals
under Documentation, and open the Introduction. You can copy and paste
commands to an R session, started in the CIP pool.
- course excercises: html, pdf.
- The French meteo data: meteo data, and R script
- The classic Irish wind data
set; an script analyzing it, the original paper on it, and a more recent one.
Assignment and test
A test is planned on 3.2.09, and will cover the material treated in the
lectures and exercises. The questions will not be of the kind "how do
I do this with R", but rather refer to the modelling itself.
The assignment will be a written report of max 5 pages (regular
fonts/margins/page size, including figures and/or tables) on a modelling
topic taken from the list with suggestions below, or otherwise approved
by me. The minimum requirements are (i) data should be analyzed, and
(ii) the analysis should explicitly address spatial variation, temporal
variation, or spatio-temporal variation.
The written report should include an introduction, a central research
question, a description of the data, a description of the analysis
and the results, and concluding remarks answering the central research
question. Write in scientific style. It is allowed to do the research
in couples, but the written report should be made individually. Hand-in
deadline is Feb 27, 12:00. German is allowed, English is encouraged
(but does not guarantee extra points). If you have used R, please attach
the R script used as appendix (additional to the max. 5 pages).
For each topic: take care that you do not compute Euclidian distances
based on long/lat coordinates. See the KML example in the exercises on
how to (re)project data.
- Analyze meteorological variables (temperature, rainfall) in a spatial or
- Analyze the Irish wind data given above; focus on one of the research
questions in the Haslett and Raftery 1989 paper, or on a different research
- Analyze air quality data in relation to EU regulations; air quality data can
be obtained from the UBA
- Analyze the sediment pollution data in R package gstat
- Try to find an optimal sampling strategy for spatial data (e.g. for the tull
data in package gstat) or for a temporal sampling problem.