Holliday et al.'s (2023) Gish Gallop: timing of the Younger-Dryas onset and Greenland platinum spike
The location of several Greenland ice cores, from Steffensen et al. (2008).
Over the next few months/year I'll post a regular series of blog posts that deconstruct Holliday et al.'s (2023) (henceforth, just Holliday et al.) "comprehensive refutation" of the YDIH. Some common themes will become apparent. Essentially, there are no refutation arguments at all in it, so their title is misleading. Remember, the meaning of 'refute' is to 'disprove', but there are no arguments in Holliday et al. that disprove the YDIH. It's also worth looking up the 'Gish gallop' fallacy, because that is what is happening here. Holliday et al.'s paper might be extremely long, but it is lacking in serious arguments, and there are certainly no refutations.
First up in this blog post is the timing of the YD climate change (specifically its onset) and the Greenland platinum spike. The relevant sections of Holliday et al. are 3.3 and 5.1. My comments are in italics.
3.3. Problematic Chronologic and Paleoclimatic Assumptions
Four assumptions are implicit in the YDIH:
1) that the climatological and environmental changes at the beginning of the YD/GS-1 (i.e., YDB) are
globally synchronous,
2) that the ages of the direct “impact indicators” are identical to that of the hypothetical impact ;
3) that the indicators are consistent in sign and magnitude with the “Impact Winter” scenario proposed
by Wolbach et al. (2018a,b); and
4) that all of those environmental changes are unique or so exceptional in a longer-term context that a
singular, as opposed to general, explanation for them is required.
1-3 are fine (provided no impact material went into Earth orbit and landed much later), but 4 is vague and needs some clarification.
In Firestone et al. (2007) we read in the abstract "We propose that one or more large, low-density ET objects exploded over northern North America, partially destabilizing the Laurentide Ice Sheet and triggering YD cooling ". In the text we further read in the section on climate "A number of impact-related effects most likely contributed to the abrupt, major cooling at the onset of the YD and its maintenance for ~1,000 years. Cooling mechanisms operating on shorter time scales may have included (i) ozone depletion, causing shifts in atmospheric systems in response to cooling, with the side-effect of allowing increased deadly UV radiation to reach survivors on the surface (46); (ii) atmospheric injection of nitrogen compounds (NOx), sulfates, dust, soot, and other toxic chemicals from the impact and widespread wildfires (46), all of which may have led to cooling by blockage of sunlight, with the side-effect of diminished photosynthesis for plants and increased chemical toxicity for animals and plants (46); and (iii) injection of large amounts of water vapor and ice into the upper atmosphere to form persistent cloudiness and noctilucent clouds, leading to reduced sunlight and surface cooling (46). Although these cooling mechanisms tend to be short-lived, they can trigger longer-term consequences through feedback mechanisms. For example, noctilucent clouds can reduce solar insolation at high latitudes, increasing snow accumulation and causing further cooling in a feedback loop. The largest potential effect would have been impact-related partial destabilization and/or melting of the ice sheet. In the short term, this would have suddenly released meltwater and rafts of icebergs into the North Atlantic and Arctic Oceans, lowering surface-ocean salinity with consequent surface cooling. The longer-term cooling effects largely would have resulted from the consequent weakening of thermohaline circulation in the northern Atlantic (54), sustaining YD cooling for ~1,000 years".
Furthermore, in Wolbach et al. (2018a) we read in the abstract "The ice record is consistent with YDB impact theory that extensive impact-related biomass burning triggered the abrupt onset of an impact winter, which led, through climatic feedbacks, to the anomalous YD climate episode." In the text we further read "The impact event destabilized the ice-sheet margins, causing extensive iceberg calving into the Arctic and North Atlantic Oceans (Bond and Lotti 1995; Kennett et al. 2018). The airburst/impacts collapsed multiple ice dams of proglacial lakes along the ice-sheet margins, producing extensive meltwater flooding into the Arctic and North Atlantic Oceans (Teller 2013; see Kennett et al. 2018 for summary and references). Destabilization of the ice sheet also may have triggered extensive subglacial ice-sheet flooding, leaving widespread, flood-related landforms across large parts of Canada (Shaw 2002). The massive outflow of proglacial lake waters, ice-sheet meltwater, and icebergs into the Arctic and North Atlantic Oceans caused rerouting of oceanic thermohaline circulation. Through climatic feedbacks, this, in turn, led to the YD cool episode (Broecker 1997; Teller 2013; Kennett et al. 2018). Unlike for typical warm-to cold climate transitions, global sea levels rose up to 2–4 m within a few decades or less at the YD onset, as recorded in coral reefs in the Atlantic and Pacific Oceans (Bard et al. 2010; Kennett et al. 2018). Multiple impact-related drivers caused warm interglacial temperatures to abruptly fall to cold, near-glacial levels within less than a year (Steffensen et al. 2008), possibly in as little as 3 mo (Manchester and Patterson 2008)."
