Assimilation of Time-Averaged Pseudoproxies for Climate Reconstruction

Abstract

The efficacy of a novel ensemble data assimilation (DA) technique is examined in the climate field reconstruction (CFR) of surface temperature. A minimalistic, computationally inexpensive DA technique is employed that requires only a static ensemble of climatologically plausible states. Pseudoproxy experiments are performed with both general circulation model (GCM) and Twentieth Century Reanalysis (20CR) data by reconstructing surface temperature fields from a sparse network of noisy pseudoproxies. The DA approach is compared to a conventional CFR approach based on principal component analysis (PCA) for experiments on global domains. DA outperforms PCA in reconstructing global-mean temperature in all experiments and is more consistent across experiments, with a range of time series correlations of 0.69–0.94 compared to 0.19–0.87 for the PCA method. DA improvements are even more evident in spatial reconstruction skill, especially in sparsely sampled pseudoproxy regions and for 20CR experiments. It is hypothesized that DA improves spatial reconstructions because it relies on coherent, spatially local temperature patterns, which remain robust even when glacial states are used to reconstruct nonglacial states and vice versa. These local relationships, as utilized by DA, appear to be more robust than the orthogonal patterns of variability utilized by PCA. Comparing results for GCM and 20CR data indicates that pseudoproxy experiments that rely solely on GCM data may give a false impression of reconstruction skill.

On the utility of proxy system models for estimating climate states over the common era

“…The specific DA implementation used in this study follows closely that of Steiger et al [2014]. Briefly, it uses an off-line approach, wherein the prior ensemble, x b , consists of annually averaged climate states drawn from a climate model simulation; the ensemble is not integrated forward in time because of the massive computational constraints involved and because online DA for paleoclimate reconstructions appears to provide little improvement in skill over off-line DA, at least for atmospheric variables [Matsikaris et al, 2015].…”

“…Briefly, it uses an off-line approach, wherein the prior ensemble, x b , consists of annually averaged climate states drawn from a climate model simulation; the ensemble is not integrated forward in time because of the massive computational constraints involved and because online DA for paleoclimate reconstructions appears to provide little improvement in skill over off-line DA, at least for atmospheric variables [Matsikaris et al, 2015]. The most important difference from Steiger et al [2014] is that we use PSMs for generating y and for computing Hðx b Þ (in most cases). This provides a more realistic reconstruction framework than that typically employed in pseudoproxy experiments.…”

“…period [e.g., Goosse et al, 2006Goosse et al, , 2010Widmann et al, 2010;Steiger et al, 2014]. One central aim of this paper is to scrutinize the implications of this assumption for climate reconstruction purposes.…”

“…We first investigate the utility of PSMs for use in paleoclimate data assimilation [Steiger et al, 2014;Hakim et al, 2016]. We explore this within a synthetic, pseudoproxy framework [see Smerdon, 2012, for a review] because it allows for a controlled test bed for both the PSMs and the reconstruction technique.…”

Multi-timescale data assimilation for atmosphere–ocean state estimates

“…For the update calculations we employ an ensemble square-root Kalman filter with serial observation processing, applied to time averages (see Steiger et al, 2014, for a detailed DA algorithm and fuller discussion of DA terminology). We extend the technique of Dirren and Hakim (2005), Huntley and Hakim (2010), and Steiger et al (2014) by iteratively applying the state-update equations across multiple timescales by leveraging the serial observation processing approach to the Kalman filter (Houtekamer and Mitchell, 2001).…”