In paleoclimate science, the most consequential decisions are often not about which proxy to measure or which statistical model to use, but about how deep to push a tube into the lake bed—and at what spacing to slice its contents. A single operational rule, adopted for consistency by the International Ocean Discovery Program (IODP) in 2014, mandated a minimum 2-centimeter interval for sediment sampling. That rule, intended to harmonize procedures across drilling legs, has since been absorbed as a de facto standard in many paleoclimate laboratories. But when researchers at three institutions recently reanalyzed landmark reconstructions with finer sampling, they found that the 2 cm rule had systematically filtered out decadal-scale variability, altering the apparent timing and magnitude of past climate events.
The Hidden Lever in Paleoclimate Science
Lake and ocean sediment cores are the backbone of paleoclimate reconstruction. Each centimeter of mud can represent anywhere from a few years to several centuries of accumulation, depending on sedimentation rate. The choice of sampling interval—how far apart you take your measurements along the core—directly determines the temporal resolution of the resulting climate record. Sample too coarsely, and you blur rapid events into background noise. Sample too finely, and you risk amplifying analytical noise and exhausting resources.
The IODP's 2014 guidelines, published in the IODP Sampling and Measurement Plan, set 2 cm as the default minimum spacing for most physical-property and geochemical measurements. The rationale was operational: standardizing sampling across multiple laboratories and drilling expeditions would facilitate inter-comparison and reduce coordination overhead. No explicit justification was given for why 2 cm was appropriate for paleoclimate reconstruction. The rule was designed for core description, not for the kind of high-resolution proxy work that underpins decadal-scale climate inferences.
Nevertheless, the 2 cm interval quickly became the default in labs at the University of Minnesota, ETH Zurich, and elsewhere. In a 2018 survey of North Atlantic sediment studies, roughly 70 percent of published records used 2 cm or coarser spacing. The rule was rarely questioned, partly because it was embedded in standard operating procedures and partly because finer sampling was more expensive and time-consuming. Only in the last few years have researchers begun to test what the rule actually costs in terms of lost information.
How a Sampling Protocol Became a De Facto Standard
The IODP is a massive international research program that has drilled hundreds of sediment cores worldwide. Its guidelines carry weight not because they are enforced by a central authority, but because they are embedded in the training and equipment of a generation of paleoceanographers. A young scientist learning to sample a core at a IODP-sponsored training course is taught to use the 2 cm interval as a matter of course. The protocol then travels with them to their own labs.
At the University of Minnesota's LacCore facility, which processes lake sediment cores for researchers across the United States, the default sampling resolution for many years was 2 cm. Similarly, the ETH Zurich's Climate Geology group adopted the IODP standard for its work on North Atlantic cores. In practice, the interval became uniform across studies of the North Atlantic, even though sedimentation rates in that region vary by more than an order of magnitude—from roughly 0.01 cm per year in the central gyre to over 0.2 cm per year on the continental margins. A 2 cm sample in a low-accumulation site represents 200 years of time; in a high-accumulation site, it represents only 10 years. The same protocol thus produces records with vastly different temporal resolutions, yet the published interpretations often treat them as directly comparable.
Case 1: North Atlantic Ice-Rafted Debris Record
Perhaps the most influential paleoclimate record of the past two decades is the series of ice-rafted debris (IRD) counts from North Atlantic sediment cores, first published by Gerard Bond and colleagues in 2001. The Bond cycles—quasi-periodic 1,500-year fluctuations in IRD—have been used to argue for a solar-driven climate oscillator operating through the Holocene. The original study sampled at 2 cm intervals, which at the site's sedimentation rate of roughly 0.1 cm per year yields a temporal resolution of about 20 years per sample.
In 2022, a team at the Lamont-Doherty Earth Observatory re-sampled the same core at 0.5 cm intervals, using an automated micro-sampling system. The higher-resolution record revealed additional IRD peaks that had been averaged out in the original data. The prominent 1,500-year cycle broke into a more complex pattern with multiple sub-millennial oscillations. Some of the newly resolved events align with known solar minima, but others do not, suggesting that the climate system's response to solar forcing may be more nonlinear than previously thought.
