High-frequency climate forcing causes prolonged cold periods in the Holocene

The study addresses whether the Holocene, particularly the mid-to late-Holocene, was climatically stable or punctuated by prolonged cold periods. While the last 800 years are relatively well understood and clearly influenced by volcanic forcing (with pronounced events such as the Little Ice Age and mid-sixth century cooling), proxy scarcity and biases earlier in the Common Era and farther back into the Holocene limit reliable reconstructions of high-frequency variability. Multi-proxy reconstructions across the Holocene often feature low resolution, age uncertainties, seasonal and spatial sampling biases, and thus tend to under-represent interannual–centennial variability and the occurrence of LIA-like events. The research question is to determine the role of high-frequency external forcings—especially volcanic eruptions—in driving annual to centennial climate variability over the past 8000 years, and whether such forcings can explain multi-decadal to multi-centennial cold periods previously underrepresented in proxy compilations.

- Last millennium: Robust agreement exists between proxy-based reconstructions and climate models for the CE, with volcanic eruptions recognized as a primary driver of pre-industrial variability, including the Little Ice Age. High-resolution proxies (e.g., tree rings) and models identify extremely cold years/decades and major events (mid-sixth century, LIA), with documented societal impacts. Mechanistic studies emphasize atmosphere–ocean–sea-ice interactions sustaining post-eruption cooling. - Holocene-scale reconstructions: Holocene multi-proxy datasets (e.g., Temperature12k) often lack the resolution and precise dating to capture high-frequency variability, leading to the perception of relative climatic stability and an absence of LIA-like global events. Regional high-resolution records increasingly indicate multi-decadal to multi-centennial cold periods across the Northern Hemisphere, attributed to volcanic/solar forcing, meltwater inputs, or AMOC changes. Prior Holocene modeling/assimilation efforts seldom included realistic high-frequency volcanic forcing; earlier volcanic reconstructions relied on limited ice cores with synchronization and dating issues. - Solar forcing: Grand solar minima have been implicated in some cold periods, but several studies argue solar variability alone is too weak to generate persistent multi-centennial cooling without additional feedbacks or forcings. - Holocene conundrum: Discrepancies between modeled and reconstructed long-term Holocene temperature trends are debated (spatial/seasonal proxy biases vs. model/forcing limitations), but this work focuses on sub-centennial variability, not resolving the conundrum directly.

- Model: Max Planck Institute Earth System Model (MPI-ESM1.2). Atmosphere (ECHAM6) at T63 (~1.9°×1.9°), 47 vertical levels up to ~80 km. Ocean (MPIOM) nominal 1.5° horizontal resolution with 40 vertical levels. - Experiments: Two transient simulations spanning 6000 BCE to 1850 CE. 1) Orbital + GHG run: Forced with orbital variations and greenhouse gas concentrations. 2) All forcing run: Includes orbital, GHG, plus land-use changes (from 850 CE), stratospheric ozone (varying with solar), solar irradiance, and prescribed volcanic aerosols. - Forcing datasets: - Volcanic: HolVol1.0 for 6000 BCE–20 BCE; eVolv2k for 20 BCE–1850 CE, derived from bipolar ice-core sulfur deposition. Easy Volcanic Aerosol (EVA) generator to translate sulfur injections to spectrally resolved aerosol properties. Stratospheric aerosol optical depth at 550 nm (AODNH) used as forcing metric (NH mean, similar to global mean). - Solar irradiance: New reconstruction (varies ozone accordingly). - Land use: Implemented from 850 CE (affects only last millennium). - Event classification and detection: - Large eruptions: Annual AODNH peak > 0.08 (Tambora ~0.26). - Double eruptions: Two large eruptions within 10 years; classified distinctly from single events. - Multiyear cold periods (annual scale): Defined as periods with >2 consecutive years with detrended NH annual mean temperature below 2σ or 3σ thresholds, where σ is from the detrended orbital+GHG control. First year of each event identified. - Multi-centennial events: 200-year running mean NH temperature anomalies (detrended) analyzed against 200-year accumulated AODNH to identify long-lasting cold periods (significant on 2σ level). - Statistical approach: - Detrending: Fit and remove a second-degree polynomial from NH annual mean temperature. - Significance: 2σ and 3σ thresholds derived from similarly filtered orbital+GHG control. Events exceeding thresholds assessed as significant. - Correlation: Pearson correlation between temperature anomalies and AODNH for various filtered series; significance assessed by t-test at 95% confidence. - Diagnostics of mechanisms: Evaluate co-variations of NH air temperature, upper ocean heat content, Arctic sea-ice extent, ocean heat transport at 60°N, and North Atlantic subpolar gyre strength to infer ocean–sea-ice feedback processes. - Model–proxy comparison: Compare simulated high-latitude (60°–90°N) summer temperatures (annual and 100-year means) with Temperature12k reconstructions (mean and 5–95th percentile).

