I have used this exercise occasionally for my graduate course in Quaternary Environments.

The students must have some skill with Excel or other spreadsheet software, including importing data, cleaning data sets, formulas, and basic statistics. We have already discussed the strengths and weaknesses of ice core data (closure time, freeze-thaw, dating methods...) by the time they undertake this activity.

The activity is effectively a lab assignment, although they work both in and out of lab on it. It is about 1/3 of the way into the course, after discussion in class and text of ice cores.

This exercise has several content goals. The first is to understand the nature of an actual data set, warts and all. At a broader scale, it is to visualize the variance in a long data set and conceptualize partitioning that variance (trend, cycle, noise...). At the broadest scale, when we discuss the outcomes of each of the students or groups working with first the same, then different ice core records, it is to understand the nature of both the short- and long-term variability in the ice core record—cycles, storm-tracks, storm events...

Each student or group must demonstrate an understanding of the system they are studying. They "know" about Milankovitch cycles, and that ice core data are one of the archives of that data, but they need to think on a noisier scale. The ice core data are still being parsed, and our institutional understanding still has gaps.

This exercise provides a context in which to work on complex spreadsheeting skills, including interpolation of data, offsetting data to determine correlation lags/leads, and graphical display.

Students access the ice core data archived at Lamont-Doherty Geological Observatory. They select a core (Greenland, Antarctica, Quelcaya), pose a working hypothesis regarding the data, import the data in an Excel-readable format, and examine the data to determine correlations between variables and cause/effect as recorded in leads and lags. They generate a written and graphical analysis of the data and, in the next lab period, discuss the similarities and differences among their group outputs in terms of demonstrated correlations, assumptions required, effects of latitude, and any other item that arises.

The students submit their final Excel file and write-up. Because they are free to seek relationships among a number of variables, there is no "right answer". Excellent submissions are well thought out, well illustrated, and well written. They include multiple explanations and probably hierarchy of those explanations. In contrast, poor submissions lack sophisticated illustrations, draw over-simplified conclusions, and propose single answers for complex responses. [As a graduate course with small enrollment and teams of two working, the responses are almost always of high standard.]

The data are archived at the International Research Institute for Climate and Society - http://ingrid.ldgo.columbia.edu/SOURCES/.ICE/.CORE/.
Note that many other data sets are also accessible through the Archive home page - http://iridl.ldeo.columbia.edu/ (more info) .