Pacific decadal variability over the last 2000 years and implications for climatic risk

The Pacific decadal variability (PDV) significantly influences various climate impacts across the Pacific Basin, including drought, flood, wildfire, tropical cyclone risk, and global temperature. Recent research also links PDV to substantial changes in high-latitude Southern Hemisphere and Antarctic climate, such as abrupt sea ice extent declines and regional Antarctic surface air temperature fluctuations. These studies primarily rely on short-term (mostly satellite-era) high-latitude observations, potentially misrepresenting long-term variability. Understanding PDV's behavior is crucial for both regional climate outcomes affecting Pacific Basin nations and the observed abrupt changes in the Antarctic ice sheet and sea ice. The relative contributions of internally generated processes versus external forcings (atmospheric, radiative, volcanic, and anthropogenic) in generating decadal variability, including modes like the IPO and AMO, remain debated. Regardless of the underlying cause, both observed and paleoclimate data often assume that basin-wide PDV manifests as positive and negative phases of similar length (around 20–30 years). However, an underlying, internally generated process causing decadal-scale oscillations like the IPO in paleoclimate studies hasn't been definitively identified. Recent work suggests that multi-decadal volcanic forcing might trigger the AMO, based on long observations and climate simulations. This raises questions about the role of internal mechanisms initiating decadal-scale modes like the IPO and AMO, prompting further investigation using paleoclimate reconstructions. Observational IPO indices have too few phase changes to definitively determine the driving mechanism of the IPO – whether internal (deep ocean) modulation, stochastic (atmospheric) forcing, or another mechanism. Determining the frequency and distribution of IPO phases is essential for understanding Pacific state transitions, associated climate risks, and reconciling climate projections with natural variability. Multi-centennial, annually resolved reconstructions of PDV are needed to achieve this. A recent meta-analysis of existing reconstructions suggests that observation-era IPO phase lengths are similar to those from reconstructions spanning recent centuries, but phase lengths/prevalence might have differed before the past 400–500 years. However, limited availability of longer reconstructions hampered detailed analysis. This study leverages new ice core data to improve and extend a previous IPO reconstruction derived from Law Dome ice cores in East Antarctica. The refined reconstruction extends the record by 1000 years, spanning the Common Era (CE –10 to 2016) at annual resolution. The same reconstruction methods are employed, demonstrating high skill against observational data. While the reconstruction exhibits a small, statistically significant trend, this is attributed to more frequent/longer negative phases in recent centuries and is not detrended. A statistical treatment of the detrended data confirms the robustness of the findings.

Extensive research has explored the relationship between Pacific decadal variability (PDV) and various climate impacts. Studies have linked PDV to changes in Pacific Basin drought, flood, wildfire, and tropical cyclone risk, as well as global temperature patterns. More recently, PDV's role in influencing high-latitude Southern Hemisphere and Antarctic climate has been investigated, demonstrating its impact on sea ice extent and regional Antarctic surface air temperature. However, most of this work relies on relatively short observational datasets, particularly from the satellite era, potentially limiting their ability to capture long-term variability. The mechanisms driving decadal variability, including the roles of internal and external forcings, remain a subject of ongoing debate. Studies have proposed competing hypotheses, such as internally generated oscillations versus external forcings like volcanic eruptions and anthropogenic aerosols influencing the Interdecadal Pacific Oscillation (IPO) and the Atlantic Multidecadal Oscillation (AMO). While some research suggests volcanic forcing may play a significant role in multi-decadal climate oscillations, the contribution of internal variability to IPO phase changes remains uncertain. Existing observational indices of the IPO are limited by the short duration of the data record, hindering the precise determination of the IPO's driving mechanism and the frequency of different phases. The need for multi-centennial, annually resolved reconstructions of PDV has been consistently emphasized, aiming to provide a more comprehensive understanding of long-term variability and its impact on climate projections. Previous studies using various reconstruction methods have produced varying results, often with disagreements on the lengths and prevalences of positive and negative phases.

This study refines and extends a previous IPO reconstruction from Law Dome ice cores in East Antarctica, significantly increasing the temporal coverage. The new reconstruction uses annually resolved data spanning the Common Era (CE 1 to 2016), doubling the previous reconstruction's length. The analysis incorporates improved annual dating techniques and incorporates higher-resolution samples to reduce the risk of inhomogeneity. The reconstruction methods used include a piecewise linear fit (PLF) method and a novel Gaussian kernel correlation reconstruction method, which is robust to missing or unevenly sampled data. The Gaussian kernel correlation method is used to reduce the impact of missing or unevenly sampled data, creating an ensemble of 2000 potential reconstructions. Input datasets for the reconstruction were the log-transformed warm and cool-season average sea salt concentrations, and the annual snowfall accumulation rate. The reconstruction methods have been rigorously tested against observational data to ensure reliability and accuracy. Data validation involved testing for skill and determining the optimal number of basis functions based on root mean square error and generalized cross-validation scores, ultimately selecting nine basis functions for PLF and producing an ensemble of 2000 reconstructions for the Gaussian kernel correlation method. Spectral and coherence analyses were conducted using the MTM toolkit to examine the frequency characteristics of the reconstructed IPO and assess its relationship with other reconstructions. The study compares the reconstructed IPO with the observed Parker et al. IPO index and the Buckley et al. IPO reconstruction, facilitating a comprehensive analysis of PDV across different datasets. The study further explores the implications of this understanding of PDV for eastern Australia's hydroclimate. The analysis includes testing for differences in rainfall distributions between positive and negative IPO phases, positive and neutral phases, and neutral and negative phases to better understand the impact of different phase lengths and frequencies on rainfall and runoff variability. This analysis utilizes both regional and station-level rainfall data sourced from the Australian Bureau of Meteorology's high-quality station network.