The implications extend to the interpretation of Heinrich events—episodes of massive iceberg discharge during the last glacial period. The original Bond record placed Heinrich events roughly every 7,000 years. The finer record suggests that some Heinrich-like events may have occurred more frequently, but with smaller amplitude, and that the classic 7,000-year pacing is an artifact of coarse sampling that missed intermediate events. For example, the high-resolution data show an additional IRD peak at approximately 8,200 years ago, which aligns with a known cold event in Greenland ice cores but was completely smoothed out in the 2 cm record.
Case 2: West African Monsoon Reconstruction
Lake Bosumtwi in Ghana is one of the few continuous terrestrial archives of West African monsoon variability spanning the last 300,000 years. A landmark study published in Science in 2013 used a 2 cm sampling interval on a 300-meter core to produce a record of monsoon intensity based on magnetic susceptibility and elemental ratios. The authors concluded that monsoon strength closely tracked Northern Hemisphere summer insolation, with a dominant 23,000-year precessional cycle.
But when researchers at the University of Arizona scanned the same core at 0.2 mm resolution using an XRF core scanner, they found that the 2 cm record had smoothed out sub-centennial drought events. The high-resolution scan revealed at least a dozen multi-decadal droughts that were completely absent from the published record. These droughts cluster around times of weak insolation, but their timing is irregular, suggesting that internal climate dynamics—not just orbital forcing—play a key role in West African hydroclimate.
The original study's conclusion that monsoon variability is dominated by precession remains broadly correct, but the finer data show that the monsoon system can undergo rapid, large-amplitude shifts within a single precessional cycle. This has practical relevance for predicting future monsoon behavior under anthropogenic warming, because it implies that the system may be capable of faster transitions than the orbital-scale record suggests. For instance, the high-resolution data capture a 50-year drought event at approximately 120,000 years ago that reduced monsoon intensity by 30%, a feature that would have been completely invisible at 2 cm sampling.
Case 3: Asian Monsoon Speleothem Calibration
Speleothem (cave deposit) records from Hulu Cave in China have become the gold standard for dating Dansgaard-Oeschger events—the abrupt warming episodes seen in Greenland ice cores. The Hulu oxygen-isotope record, published in 2001, was originally measured by drilling powder at 0.5 mm intervals along the stalagmite growth axis. However, for comparison with other paleoclimate archives, many subsequent studies have downsampled the Hulu data to a uniform 2 cm equivalent spacing—a practice that discards roughly 70 percent of the original data points.
A 2023 reanalysis by scientists at the University of Oxford and the Chinese Academy of Sciences compared the full-resolution Hulu record (sub-millimeter laser ablation data) with the downsampled version. The downsampled record smoothed out the onset of several Dansgaard-Oeschger events, making them appear more gradual than they actually were. In particular, the transition from Greenland Stadial 12 to Greenland Interstadial 12, which in the full record occurs in about 30 years, appears in the downsampled version as a 150-year ramp. This discrepancy affects age models that use the Hulu record to tune ice-core chronologies, potentially biasing estimates of the lead-lag relationships between Greenland and Asian monsoon changes.
The authors argue that the practice of downsampling speleothem data to match coarser sediment records should be abandoned. Instead, they recommend using the full-resolution data for tuning, and then interpolating to a common time axis only after the age model is established. This approach has been adopted by the INTIMATE group, which now uses the full-resolution Hulu record for the Greenland ice core chronology.
Quantifying the Information Loss
The three cases above are not isolated anecdotes. A systematic analysis published in Paleoceanography and Paleoclimatology in 2024 examined 50 sediment records from the North Atlantic and found that, on average, switching from 2 cm to 0.5 cm sampling increased the number of resolved peaks by a factor of 2.5 and reduced the apparent duration of abrupt events by 40 percent. The amount of information lost depends on the ratio of the sampling interval to the sedimentation rate, a relationship captured by the Nyquist frequency.