Meanwhile in Wolbach et al. (2018b) we read in the abstract "Thus, existing evidence indicates that the YDB impact event caused an anomalously large episode of biomass burning, resulting in extensive atmospheric soot/dust loading that triggered an “impact winter.” This, in turn, triggered abrupt YD cooling and other climate changes, reinforced by climatic feedback mechanisms, including Arctic sea ice expansion, rerouting of North American continental runoff, and subsequent ocean circulation changes." In their text we further read "Thus, the negative effects of AC/soot might have persisted for 6 wk or more at the YD onset, blocking all sunlight and causing rapid cooling. Reduced insolation is also expected from the injection of comet dust to the upper atmosphere, as discussed in part 1 (Wolbach et al. 2018). If so, the lack of sunlight would have had widespread and catastrophic biotic effects, including insufficient light for plant photosynthesis and growth. At the same time, North Atlantic deepwater formation ceased, thus throttling the so-called ocean conveyor and triggering a sustained decrease in near-global temperatures. The changed state of oceanic circulation in the North Atlantic maintained YD cold temperatures for ∼1400 y, until the system reverted to its previous state (Broecker 1997; Kennett et al. 2018)."
Clearly, the YDIH proposes the usual AMOC-weakening mechanism was triggered by the impact via de-stabilisation of northern ice-sheets. Moreover, this effect was synchronous with a brief impact winter of unknown but significant magnitude at the YD onset. One of the aims of YDIH research is discover the magnitude of this latter effect. So the YDIH is consistent with the AMOC-weakening mechanism and can only be differentiated from it at short-timescales while the impact winter is acting; long-timescales will be dominated by the AMOC-weakening process. Thus, high-resolution paleoclimate data is needed that focusses on the onset of the YD.
If any supposition were shown to be false, that would discredit the YDIH.
However, we don't know how common ET impact triggers are, so we need to be careful. They could be much more common than Holliday et al. suppose.
The first three assumptions jointly propose that the climatological effects generated by a
purported impact produced synchronous changes in other environmental systems. Furthermore,
these changes occurred in a direction consistent with that predicted by the YDIH, i.e., the
“Impact Winter” i.e., “a sustained decrease in near-global temperatures” that occurred at the
beginning of the YD/GS-1 (Wolbach et al., 2018b). However, this did not occur (see also Section 13.7).
No evidence is provided here for this latter statement. Which is odd, considering this section is where such evidence should be found. Moreover, this statement is presented as fact, which is uncommon in scientific literature. How do they know it did not happen? Normally, such statements are qualified.
The notion of synchroneity is central to the YDIH: “If an ET event caused the YD, then
within the limits of dating precision, the YDB will have the same age everywhere. If on the
contrary, different YDB sites have different ages and especially if those ages spread over a
significant amount of time, that would falsify the claim of an instantaneous event” (Powell 2022,
p 27). However, it is important to note that while an ET impact might be presumed to produce
globally synchronous environmental changes, it is not the only mechanism that could do so. In
particular, the AMOC/ACC hypothesis described in Section 2 also predicts that the climate
changes in the North Atlantic region will be registered globally, without an appreciable lag.
The “Impact Winter” scenario holds that at the YDB “The radiant and thermal energy
from multiple explosions triggered wildfires that burned ∼10% of the planet’s biomass,
producing charcoal peaks in lake/marine cores that are among the highest in 368,000 y”
(Wolbach et al., 2018a, p. 179). Wolbach et al. propose that the smoke from those fires “might
have persisted for 6 wk or more at the YD onset, blocking all sunlight and causing rapid
cooling.” The reduced insolation “would have had widespread and catastrophic biotic effects,
including insufficient light for plant photosynthesis and growth” (Wolbach et al., 2018b, p 200).
They also propose that “The impact event destabilized the ice-sheet margins, causing extensive
iceberg calving into the Arctic and North Atlantic Oceans” which “collapsed multiple ice dams
of proglacial lakes along the ice-sheet margins, producing extensive meltwater flooding into the
Arctic and North Atlantic Oceans” (Wolbach et al., 2018a, p. 179). Note that except for the
“impact event” feature, this latter proposal is consistent with elements of the AMOC/ACC
hypothesis; see Section 2). By using analogies with nuclear war and other ET impact scenarios,
Wolbach et al. (2018a,b) clearly propose that the impact and its immediate effects occurred on
timescales too short to resolve using standard radiocarbon dating.