- High-frequency forcing impact: The all forcing run exhibits frequent 1–8 year cold anomalies absent in the orbital+GHG control. Identified 48 multiyear cold periods (annual scale) in the all forcing run linked to large eruptions versus only 6 in the orbital+GHG run at the 2σ level. - Forcing–response relationships: - Annual/decadal scale: Strong negative correlation between temperature and volcanic forcing (AODNH), r ≈ -0.6; no significant correlation with solar irradiance (r ≈ 0.06). Volcanic forcing magnitude exceeds solar except during major grand minima (e.g., Maunder: volcanic ~ -0.32 W m^-2; solar ~ -0.23 W m^-2). - Decadal accumulation: Cooling duration scales with 10-year accumulated AODNH; stronger decadal AODNH yields longer cooling, especially for 3σ events. - Double eruptions: Large double events (AODNH peak > 0.08 within 10 years) produce up to ~10 years of significant NH cooling vs. ~5 years for comparable single events. The 536/540 CE double event stands out as one of the largest Holocene decadal temperature perturbations. The strongest modeled decadal NH cooling occurs around 5230 BCE, associated with the very large 5229 BCE eruption. - Multi-centennial cold periods: Eleven NH long-lasting (200-year scale) cold periods identified over the past 8000 years, ten of which coincide with the highest accumulated AODNH intervals. The 200-year mean temperature anomaly and accumulated AODNH correlate strongly (r = -0.73, p < 0.01). The three strongest events occur early in the analyzed mid-Holocene interval. The LALIA ranks 3rd and the LIA 5th in 200-year cooling magnitude. The strongest multi-centennial anomaly centers around 3895 BCE (5.9 ka), with a 200-year mean NH cooling up to -0.18 K. - Event catalog (examples): Periods include the 7.6 ka, 6.6 ka, 6.4 ka, 5.9 ka (strongest), 5.1 ka, 3.6 ka, 2.6 ka, 2.4 ka, 2.1 ka events, plus LALIA (520–720 CE) and LIA (1640–1840 CE). Many periods featured numerous eruptions larger than Pinatubo and notable events (e.g., Mazama, Aniakchak II, Okmok, Tambora, Laki, Cosigüina). - Mechanisms: Long-lasting cold periods coincide with decreased global upper-ocean heat content, increased Arctic sea-ice extent, reduced ocean heat transport at 60°N, and a weakened North Atlantic subpolar gyre, consistent with ocean–sea-ice feedbacks sustaining cooling after volcanic clusters. - Model–proxy comparison: Simulated high-latitude summer temperatures agree with an overall decreasing Holocene trend seen in Temperature12k, with 200-year means near the lower bound of proxy uncertainty. The model reveals additional NH multi-centennial cold periods not captured by sparse or seasonally biased proxies. - Recurrence and regional RCC: Average recurrence of long-lasting cold periods is ~700 years, with a Holocene Quiet Period (3000–2000 BCE) lacking elevated volcanic activity. Rapid climate change (RCC) transitions (e.g., Roman Warm Period to LALIA; 4130–3900 BCE) show strong regional amplitudes (up to ~0.8 °C) in the Barents–Kara Seas, highlighting sea-ice feedbacks and potential societal relevance. - Overall: High-frequency volcanic forcing, especially clustered eruptions, is essential to reproduce the amplitude and timing of Holocene sub-centennial to multi-centennial temperature variability, challenging the notion of Holocene climatic stability in global proxy compilations.

The simulations demonstrate that the mid-to late-Holocene was punctuated by numerous prolonged cold intervals driven primarily by volcanic activity, contrary to the stability suggested by low-resolution global proxy compilations. The strong negative correlation between accumulated AODNH and 200-year temperature anomalies and the distinct impact of double eruptions substantiate volcano-cluster-induced prolonged cooling maintained by ocean–sea-ice feedbacks. Solar forcing can add to cooling during grand minima but is generally too weak to generate multi-centennial cold periods by itself. These results reconcile regional evidence of Holocene cold periods with physically consistent mechanisms and quantify the temporal structure (frequency, duration, and intensity) of such events. The findings underscore limitations in traditional multi-proxy reconstructions—chiefly temporal resolution, dating, and seasonal biases—that likely obscure high-frequency variability and under-detect LIA-like events. The documented recurrence (~700 years) and identified RCC transitions provide context for evaluating potential societal impacts, consistent with historical cases (e.g., LALIA, LIA, Okmok II, Aniakchak II) and archaeological inferences of social change during cool intervals. The study advances understanding of Holocene variability, emphasizing the necessity of realistic high-frequency forcing in models to assess past climate risk and to evaluate model performance for future projections.

Using a fully coupled Earth system model with updated high-resolution volcanic and solar forcing, the study identifies 11 multi-centennial Northern Hemisphere cold periods over the past 8000 years. The integrated effect of volcanic forcing, amplified and sustained by ocean–sea-ice feedbacks, explains all identified long-lasting cold periods in the simulation, while solar forcing contributes additively during some grand minima but is generally secondary. The results overturn the notion of a climatically stable mid-to late-Holocene, reveal the critical role of eruption clustering and double events in prolonging cooling, and provide mechanistic insights into associated ocean–cryosphere adjustments. Model–proxy comparisons indicate that proxy reconstructions may under-represent high-frequency variability due to resolution and sampling biases, highlighting the need to update chronologies, improve age models, and expand spatial/seasonal coverage (especially SH and boreal winter). Future work should include ensembles and additional simulations to probe background-state dependence and to quantify uncertainties in forcing reconstructions and feedback strengths.

- The simulations begin in the mid-Holocene and exclude ice-sheet and freshwater forcing, limiting insights into early Holocene dynamics and the broader Holocene temperature conundrum. - Only a single long transient realization is presented; additional simulations/ensembles are needed to assess internal variability and background-state dependence of responses to high-frequency forcing. - Focus is on the Northern Hemisphere due to proxy limitations; Southern Hemisphere responses are not comprehensively assessed. - Land-use forcing is only included from 850 CE, affecting just the last millennium. - Volcanic forcing reconstructions, while state-of-the-art and extended, still carry uncertainties in magnitude, spatial distribution, and timing that can influence simulated responses. - Proxy-model comparisons are constrained by seasonal/spatial proxy biases and low temporal resolution in many records, complicating direct validation of high-frequency variability.