Ingo Bethke

We present a new paleo reanalysis based on an adaptation of the Norwegian Climate Prediction Model (NorCPM), in which the Norwegian Earth System Model is equipped with an ensemble Kalman filter to assimilate hundreds of annually resolved proxy records, including tree rings, corals, and ice cores, over the last 500 years. First, an offline 30-member reanalysis is produced. This is followed by a dynamically coupled 10-member reanalysis obtained through atmospheric wind nudging from the offline reconstruction, allowing the ocean to respond to reconstructed atmospheric circulation variability. This approach enables an explicit representation of ocean dynamics and associated atmosphere-ocean feedback, which are typically indirectly represented in existing multi-century reanalyses. The offline reanalysis successfully reproduces large-scale atmospheric variability and hydroclimate variability over the 20th century across the tropics, mid-latitudes, and polar regions, and compares well with existing paleo reconstructions. The wind-nudged ensemble further captures ocean variability and exhibits more pronounced multi-decadal variability, highlighting the role of wind-driven ocean adjustment in shaping low-frequency climate variability. Challenges remain, particularly in the Southern Ocean where the wind-nudged ensemble shows strong sensitivity to wind variability. Nevertheless, these results demonstrate a promising pathway toward fully coupled ocean-atmosphere reanalyses spanning the past centuries, opening new opportunities to investigate the mechanisms underlying low-frequency variability and coupled feedback.

Atmospheric reanalyses extending before 1979, including ERA5, show a substantial cold bias over Antarctica during 1958–1978. Because Southern Ocean sea surface conditions (SSCs) are poorly constrained in this period, we assess their role using two ensembles of a global atmospheric model forced by different SSC datasets. The first uses the widely employed HadISST product, while the second is based on a recent reconstruction that assimilates continental observations. Our results show that SSC differences strongly alter the Antarctic surface climate, but explain only about 30 % of the ERA5 cold bias ERA5 in our model experiments, indicating that additional factors contribute to this bias.

Upper-level zonal wind is a robust indicator for predicting episodes of consecutive dry days on subseasonal timescales in Norway. To construct a predictor that facilitates the predictive modeling of consecutive dry days, we define a regime-based index by projecting instantaneous upper-level flow anomalies onto three dominant circulation regimes identified using an agglomerative clustering algorithm. Next, we evaluate the skill of Integrated Forecast System (IFS) in predicting the regimes.Results show that the raw probabilistic forecasts of the index are systematically over-dispered. To address this deficiency, we apply a Bayesian recalibration framework that combines a prior distribution derived from observed climatology with a likelihood function that represents the conditional dependence of observations on the ensemble mean. The resulting posterior distribution serves as the calibrated probabilistic forecast.This modeling framework offers two key advantages: (1)Intrinsic Recalibration: when the forecast contains no usable predictive information and the likelihood converges toward the model climatology, the posterior automatically reverts to the observed climatology, ensuring well-calibrated probabilistic forecasts even in the absence of raw forecast skill (2) Sampling from complete samples: Bayesian method infers the posterior distribution efficiently from a shorter reforecast-reanalysis join sample (two times per week, 20 years), and a longer reliable reanalysis data (ERA5, ~40 years). This mitigates sampling limitations commonly encountered in non-Bayesian calibration methods, which rely solely on the joint forecast–observation sample and can therefore produce suboptimal distributional estimates, particularly when hindcast datasets are short.

Instrumental observations only capture a short interval of the climate history of the Earth, and are insufficient to fully constrain low-frequency variability, internal dynamics, and the response of the climate system to changing background states. Paleoclimate archives, by contrast, document a wide range of past climate changes, yet translating this information into Earth System Models (ESMs) to enhance their performance and future projections remains a major challenge. Paleoclimate Data Assimilation (PDA) provides a promising pathway to bridge this gap by combining proxy records and ESMs within a physically consistent framework. Here we present a paleo reanalysis based on an adaptation of the Norwegian Climate Prediction Model (NorCPM), in which an ensemble Kalman filter is used to assimilate hundreds of annually resolved proxy records (including coral, tree-ring, and ice-core records) back to 1600 CE. Unlike many existing paleo reanalyses for past centuries, which primarily constrain the atmospheric state and only indirectly represent ocean variability, our approach explicitly accounts for ocean dynamics. By nudging three dimensional atmospheric wind fields derived from the paleo atmospheric reanalysis, we generate a dynamically consistent coupled reanalysis, by simulating the response of the ocean to large-scale wind variability, as well as their associated impacts through thermodynamical feedbacks. The climate reanalysis represents both forced and internal variability over the last four centuries and shows good agreement with independent instrumental observations. By construction, this approach yields a dynamically coherent, multivariate reconstruction that goes beyond traditional proxy reconstructions and enables direct investigation of climate dynamics and feedback. Here, we focus on methodological aspects and perspectives of PDA, highlighting how paleo reanalyses can (i) constrain modes of low-frequency variability and their stability across different climate states, and (ii) evaluate and refine the calibration of the ESMs beyond the instrumental period. Such approaches are essential for improving confidence in future climate projections, particularly with respect to long-timescale variability, feedback, and the potential for abrupt transitions in the Earth system.