Exactly. The YD period is expected to be similar to other AMOC-related climate change events except for a brief (sub-annual?) impact winter at the YD onset. Thus, low-resolution paleoclimate data is not expected to distinguish them. Only high-resolution data at the YD onset is expected to show any unusual climate change effects.
An additional complication in assessing synchroneity of environmental
changes at the YDB using radiocarbon dating arises from the presence of an “age plateau” in
radiocarbon calibration curves at the YDB (see Section 5.1). This has the effect of compressing a range
of calendar ages into a shorter span of radiocarbon ages, which can contribute to a greater sense of
synchroneity among radiocarbon ages than is real.
This is not evidence against the YDIH, but is useful information.
On the global scale, there is also substantial evidence for synchroneity of the YDB, but
with some relatively short (decadal) lags in the expression of the accompanying climatic changes
between the Greenland ice-core records and high-temporal-resolution records elsewhere (Cheng
et al., 2020; Nakagawa et al., 2021). For example, recent comparisons of the Greenland ice-core
records with a global network of U-Th dated speleothem records (Cheng et al., 2020), and the
annually laminated sedimentary record of Lake Suigetsu, Japan (Nakagawa et al., 2021), strongly
support the notion of global synchroneity of the onset of the YD/GS-1. None of the records
examined by Cheng et al., or Nakagawa et al. depend on conventional radiocarbon dating, and
both incidentally also support the use of the Greenland ice-core chronology, i.e., GICC05, as the
“master” one.
It must be remembered that there is estimated a ~ 140 yr uncertainty in counting ice layers up to the YD onset in the GICC05 chronology. Therefore, this statement is misleading, as the speleothem and other records cannot reliably distinguish between different the ice-core chronologies. Also, no evidence is provided against the YDIH here.
Both studies admit the possibility of regional lags in the full expression (as
opposed to the initiation) of the climatic response to North Atlantic-focused abrupt climate
changes and relate those to large-scale atmospheric and oceanic circulation mechanisms. Both
studies also propose and test hypotheses about those linkages that invoke elements of the
AMOC/ACC hypothesis (Cheng et al., 2020, p. 23414; Nakagawa et al., 2021, their sec. 5.4). In
general, paleoclimatic and paleoenvironmental records do support the notion of a globally
synchronous beginning of the YD/GS-1, but as is the case of the latter two examples, an impact
is neither required nor invoked, and Cheng et al. (2020) explicitly consider and reject it.
The speleothem records in Cheng et al. have insufficient resolution to reject the possibility of an initial, brief impact winter. In fact, their analysis is probably flawed, and their data are fully consistent with the YDIH. We will see this later.
Above is part of Figure 11 from Nakagawa et al. (2021) - the highest resolution plot in their paper focussed on the YD onset (Suigetsu lake varve in blue, NGRIP ice core deuterium excess in green). For Suigetsu, the YD onset is estimated from the start of the ramp function to be 12,845.5 +- 44.5 BP on the IntCal20 radiocarbon scale. For NGRIP, the YD onset is estimated from the deuterium excess measurement to be 12,846 +- 4 BP on the GICC05 scale. So these records are apparently in agreement. However, the timescales of these plots cannot be compared with much confidence since the uncertainty in the GICC05 chronology is around 140 years. Thus, the data in Nakagawa et al. (2021) cannot be used to refute the YDIH.
Larsson et al. (2022), by supplementing the usual dating methods applied to European
paleoenvironmental records with tephrochronology (to chronologically align records), were
unable to support a long-standing hypothesis that the YD/GS-1 climate reversal was
asynchronous across Europe on a regional scale. Similarly, Reinig et al. (2021), while focusing
on the determining the age of the Laacher See eruption, noted that “Our study demonstrates that
the Greenland GI-1–GS-1 transition coincided with the European AL/YD cooling. The temporal
match between Greenland ice core and central European climate proxies suggests that the last
major Northern Hemisphere cooling interval before the Holocene was initiated and steered by an
abrupt climate system change that instantly affected the whole North Atlantic region.” Likewise,
Engels et al. (2022), examined the vegetation change across Europe at the YDB, and concluded
that it “shows instant and, within decadal scale dating uncertainty, synchronous response of the
terrestrial plant community to Late-Glacial climate change across northwest Europe.”
Consequently, the first assumption of global synchrony in the beginning of the YD/GS-1 is
probably true.
As expected of the YDIH, so not evidence against the YDIH.
