Quaternary Environments
Spring, 2002

Ice Core Exercise

Ice cores are one of the best terrestrial repositories of environmental data, including as much as 400,000 years of interpretable paleoenvironmental information.  Some of this information is available to you in a near-raw format!  What I would like you to do is to investigate some of these data.

The data are archived at and downloadable from http://ingrid.ldgo.columbia.edu/SOURCES/.ICE/.CORE/ .

What I would like you to do is to investigate relationships between any two variables accessible from that source.  Note that Quelccaya is a short core with little to compare to the longer records.  The GRIP core spans from 8 ky to 40 ky BP and includes methane and oxygen isotope data.  The Vostok data spans the last 160,000 years and lists 8 values (including inferred temperature difference from present).  So - what interests you?  The similarity between Greenland and Antarctic oxygen isotope compositions?  Whether methane or CO2 is a better predictor of climate?  Whether climate lags CO2 or vice-versa?  Please investigate one such issue, by next Tuesday at 5 PM, in singles or groups of two, following the methodology below:

• Write your question (perhaps in the form of a testable hypothesis!)
• Download the data you need to answer your question.  The preferred format is .tsv, which is readily importable into Excel, Quattro, Lotus, or another spreadsheet.
• Import your data.  If it doesn't parse into separate columns automatically, use Excel's Data...Text to Columns... capability, or equivalent, to generate one or more columns of data for each variable.
• Transform your data to a common scale.  That scale could be depth or age in a single core, but must be age between cores.  In order to compare data sets, you must have an array of numbers, with no gaps, in the form Z, var1, var2, where Z is depth or age and var1 and var2 are your variables of interest.  [This is the hard part of this exercise.]
• Correlate your data.  Use the statistical analytical tools available to you (e.g., Excel's Tools...Data Analysis..., Correlation) to answer your question.  To test the presence of a lag or lead simply shift one variable column up or down a row at a time, calculating correlation, to see if the correlation improves or worsens.  Do so enough steps to demonstrate worsening to insignificance.
  • Potential problem: if the Data Analysis package has not yet been invoked on your copy of Excel, it may not show up in the Tools menu.  Use Excel Help to find the instructions to Install the Analysis Toolpak - an add-in that should be lying dormant within the software.
• Submit a short (<2 page) write-up describing your question, your approach, your outcome, and the statistics that support it.