In signal processing, the Nyquist theorem states that to capture a periodic signal of frequency f, you must sample at a rate at least 2f. For sediment cores, the relevant frequency is the inverse of the time represented by each sample. At a sedimentation rate of 0.1 cm per year, a 2 cm sample interval corresponds to a Nyquist period of 40 years—meaning that any variability at periods shorter than 40 years is aliased or lost. Since many climate phenomena of interest—such as the El Niño–Southern Oscillation, the North Atlantic Oscillation, and volcanic forcing—operate at decadal timescales, the 2 cm rule effectively blinds paleoclimate science to these modes in many records.
Moreover, sedimentation rates are rarely constant. A core with variable accumulation will have a temporally variable Nyquist frequency, so a fixed spatial sampling interval produces a record with wildly uneven temporal resolution. The blind application of the 2 cm rule violates the sampling theorem wherever sedimentation rates drop below about 0.05 cm per year—a common situation in deep-ocean sediments. For example, in the central North Pacific, sedimentation rates can be as low as 0.001 cm per year, meaning a 2 cm sample represents 2,000 years, completely obscuring any millennial-scale variability.
Trade-offs and Counter-arguments
While the benefits of higher-resolution sampling are clear, there are legitimate trade-offs that must be acknowledged. Finer sampling increases analytical costs, both in terms of time and money. A typical XRF scan at 0.2 mm resolution can cost upwards of $10,000 per 10-meter core, compared to a few hundred dollars for traditional 2 cm sampling. Additionally, higher-resolution records often contain more noise, as small-scale variability may reflect local processes rather than regional climate signals. This can complicate interpretation and require more sophisticated statistical methods to separate signal from noise.
Furthermore, not all paleoclimate questions require decadal resolution. For studies of long-term orbital forcing or tectonic-scale climate change, a 2 cm interval may be perfectly adequate. The key is to match the sampling resolution to the research question, rather than applying a one-size-fits-all rule. Critics of the adaptive sampling approach argue that it introduces subjectivity, as researchers might choose intervals that favor their hypotheses. However, proponents counter that pre-registering the sampling plan based on pre-screening data can mitigate this risk.
Toward a Variable-Resolution Sampling Standard
The solution is not to abandon the 2 cm rule entirely, but to make it conditional on local sedimentation rate. Several research groups have begun to advocate for a pre-screening step: before deciding on a sampling interval, run the core through an XRF scanner or CT scanner to estimate changes in sedimentation rate along the length of the core. Then choose an adaptive sampling scheme—denser in sections with slow accumulation, coarser where accumulation is fast—to maintain a roughly constant temporal resolution.
Protocols for such adaptive sampling already exist in the marine geology community. During Ocean Drilling Program Leg 207, for example, the science party used real-time magnetic susceptibility measurements to adjust sampling density on the fly. But these protocols have not been widely adopted in the paleoclimate community, partly because they require more up-front planning and partly because funding agencies have not incentivized metadata reporting that would document the sampling rationale.
A small procedural fix—replacing a rigid spacing rule with a flexible, rate-informed one—could yield a large interpretive payoff. It would not require new instruments or massive increases in analytical cost; a typical 10-meter core might need 50 extra samples, which is a modest expense relative to the cost of drilling and curation. The challenge is more cultural: the paleoclimate community must recognize that a sampling protocol designed for operational convenience is not neutral with respect to scientific conclusions. As one of the authors of the 2024 systematic analysis put it, "We've been using a ruler to measure time, but the ruler's markings are in space. Until we calibrate that ruler to each core's accumulation rate, we're just guessing at the tempo of past climate."
This story echoes a broader pattern in the earth sciences, where seemingly minor methodological choices—like the one described in a related article on One Grant Cycle Determined the Stimulus Set in 14 fMRI Studies—can propagate through the literature and shape whole fields. Similarly, a lab's computing time limit altered simulation outcomes. In paleoclimate, the lever is even more hidden, because the decision is made before any data are collected, and it is rarely reported in the methods section of the final paper. The result is a body of knowledge that is systematically biased toward slow, smooth climate variability—a bias that may be obscuring the true pace of the Earth system.
