A Proxy System Modeling Toolbox for Comparing Water Isotope Observations to Simulations

Abstract:

Simulations which integrate both climate physics and the processes by which climate variations are imprinted in and sampled from paleoclimate archives may facilitate differentiation of the climate signal from random and systematic sources of uncertainty. We simulate the former using a newly developed efficient water-isotope-enabled atmospheric GCM, SPEEDY-IER (Molteni, 2003, Dee et al., submitted), and the latter using a toolbox of proxy system models (PSMs, Evans et al., 2013), synthesized, organized and coded within a self-consistent framework (Dee et al., in prep). SPEEDY-IER is forced with SSTs from the Last Millennium PMIP3 integration of the CCSM4 model (Landrum et al., 2012); relevant climate and isotope variables are extracted from the GCM simulation and used to drive PSMs. Through comparing simulated climate fields to simulated observations, we evaluate the extent to which linear and univariate calibrations on local temperature are valid, given bias in the simulated SST, moisture divergence, and associated isotopic composition of water vapor and precipitation. Taking this a step further, PSMs that incorporate the physical, biological, structural, and time-uncertain aspects of each proxy system help to explicitly quantify the errors accompanying the assumption of linear univariate response of proxy systems to climate forcing.

We demonstrate the utility of the PSM toolbox with an integrative multi-PSM simulation spanning a realistic pan-tropical pacific proxy network of tree-ring cellulose, speleothem, and ice core oxygen isotopic composition (δ¹⁸O). The multi-PSM simulation is used as a testing ground to assess the robustness of frequently invoked teleconnections relating tropical SSTs to terrestrial hydroclimate proxies. By exploring modeled connections between ocean climate and the proxies (both by individual proxy class and for the entire network), we identify which tropical SST signals can be captured by the proxy network, track the individual contribution of sensor, archive, and observational proxy system sub-models to the distortion of the original climate signal, and leverage the PSM framework to investigate why some signals may be filtered out of the final observations.