As to the second assumption, synchroneity in the response of impact indictors, the Impact
Winter scenario places strong demands on the quality of local chronologies for establishing
synchroneity among purported YDIH indicators (especially the direct impact indicators), and
with global records. An environmental or biotic change that is simply in the chronological
neighborhood of the beginning of the YD/GS-1 does not automatically imply that the
environmental change supports the YDIH. It must be exactly synchronous with or slightly postdate
the YDB, but not occur before. This situation also implies that where available, absolute
chronologies produced by annual layer-counting or radiometric dating methods not subject to
calibration errors should be preferred when discussing questions of timing or synchroneity. The second
assumption, synchroneity of impact indicators, is unlikely to be supported in all cases
(see Section 5 for a fuller discussion of dating issues among purported YDIH impact indicators).
Again, no evidence against the YDIH is provided here. The relevant part of section 5 is discussed later.
The third assumption is that the climate changes at the beginning of the YD/GS-1 were
consistent in sign, pattern, and magnitude with the “Impact Winder scenario”. Climate-model
simulations of nuclear war scenarios offer insight into the spatial and temporal expression of an
impact winter. One of the more robust features of such simulations is the nearly global
uniformity in the sign of the climate change: decreases in surface air temperature nearly
everywhere and in all seasons (except some scattered high-latitude regions, related to
stratospheric warming), persisting for several years (Robock et al., 2007a; 2007b). The
hypothesized YDB impact should inject significant particulate in the atmosphere, and
simulations of the response to the widespread injections of aerosols by volcanism or
geoengineering solutions (Zhao et al., 2021) also show uniformly negative changes in surface air
temperature.
Wolbach et al. propose a briefer impact winter than "several years". Still, no evidence against the YDIH is presented.
Fastovich et al. (2020, abstract) used a variety of paleovegetation indicators to show that
“YD cooling was pronounced in the northeastern United States and muted in the north central
United States. Florida sites warmed during the YD, while other southeastern sites maintained a
relatively stable climate.” Further illustrating the complexities of climate during and before the
YDC, Griggs et al. (2022) show that climate in the eastern Great Lakes during LGIT was antiphased
to the stereotypical warmer B-A/GI-1 and colder YD/GS-1 (owing to the effects of the
Great Lakes and other periglacial lakes on regional climate, see also Hostetler et al., 2000). In a
similar fashion, Schenk et al. (2018) synthesized multi-proxy biotic evidence from Europe and
showed that summer temperatures remained high over the YD/GS-1, a result consistent with high-
resolution climate simulations they describe that employed an experimental design
consistent with the AMOC/ACC hypothesis.
Neither Fastovich et al. (2020) or Griggs et al. (2022) have anywhere near the resolution necessary to refute the YDIH. Schenk et al. (2018) use computer simulations which explicitly do not consider the YDIH, and so are irrelevant.
Other lines of evidence are inconsistent with the “Impact Winter” scenario. Extinction of
megafauna is the clearest example, a claim discussed and dismissed in Section 3.2.
As it is a secondary effect, Megafaunal extinction is not primary evidence for/against the YD impact and its effect on climate. However, Stewart et al. (2021) state "we think our findings point more clearly to the onset of the YD as driving megafauna declines and extinctions", and "there is evidence to suggest that the YD involved a specific set of climatic and ecological changes that may have been particularly devastating to megafauna populations". Moreover, O'Keefe et al. (2023) have just published the most detailed and accurate study of N. American extinctions near the YD onset. This study is important because to date the megafaunal remains they only use AMS radiocarbon dates from collagen in bones, which avoids many of the pitfalls of earlier radiocarbon dating attempts on megafaunal remains. Their data reflects the situation in the region around the Labrea tar pits in California, and it fully supports the YDIH (see a previous blog post for details). A similar study by Conroy et al. (2021) focussed on Alaska and the Yukon, also mainly using AMS radiocarbon data from bone and tooth collagen, gives a similar result. Thus, the megafaunal extinction evidence, when accurate radiocarbon dates only are used, is fully consistent with the YDIH and does not refute it.
The occurrence of catastrophic wildfires at the YDB is another well-known claim. Evidence for fires
is ubiquitous in lake and other stratigraphic records (Marlon et al., 2013) and provides no
evidence for a unique burning event at the YDB, further discussed in Section 9.
Other paleobotanical records from across North America show a wide variety of changes before,
during, and after the YDC but nothing unique at the YDB (e.g., Meltzer and Holliday 2010;
Straus and Goebel, 2011; Eren, 2012).
