HR: 1330h
AN: PP52A-0952    [PDF]
TI: Examination of Long-Term Proxy Stationarity Using AOGCM Output: AD 1000-1999
AU: * Wahl, E
EM: wahl@ucar.edu
AF: National Center for Atmospheric Research, 3450 Mitchell Lane, Boulder, CO 80301
AU: Ammann, C
EM: ammann@ucar.edu
AF: National Center for Atmospheric Research, 3450 Mitchell Lane, Boulder, CO 80301
AB: The stationarity of proxy-based climate reconstructions beyond their modern calibration/verification periods is examined using AOGCM output to construct simulated "proxies", whose performance can be compared to "real" (within-model) climate variables over long time spans. Transient monthly output of the NCAR Paleo-CSM 1.4 model for the period 1000-1999 is employed to examine a set of such proxies for seven locations on the Pacific Rim whose modern climates are varyingly teleconnected to ENSO. These areas and model were chosen because: a) they contain real-world proxy records used in ENSO reconstructions; b) their "within-model" relationships with ENSO over 1871-1999 are close to those in instrumental data; and c) Paleo-CSM represents reasonable ENSO amplitudes. Simulated proxies were constructed by adding white noise to the teleconnected variable at each site sufficient to degrade the correlation between the climate variable and the noise-added time series to that observed in real-world proxies at the site. Results show that an r-square above 0.6 between a simulated proxy and the real teleconnected climate variable it mimics is needed in order for a NINO3 reconstruction based on the proxy to systematically track the variability of the real NINO3/climate teleconnection. As the proxy/climate r-square drops below 0.5, the correlation between the proxy-based NINO3 reconstructions and real NINO3 has significant reverse-trend excursions in relation to the underlying correlation between real NINO3 and the teleconnected variable mimicked by the proxy. However, even at r-square values as low as 0.34, the longest of these excursions are relatively short (50 years), although amplitude mismatches become more pronounced. (Results are based on examination of Monte Carlo experiments, since adding random noise generates a unique transient path for the simulated proxy for each replicate--any of which could reflect the random noise realized in the real-world proxy.) The results also show an underlying correlation between real NINO3 and a teleconnected climate variable greater than 0.4 during the instrumental period appears to be necessary for this relationship (and its proxy-based reconstruction) to remain statistically significant over the entire 1000-year experiment.
DE: 0300 ATMOSPHERIC COMPOSITION AND STRUCTURE
DE: 0370 Volcanic effects (8409)
DE: 1620 Climate dynamics (3309)
DE: 1650 Solar variability
DE: 3344 Paleoclimatology
SC: Paleoceanography and Paleoclimatology [PP]
MN: 2003 Fall Meeting