ASSIGNMENTS AND GRADE DETERMINATION

General Comment

Although some derivations (not proofs) are given in the notes and some of these will be discussed in class, you will not be expected to learn these nor repeat them on quizzes. There will be strong emphasis on using software to actually analyze data sets. There will be opportunities for programming but in general you will not need to do any programming in the course. The emphasis will be on the use of free software.

I. LITERATURE REVIEWS

Each student will find four different articles in the literature and write a short review of each article (4-5 pages, wordprocessed, no handwritten or typed reviews). A zerox copy of the paper should be attached to each review, be sure to give the full citation (title, author, journal, year, volume, pages)in the review for the article being reviewed

The objective in having you do these reviews is for you to become acquainted with some papers appearing in the literature of your field/discipline which have used geostatistical tools

The review should include the following: (1) A brief description of the data used in the reported research, (2) A brief description of the objective of the research, (3) A brief discussion of any problems encountered in the research and/or significant results obtained with the research, (4) A personal evaluation of the readability of the paper and the significance of the results

You do not have to explain the theory in the paper nor in general repeat any of it.

You may choose to review a theoretical paper but in general it is assumed that you will review papers that report using geostatistics rather than ones primarily focused on developing new theory

You may choose papers from any professional journal (but published no earlier than 1995)

A. One review should be on a paper using (univariate) Kriging (Ordinary or Simple, Point or Block)

This review will be due on 9 February

B. One review should be on a paper using Cokriging

This review will be due on 9 March

C. One review should be on a paper using Simulation, e.g., sequential gaussian

This review will be due on 13 April

D. One review should be on a paper using a non-linear transformation in connection with Kriging or Simulation, e.g., logarithmic, indicator, normal score

This review will be due on 8 May

You may substitute a review of a paper using Bayesian geostatistics in lieu of any one of the four topics listed above

II. SHORT TAKE HOME QUIZZES

There will be three of these.

QUIZ 1, given out on 15 February, due on 22 February

QUIZ 2, to be given out on 22 March, due on 29 MarchQUIZ 3, to be given out on 27 April, due on 3 May

Basically these are to ensure that you have read the relevant material and have some understanding and grasp of it.

III. DATA SET ANALYSIS

Each student will need a data set, it should consist of values for at least two different variables at a minimum of 75 data locations (in 2 or 3 dimensional space). This can be thought of as a table with a minimum of four columns, two for coordinates and two for the variables, minimum of 75 rows. I.e., as in a spreadsheet. Ideally this would be data that you might use in your research or at least similar data.

Email me a copy of the data set by 9 February .

You will be asked to analyze the data in various ways and then at the end of the semester you will submit a report showing the results of your analyses together with a discussion of your results.

Assuming that you use one of the R packages for your analyses, generate a history of the commands, functions, outputs, etc. to be included in your report. Incorporate the various graphs and plots into your report

Initial Analyses

            Generate a data file in an appropriate format (ASCII)

            Read the data in the program(s) you will use

            For each of the two variables, compute the sample mean and the sample variance

            Plot histograms for each of the two variables

            Generate a coded plot for each variable, i.e., a plot of the data locations with each plotted point coded (by symbol type and/or symbol color representing the data value at the plotted point)

            Fit a quadratic trend surface to the data for each variable, report the coefficients for the terms in the polynomial.

            Compute and plot sample variograms for each variable (both omni-directional and directional using the directions 0, 45, 90, 135 degrees)

             Fit a variogram model to the sample variograms in at least two ways, e.g. by "eye", by least squares, by ML or REML. Report both models in your report

             Use cross-validation to evaluate the fit of both of the models obtained for both variables

Construct a grid object covering the region of your data locations and using either Ordinary or Universal kriging (which ever is appropriate for your data) to interpolate one of the two variables. Generate a plot showing the results

Fit a a LCM to the sample variograms/sample cross variogram for the two variables. Apply Ordinary or Universal cokriging to interpolate the variables. You may use the same grid as for kriging. Generate plots showing the results

For one of the two variables, simulate 4 realizations. You may use the same grid as for kriging/cokriging. For each realization generate a plot showing the results

Be sure to label all plots and tables. List and identify all variogram, LCM models

Summarize your results, list any problems or difficulties.