The ocean has absorbed approximately 30% of global carbon dioxide emissions since the pre-industrial era, thereby mitigating climate change. The uptake of atmospheric carbon depends on coupled changes in ocean circulation, atmospheric CO2 forcing, and ocean biogeochemical processes, and exhibits pronounced variability on interannual to decadal timescales. However, the scarcity of in situ observations makes it difficult to robustly quantify the magnitude, temporal variability, and drivers of oceanic carbon uptake over the period of 1960-2020.Using multiple observation-based data products together with numerical experiments based on the Community Earth System Model version 2 (CESM2), we examine the variability of air-sea carbon fluxes on interannual to decadal timescales. Consistent with previous studies, we find that externally forced variability driven by anthropogenic emissions, along with natural variability associated with climate modes such as the El Niño–Southern Oscillation (ENSO) and the Interdecadal Pacific Oscillation (IPO), are key contributors to carbon-flux variability over the past decades. Earth System Models including the CESM2 tend to underestimate the magnitude of this variability because they struggle to capture complex physical and biogeochemical processes as well as inter-decadal climate variability. Assimilating realistic anomalies in observed surface winds improves the representation of variability in the equatorial ocean, enhancing the model’s ability to reproduce observed changes in ocean carbon uptake since the 1960s. Additional assimilation of observed ocean temperature and salinity anomalies further improves extra-tropical variability, although model performance degrades in the equatorial regions.Based on these assimilated simulations, we demonstrate how a large-scale climate mode in the Pacific led to a redistribution of both natural and anthropogenic dissolved inorganic carbon in the global ocean, accompanied by a slowdown of ocean carbon sink during the 1990s (the so-called carbon-sink hiatus). Finally, we discuss the inferred decadal variability of the land carbon sink, estimated by incorporating the newly constrained ocean carbon sink into the global carbon budget.

Climate models are commonly run at resolutions too coarse to resolve mesoscale ocean dynamics, and therefore lack oceanic eddies and fronts that strongly influence air-sea exchange. This leads to an underestimation of the ocean’s role in driving atmosphere–ocean interactions in western boundary current regions, with implications for simulated climate variability and change. We explore whether the effects of mesoscale sea surface temperature (SST) features on large-scale circulation can be represented in a standard resolution climate model using a partially coupled “pacemaker” configuration of the Norwegian Earth System Model version 2 (NorESM2). The setup introduces mesoscale SST features from a high-resolution (0.125°) ocean into the standard-resolution coupled model grid (1° ocean and atmosphere). Focusing on the Kuroshio Current, we find that mesoscale SST features amplify ocean-to-atmosphere turbulent heat fluxes, as expected, and also produce notable free tropospheric responses (a robust local strengthening of the North Pacific storm track at low levels and a poleward shift aloft). The results offer a proof of concept that 1° climate models can capture the broader climate impacts of small-scale oceanic variability without explicitly resolving it, opening promising pathways to improve predictions and projections.

Atmospheric reanalyses extending before 1979, including ERA5, show a substantial cold bias over Antarctica during 1958–1978. Because Southern Ocean sea surface conditions (SSCs) are poorly constrained in this period, we assess their role using two ensembles of a global atmospheric model forced by different SSC datasets. The first uses the widely employed HadISST product, while the second is based on a recent reconstruction that assimilates continental observations. Our results show that SSC differences strongly alter the Antarctic surface climate, but explain only about 30 % of the ERA5 cold bias ERA5, indicating that additional factors contribute to this bias.

Atmospheric reanalyses extending before 1979, including ERA5, show a substantial cold bias over Antarctica during 1958–1978. Because Southern Ocean sea surface conditions (SSCs) are poorly constrained in this period, we assess their role using two ensembles of a global atmospheric model forced by different SSC datasets. The first uses the widely employed HadISST product, while the second is based on a recent reconstruction that assimilates continental observations. Our results show that SSC differences strongly alter the Antarctic surface climate, but explain only about 30 % of the ERA5 cold bias ERA5, indicating that additional factors contribute to this bias.

Climate models are commonly run at resolutions that do not represent mesoscale ocean dynamics and therefore lack the oceanic eddies and fronts that strongly influence atmosphere–ocean exchange. As a result, these models tend to underestimate the ocean’s role in driving atmosphere–ocean interactions in western boundary current regions, with implications for simulated climate variability and change (Stevens & Bony, 2013; Hewitt et al., 2022). Here, we investigate whether the effects of mesoscale sea surface temperature (SST) features (those associated with eddies and fronts) on large-scale circulation can be represented in a standard resolution climate model. We develop a partially coupled “pacemaker” configuration of the Norwegian Earth System Model version 2 (NorESM2) that allows us to introduce mesoscale SST features from a high-resolution (0.125°) ocean in the standard coupled model grid (1° in both ocean and atmosphere). Focusing on the Kuroshio Current, this amplifies the ocean-to-atmosphere turbulent heat fluxes in the Kuroshio region, as expected, but also produces notable effects in the free troposphere. Even with a 1° atmosphere, mesoscale SST features induce a robust local strengthening of the North Pacific storm track at low levels and a poleward shift aloft. The results demonstrate that, although typical climate models cannot resolve mesoscale dynamics, there is clear potential to incorporate their effects. With the growing body of observational products for atmosphere-ocean exchange (Tang et al., 2024), developing more realistic parameterizations of mesoscale-driven SST variability to improve the types of climate models used for future projections appears promising.