As for all other archaeological and paleontological data, including charcoal, it should be compared to the YD boundary to be certain of its context. Opponents of the YDIH, including Marlon et al. (2013) never do this. However, it is already clear that the latest work by O'Keefe et al. (2023) (see the previous blog post for details), Conroy et al. (2021) and earlier work by Gill et al. (2009) are completely consistent with the YDIH, despite claims to the contrary by those authors who prefer to present an alternative story. Therefore, we can have little confidence in Holliday et al.'s claims. Instead, it is clear that there is evidence for exceptional wildfires at the YD onset.
Similarly, geomorphic and stratigraphic records provide no evidence for a cataclysm at the YDB
(Meltzer and Holliday 2010; Holliday and Miller 2013;
papers in Gillespie et al., 2004; Straus and Goebel, 2011; Eren, 2012) (Sections 5.6, 6, 13.7).
The geochemical signals of an ET impact are often found at the base of the YD black mat, which appears to stretch over several continents. Of course, Holliday et al. deny its dating is accurate. Nevertheless, it provides ample stratigraphic evidence for a cataclysm at the YDB.
Thus, the third assumption, that the sign, pattern, and magnitude of the climatic changes at the
beginning of the YD/GS-1 are consistent with the “Impact Winter” scenario, is false, i.e., not
supported by the paleoclimatic evidence.
This is false for all the reasons above, i.e. none of the evidence they provide has sufficient resolution to detect a brief impact winter at the YD onset. So, they have not provided any refutation evidence at all.
The fourth major assumption that underlies and motivates the YDIH is that the YD/GS-1
(and especially its beginning) is a unique event, featuring the occurrence of specific controls or
exceptional or unusual levels of individual paleoenvironmental indicators not found at other
times in those records. If so, that would imply that the YD/GS-1 must require a singular
explanation (such as a comet impact, supernova, or exceptional volcanism), as opposed to a more
generic explanation with roots internal to the Earth’s climate system, such as that provided by the
AMOC/ACC hypothesis.
This is misleading because it fails to distinguish between short-term and long-term effects, and it supposes that none of the other D-O event can have been triggered by 'exceptional events'. This latter aspect is unknown since no other D-O events have been investigated in detail at the short timescales needed to resolve such events.
The variability recorded by the Greenland ice cores during the LGIT
(Figure 1), and throughout the last glacial cycle—the Dansgaard–Oeschger (D-O) events (Lynch-
Stieglitz, 2017), immediately contradicts this supposition. Within the LGIT, for example, the change in
?18O values from the warm GI-1c1 into the cooler GI-1b interval (formerly called the Inter-Allerød
Cold Period (IACP)) is almost as great as that at the beginning of the YD/GS-1, and simple inspection
of the ice-core ?18O data shows how similar in shape, duration, and amplitude the variations of
individual D-O “cycles” (e.g. GS-9 through GS-8 and into GI-7, Figure 1c) are to those during the
LGIT. Simple inspection of the data also contradicts the claims of the occurrence of exceptional values
of various paleoenvironmental indicators at the beginning and during the YD/GS-1 (Holliday et al.,
2020, table 8).
Here, Holliday et al. are referring to long timescale, or low resolution, ice core data which cannot resolve exceptional triggering events at the onset of any D-O events. Moreover, it is notable that they ignore Figure 5 from Wolbach et al. (2018a) which contradicts their claims.
This shows the onset of the YD period was highly unusual in terms of ammonium, formate and acetate ions, which are thought to indicate biomass burning. However, these are discontinuous profiles, meaning they are sampled it irregular intervals, and the data from before 15 kyr is much more sparse compared to afterwards. So it is possible these profiles have missed several more exceptional burning events over the last glacial cycle.
If the (global) climatic changes during the LGIT were unique in character, they would
stand out as an outlier if statistically (as opposed to simply visually) compared with those of the
other D-O events. Nye and Condron (2021) applied a multivariate outlier-detection approach
(Filzmoser et al., 2008) to test this hypothesis using four records that depict the abrupt (and
global) climate variations over the last glacial/interglacial cycle: the Greenland NGRIP ?18O and
CH4 records (Rasmussen et al., 2014; Baumgartner et al., 2014), as well as ?18O and CO2 records
from Antarctic ice-cores (Barbante et al., 2006; Bereiter et al., 2015). For each record and D-O
event, they characterized the shape of the variations by the magnitude of change from interstadial
to stadial, the slope (rate and direction of the change), and the median value (overall level) of the
records. They concluded that the variations during the LGIT (from B-A/GI-1 through the
YD/GS-1) were “not unique compared to those of the other D–O events recorded in the
Greenland ice core record, other than the fact that the median ?18O levels are higher due to
proximity to deglacial warming into the Holocene.
As already discussed, this kind of analysis is irrelevant since the YDIH incorporates the AMOC-weakening process. This analysis has insufficient resolution to make any claims against the YDIH.