The 2000-year reconstruction of the Interdecadal Pacific Oscillation (IPO) reveals a substantially different picture of Pacific decadal variability than previously understood from shorter observational records. The key findings include: 1. **Asymmetry of IPO Phases:** Contrary to the common assumption, positive phases of the IPO significantly outnumber negative phases. Negative phases are much shorter (5-7 years) and occur less frequently (10-14%) than positive phases (29-51%). This challenges the prevailing notion of roughly equal durations for positive and negative phases. 2. **Predominance of Neutral-Positive State:** The IPO exhibits a predominantly neutral-positive state over the last 2000 years, with infrequent, short-lived excursions into negative phases. The long-term average state of the Pacific is neutral-positive. 3. **Implications for Eastern Australia:** Eastern Australia's rainfall is significantly lower during positive IPO years than negative IPO years. However, the rainfall during positive and neutral phases is statistically indistinguishable. Therefore, focusing on the neutral-positive phase is crucial for accurately assessing long-term hydroclimatic risk. 4. **Hydroclimatic Risk Re-evaluation:** The infrequent and short duration of negative IPO phases necessitates a re-evaluation of climate risk assessments. The assumption of roughly equal positive and negative phase durations is incorrect. Drought risk in eastern Australia, for instance, is essentially the same for positive phases as for the overall neutral-positive state. This has significant implications for water resource management and planning. 5. **Unusual Nature of Mid-20th Century:** The significant negative phase of the IPO during the mid-20th century appears highly unusual in the context of the last 2000 years. The observational period is not representative of the typical long-term variability. 6. **Rainfall Reduction:** Annual rainfall during neutral-positive IPO years is significantly lower than during negative phases across all domains of eastern Australia (p < 0.01–0.08). Rainfall reduction from negative to positive phases or negative to neutral-positive ranges from 6-14% for various regions of eastern Australia. This translates to a 3–4 times greater decrease (6–16%) in annual runoff/reservoir inflows due to the non-linear relationship between rainfall and runoff.

This study's findings significantly challenge the prevailing understanding of Pacific decadal variability and its associated climate risks. The substantial asymmetry between positive and negative IPO phases, with the predominance of a neutral-positive state, necessitates a recalibration of climate risk assessments. The over-reliance on relatively short observational records has likely led to a biased view of IPO behavior, overemphasizing the influence of negative phases. The results highlight the importance of utilizing longer-term paleoclimate reconstructions to better understand long-term climate variability and accurately assess climate risks. The case study of eastern Australia demonstrates the significant implications of this revised understanding for water resource management. The findings suggest that current water infrastructure planning and management strategies, based largely on the recent observational record, may be inadequately prepared for the long-term predominance of neutral-positive IPO conditions. The substantial reduction in annual runoff and reservoir inflows associated with neutral-positive phases poses a considerable threat to water security in eastern Australia. The unusual nature of the mid-20th-century negative IPO phase also warrants further investigation. This exceptionally prolonged negative phase may represent an anomaly or possibly a shift towards a new regime with more frequent and/or longer negative phases, necessitating distinct risk management approaches.

This study offers a substantially revised understanding of Pacific decadal variability (PDV) based on a 2000-year reconstruction of the Interdecadal Pacific Oscillation (IPO). The results demonstrate the predominance of a neutral-positive state, punctuated by infrequent and short negative phases, which differs greatly from the interpretation derived from shorter, more recent observational data. This new understanding has profound implications for climate risk assessments across the Pacific Basin, particularly concerning the long-term prevalence of neutral-positive conditions and their implications for water resources in regions such as eastern Australia. Future research should focus on understanding the mechanisms driving the initiation and termination of negative IPO phases, potentially involving internally generated variability, external forcings (volcanic and/or anthropogenic), and their potential interactions. The unusual length and frequency of negative phases in recent centuries relative to the past two millennia also merits investigation to determine whether this is a temporary anomaly or signals a shift towards a new climate regime.

While this study significantly extends the understanding of IPO behavior, several limitations exist. The reconstruction relies on proxy data from Antarctic ice cores, which may not perfectly capture all aspects of Pacific-wide climate variability. The analysis focuses on the IPO index as a measure of PDV, while other indices exist and might yield different perspectives. The case study on eastern Australia provides valuable insights but further regional studies are needed to fully appreciate the global implications of this work. Also, the reconstruction itself has limitations, and uncertainty is incorporated into the statistical treatment of the results.