A pseudoproxy emulation of the PAGES 2k database using a hierarchy of proxy system models

Abstract

Paleoclimate reconstructions are now integral to climate assessments, yet the consequences of using different methodologies and proxy data require rigorous benchmarking. Pseudoproxy experiments (PPEs) provide a tractable and transparent test bed for evaluating climate reconstruction methods and their sensitivity to aspects of real-world proxy networks. Here we develop a dataset that leverages proxy system models (PSMs) for this purpose, which emulates the essential physical, chemical, biological, and geological processes that translate climate signals into proxy records, making these synthetic proxies more relevant to the real world. We apply a suite of PSMs to emulate the widely-used PAGES 2k dataset, including realistic spatiotemporal sampling and error structure. A hierarchical approach allows us to produce many variants of this base dataset, isolating the impact of sampling bias in time and space, representation error, sampling error, and other assumptions. Combining these various experiments produces a rich dataset (“pseudoPAGES2k”) for many applications. As an illustration, we show how to conduct a PPE with this dataset based on emerging climate field reconstruction techniques.

Revisiting a Constraint on Equilibrium Climate Sensitivity From a Last Millennium Perspective

“…This caveat contributes to the uncertainties in the emergent constraint derived from the reconstructed GMST statistics. We investigate these possibilities using pseudoproxy experiments (Figure S5 in Supporting Information ; F. Zhu et al., 2023), which suggest that there would be great value in increasing the quality and spatial coverage of paleoclimate proxies over the Common Era, perhaps according to optimal sampling protocols (Comboul et al., 2015).…”

cfr (v2024.1.26): a Python package for climate field reconstruction

“…The psm module incorporates classes for multiple popular proxy system models (PSMs; Evans et al, 2013), including a univariate linear regression model (Hakim et al, 2016;Tardif et al, 2019) (Linear), a bivariate linear regression model (Hakim et al, 2016;Tardif et al, 2019) (Bilinear), a tree-ring width model VS-Lite (Tolwinski-Ward et al, 2011Zhu and Tolwinski-Ward, 2023) (VSLite), a simple lake varve thickness model (Zhu et al, 2023a) (Lake_VarveThickness), the ice core δ 18 O model (Ice_d18O), coral δ 18 O model (Coral_d18O), and coral Sr/Ca ratio model (Coral_SrCa) adopted from PRYSM (Dee et al, 2015), among which Linear and Bilinear are commonly used in paleoclimate data assimilation (PDA), and others are more useful to generate pseu-doproxy emulations (Zhu et al, 2023a, b). A summary of available PSMs in cfr is listed in Table 5.…”

Separating Common Signal From Proxy Noise in Tree Rings

“…All spectra were computed using the multi-taper spectral estimate, and spectral slopes (ß) were calculated as the linear relationship between frequency and power on a log-log scale to describe the shape of the spectrum (Thomson, 1982). We note here that our signal-to-noise ratio estimation is based on chronology variance as in Fisher et al (1985) rather than standard deviation, which has been used in some studies (Mann et al, 2007;Smeardon, 2012;Zhu et al, 2023). This is because power spectral density is expressed in variance per frequency unit, and is thus more common in signal processing (Jenkins & Priestley, 1957).…”

“…Our results address the question of how to correctly define the structure of the noise term in pseudoproxy experiments and data assimilation frameworks where proxy timeseries are integrated with climate models (Dee et al, 2017;Jones et al, 2009;Smeardon, 2012;Steiger et al, 2014). In past such experiments, a variety of different noise models have been used ranging from blue (i.e., noise that diminishes with timescale) (Mann & Rutherford, 2002;Mann et al, 2007), to white (i.e., no relationship to timescale) (Lee et al, 2008;Mann et al, 2005;Von Storch et al, 2004) to red (noise that increases with timescale) (Von Storch et al, 2009;Zhu et al, 2023). Here, we estimate the structure of proxy noise directly from the data and show that noise in tree-ring width records resembles a power-law (slope of 0.8), and in tree-ring density records exhibit a less-steep positive noise spectrum (slope of 0.5-0.6).…”