The higher median ?18O is also not unique to
the BA/YD, as D–O events 2, 20, and 23 exhibit a similar phenomenon, which we attribute to
their occurrence proximal to long-term global climate fluctuations” (Nye and Condron 2021, p1419).
This observation discredits the notion that the world was already in an interglacial mode
but returned to a glacial one from the consequences of an impact.
But this is not a claim of the YDIH. In any case, this is a matter of interpretation. We considered three different scenarios above and found we could not distinguish between them with this kind of evidence. That is, we cannot decide from this evidence alone whether i) the Holocene might have followed directly from the Bo-Al period if the YD impact event had not occurred (Broecker's suggestion), or ii) The YD period is no different to earlier D-O -glacial fluctuations (Holliday et al.'s suggestion), or iii) The YD period would have happened anyway eventually, even without the YD impact (my suggestion).
Nye and Condron (2021, p1419) conclude “The non-uniqueness of the BA/YD’s shape is clearly
indicated by the statistical indistinguishability of the changes in the Greenland ice core record with the
other D–O events, especially in terms of its ?18O variability.”
But this low-resolution analysis cannot refute the YDIH.
Contemporaneously with the Firestone et al. (2006, 2007) publications, the AMOC/ACC
hypothesis, involving catastrophic drainage of Lake Agassiz (Broecker et al., 1989; Broecker,
2006) was also being discussed, and there was at the time the perception that an abrupt climate
reversal at the close of a glacial period (a termination) was unique to the last one, or at least not
common during terminations (Carlson, 2008; see also Cheng et al., 2009). However, subsequent
work shows that abrupt climate reversals during terminations, and D-O-type variations during
glacial periods, were pervasive throughout the latter half of the Quaternary (and probably
longer). Abrupt climate changes that accompany meltwater events have been identified at the end
of the penultimate glaciation (MIS 6, 191 - 132 ka; Capron et al., 2014; see also Menviel et al.,
2019), and the one before that (MIS 8, 300 - 243 ka; Pérez-MejÃas et al., 2017). In fact, D-Otype
events over the past two glacial cycles (i.e. since MIS 7, 243 - 191 ka) have been identified
in loess sequences (Rousseau et al., 2020), and in marine records with terrestrial indicators (e.g.
Margari et al., 2010; Sierro and Andersen, 2022), at least as far back as MIS 11 (424 - 374 ka;
Martrat et al., 2007, 2014; Oliveira et al., 2016), and also in composite speleothem records
spanning the past 0.5 My (Cheng et al., 2016).
None of this says anything about the trigger of any of these events. This low-resolution data cannot resolve any triggers.
Relative sea level can be used as a rough index of ice sheet configurations that would produce
proglacial lakes that might rapidly drain to the North Atlantic and produce abrupt YD-style climate
reversals. The relative sea-level (see Spratt and Lisiecki, 2016) that prevailed between 13 ka and 11 ka
(i.e., from the B-A/GI-1 through the early Holocene, contemporaneous with the existence of Lake
Agassiz) can be seen via inspection to occur 26% of the time over the past 800 kyr.
The ubiquity throughout the paleoclimatic record of D-O-type events consistent with the
AMOC/ACC hypothesis suggests that this hypothesis is quite robust, and that the abrupt climate
changes themselves are recurrent features of climate variability, and do not require a special
explanation.
Kennett (2019, abstract) concludes that flooding into northern oceans occurred simultaneously (within dating accuracy) from the Laurentide, Fennoscandian and Greenland ice sheets; "It is difficult to explain the triggering of such widespread synchronous changes at the margins of three relatively isolated Northern Hemisphere ice sheets; Laurentide, Fennoscandian and Greenland, and their related proglacial lakes by invoking conventional climatic and/or paleoceanographic processes. Instead, this broad range of evidence is more readily explained by catastrophic processes triggered by a cosmic impact with Earth; the YDB cosmic impact theory." However, this evidence for an impact is indirect and therefore inconclusive.
Smaller amplitude, shorter-term variations are also apparent throughout the record,
like those during B-A/GI-1 (e.g., the IACP or GI-1b, Figure 1), as well as during the Holocene.
The “8.2 ka event” is one such variation (Barber et al., 1999; Thomas et al., 2007), apparently
triggered by the final drainage of glacial lakes Agassiz and Ojibway into Hudson Bay, and thus
to the North Atlantic.
Just as for the YD onset and all other D-O events, it is possible that some of these smaller signals, like the 8.2 kyr event, were also triggered by "special events". Until they are investigated in detail with high-resolution data, we cannot make any decision about this.
