Revisiting Online and Offline Data Assimilation Comparison for Paleoclimate Reconstruction: An Idealized OSSE Study

Abstract

Knowledge of past climate conditions is crucial to understanding the climate system. Recently, data assimilation (DA) has been applied to reconstruct paleoclimate of the last millennium (e.g.,

An Analog Offline EnKF for Paleoclimate Data Assimilation

“…The default ensemble size is 100, which is consistent with that in Hakim et al (2016). The Covariance localization is used to localize the impact of observations and mitigate spurious error correlations between observations and state variables resulted from a limited ensemble size (e.g., Okazaki et al, 2021;Valler et al, 2019). The Gaspari and Cohn (GC; Gaspari & Cohn, 1999) function that is a fifth-order piecewise polynomial function is used for localization.…”

How Stable Water Isotope Measurements and Modeling Can Help Bridge the Gap between Research on Weather and Climate Time Scales

“…Second, advances in data assimilation now allow paleoclimate data to be assimilated into paleoclimate simulations (Tierney et al 2020;Okazaki et al 2021). The use of isotopic records in data assimilation is made more powerful by advances in proxy system forward-modeling approaches (Okazaki and Yoshimura 2019) and by the recent development of isotopic data compilations (e.g., Comas-Bru et al 2020).…”

A Hybrid Gain Analog Offline EnKF for Paleoclimate Data Assimilation

“…Based on an intermediate complexity general circulation model, Okazaki et al. (2021) showed that the online PDA could outperform the offline PDA when the length of predictability is longer than the averaging time of observations, and the ocean plays an important role to extend predictability that is helpful for online PDA. Although online PDA is feasible with a simplified model or a linear inverse model (e.g., Jebri & Khodri, 2023; Perkins & Hakim, 2017), online PDA could require intensive computational cost for the paleoclimate simulations with a realistic and complex model, due to the length of simulations.…”

“…Furthermore, a linear inverse model of coupled atmosphere-ocean dynamics was applied for online PDA, and the reconstructed annual-resolution atmosphere and ocean fields demonstrated the added value of dynamical memory for the proxy data and physical constraints from the atmosphere-ocean interaction (Perkins & Hakim, 2021). Based on an intermediate complexity general circulation model, Okazaki et al (2021) showed that the online PDA could outperform the offline PDA when the length of predictability is longer than the averaging time of observations, and the ocean plays an important role to extend predictability that is helpful for online PDA. Although online PDA is feasible with a simplified model or a linear inverse model (e.g., Jebri & Khodri, 2023;Perkins & Hakim, 2017), online PDA could require intensive computational cost for the paleoclimate simulations with a realistic and complex model, due to the length of simulations.…”