THE EFFECT OF TIME-AVERAGING ON PALEO-WATER ISOTOPE RECORDS
First, different archives integrate amounts of time ranging from years to thousands of years. Archives that have high resolution but short duration (e.g., mollusks, mammal teeth) are ideally suited to reconstructing short-term (seasonal to decadal) phenomena. In contrast, archives that form over larger amounts of time (e.g., soil carbonates, leaf waxes, volcanic glasses) do not capture that short-term variability, instead recording the mean state of the climate over millennia. This time-averaging behavior makes longer-term records insensitive to large-amplitude, short-term variability (e.g., storms, seasonal cycle, ENSO, sunspot cycle) and thus better suited to reconstructing changes in climate over million year time scales. Variability that appears in paleoclimate archives must be due to processes operating over time scales greater than the interval of time-averaging; in longer-term records, this typically means >10 kyr.
Second, sedimentary sequences must be sampled at a resolution sufficient to avoid undersampling the geochemical record. Undersampling can lead to missing changes in climate as well as the spurious inference of such changes. We introduce a statistical test to evaluate undersampling in sedimentary sequences for an archive with any given interval of time-averaging. The implication of this observation is that field studies that sample at a resolution too low to track climatic trends are essentially wasted effort, because they are unlikely to contribute to increasing our understanding of paleoclimate or paleoaltimetry.