Like the YD/GS-1 (and other D-O events), abrupt climate change in the
North Atlantic region is registered globally (e.g., Morrill et al., 2013), and also like the YD/GS-
1, with no discernable lag (Parker and Harrison, 2022). Because there are no others of
comparable magnitude during the Holocene, the “8.2 ka event” might be regarded as unique or
“one-off” occurrence, but there is evidence for a similar “interglacial” event during the previous
interglacial, at 125 ka, the “LILO” (“Last Interglacial Laurentide outburst”) event (Zhou and
McManus, 2022). Both the “8.2 ka” and “LILO” events are consistent with the AMOC/ACC
hypothesis, and so neither requires a special explanation.
Just because they do not require a special explanation for their long-timescale behaviour does not mean they do not require a special explanation for their short-timescale behaviour. Only high resolution data can resolve these questions.
Consequently, the primary attribute that distinguishes the YD/GS-1 is that it is simply the
most recent D-O-type event (and well within the range of annual layer-counting chronologies or
calibration of radiocarbon ages, and so comparatively well studied). It is consistent with the
AMOC/ACC hypothesis, and there is no need for a singular or exceptional explanation for the
YD/GS-1.
We can agree with this statement if only low-resolution long-timescale data is considered. But the geochemical signals of an impact at the onset of YD cooling suggest the impact triggered this particular event, and this also helps to explain other observations. Only high-resolution paleoclimate data around the YD onset is useful in this debate.
Surprisingly little discussion within the YDIH literature deals with the other 25 abrupt
climate changes within the last glacial cycle (e.g., the D-O events) with the exception of
Wolbach et al.’s (2018a,b) and Sweatman’s (2021) misunderstanding of paleofire records over
the last glacial cycle (Sections 9.1, 9.2). This is in keeping, however, with a methodological
approach of looking only where one expects to find something, and not elsewhere (see also
Section 4).
No evidence is provided here to refute the YDIH.
For example, Powell (2022, p 3) cites as an example of exceptional occurrences and
observations at the YDB by quoting Kennett’s (2019, abstract) observations that the drainage of
Lake Agassiz and the Baltic Ice Lake, as well as changes in the margins of the ice shelves at 12.9
ka, cannot be explained “by invoking conventional climatic and/or paleoceanographic processes.
Instead, this broad range of evidence is more readily explained by catastrophic processes
triggered by a cosmic impact with Earth; the YDB cosmic impact theory.” But the ice-core and
other paleoenvironmental records of the D-O events tell us that conditions sufficient to
significantly weaken the AMOC as well as produce a globally registered abrupt climate change
happened 26 times over the past 120 kyr, and hundreds of times over the Quaternary.
As mentioned by Kennett (2019), within dating accuracy, there appears to be simultaneous disruption of, and flooding from, the Laurentide, Fennoscandian and Greenland ice sheets. This coincidence is most easily explained by a large cosmic impact and should not simply be dismissed.
The YDIH evidently was conceived to solve another problem that does not exist.
To summarize, the YD/GS-1, and the YDB in particular, and the accompanying
environmental changes appear to be globally synchronous, which is consistent with the first
assumption of the YDIH, but note those changes are also consistent with the AMOC/ACC
hypothesis.
As expected of the YDIH.
The second assumption, synchroneity of the impact indicators is indeed
problematical, is deeply discussed in Section 5.
But no evidence is presented here.
The heterogeneous spatial pattern of the
environmental changes is not consistent with the “Impact Winter” scenario, and hence does not
support the third assumption,
But the low-resolution data used by Holliday is not sufficient to address this question.
and the observation that abrupt climate changes appear frequently
throughout the Quaternary, provides no support for the notion that YD/GS-1 is exceptional or
special, the fourth assumption.
Each of the triggers for these events must be investigated individually. Cosmic impact triggers might be common or rare. We simply do not know yet.
5.1. Befuddled Dating the Beginning of the YDC
A theme that runs through the YDIH literature is the effort to determine, using terrestrial
radiocarbon dates, the precise age of the beginning of the YDC. This is an unnecessary effort to
solve another nonexistent problem.
And yet Holliday et al. (2023) also claim the YDIH dating methods are insufficiently accurate. Which is it? More effort or less effort is needed to resolve this question? They can't have it both ways.
With the development of the Greenland Ice Core Chronology
2005 (GICC05) that placed the GRIP, NGRIP, and GISP2 ice cores on a common time scale
(Rasmussen et al., 2006, 2014), the age of the onset of the last stadial (Figure 1), termed the
Greenland Stadial 1 (GS-1), was established as 12,846 +/- 4 yr [BP 1950, GICC05 or 12,896 +/-
4 yr [b2k, GICC05]. This is an age very close to that given by Mayewski et al. (1993) for the
GISP2 record that played a role in the initial discussions of the YDIH (e.g., 12,859 +/- 250 yr
[BP 1950, Meese/Sowers]).
One of the general issues that arise when establishing correlations or tie-points among
multiple records is the variations in age controls and chronologies of the records. This can be
seen in the simple task of establishing the calendar age of the onset of YD/GS-1. The most
precise chronology is that provided by the Greenland ice-core records, based on annual-layer
counting and measurement, and which does not require “calibration” or conversion to an
absolute age scale (as would radiocarbon ages).
Once again, this is misleading. It is well known that even the GiCC05 chronology has a layer count that could be wrong by ~ 140 years at the onset of the YD period. Thus, it lacks precision. Holliday et al.'s implicit claim that the GICC05 chronology is somehow superior to the radiocarbon chronology is, therefore, false. There are plenty of quotes from relevant papers to support this view. Indeed, there are published data files that enable conversion from published ice core chronologies to radiocarbon chronologies.
Rasmussen et al. (2014) give an age of 12,896 +/-4 yr
[b2k, GICC05] or 12,846 [BP 1950, GICC05], a refinement of the earlier estimate of
Steffensen et al. (2008) of 12,900 yr [b2k, GICC05]. This age was determined by a statistical
change-point analysis of ?18O as well as other constituents of the NGRIP core and placed on the
GICC05 chronology (Rasmussen et al., 2006), as described by Rasmussen et al. (2014) and
Seierstad et al. (2014). The latter two references also describe the transfer of the GICC05
chronology to the older “Meese/Sowers” chronology of the GISP2 core (see Holliday et al.,
2020, table 7 for chronology sources).
Let's go through this in detail.
Above is the relevant part of Figure 2 from Steffensen et al. (2008), which is their main result, focussed on the YD onset from 11.15 to 10.65 kyr BCE.
There's a lot of information here. First, this data is mainly from the NGRIP ice core. The left-hand plots are 20-yr block-averages of d (in red) and delta 18O (in blue). d is the deuterium excess and according to Steffensen et al. is "a second order isotopic parameter that contains information on fractionation effects caused by the evaporation of source water". In other words, changes in d are thought to correlate to changes in weather patterns. delta 18O, on the other hand, is thought to be a much more direct measure of temperature at the location of the ice core. The scale on the left is the GICC05 chronology which was developed by counting layers in the NGRIP ice core. This is itself quite a difficult and uncertain exercise, so the GICC05 chronology should just be seen as a best-guess chronology. As already mentioned, it could be in error by ~ 140 years at the time of the YD onset.
The red plot shows an abrupt change in d just after 10,900 BCE, which is also indicated by the horizontal grey line. Another sudden drop in d occurs just after 10,800 BCE. The blue plot, showing the local temperature, seems to drop quite rapidly, but noisily, from just before 10,900 BCE to around 10,700 BCE.
The white region of this plot is expanded on the right where we can see more detail and higher resolution data. On this expanded region of the plot, from left to right, we have d (red) and delta 18O (blue) at higher resolution, followed by dust (orange) and calcium ions (light blue), sodium ions (purple) and ice layer thickness (green). The coloured bars on the right show the average start and end points of ramp functions fitted to this data. The colour of the bars is the same as the plot it references. The second blue bar (7) refers to the GRIP ice core rather than the NGRIP ice core. These bars show the lengths of these ramp functions at their average positions. However, the start and finish points of these ramp functions are uncertain, and this uncertainty is shown by the grey bars on the relevant plots.
So how should we interpret all this data?
Well, despite only having data for the NGRIP ice core, Steffensen et al. (2008) define the onset of the YD by the initial sudden drop in d just after 10,900 BCE. While this might be correct, it can be challenged; indeed Steffensen et al. (2008) specifically concede this in their paper. For example, should the YD onset be defined according to changes in weather patterns or changes in temperature? Or how about ice layer thickness? The advantage of using d to define the onset of the YD is that is is very precisely defined - the abrupt change in d in the NGRIP ice core appears to occur in only a few years. This is indeed convenient. But is it correct? And what does the change in d really mean? Maybe it is just a statistical fluke. Indeed, if we examine the fluctuations in d, it could be argued that some will be much larger than others purely by chance, and therefore we should not attach any special meaning to them. Since Steffensen did not perform any statistical evaluation of the distribution of these fluctuations, we cannot know whether this jump is expected by chance or whether it is special in some sense. In fact, if we look at the thick red plot in Steffensen's figure above, we see several large jumps in the deuterium excess, d, over this time period with a similar size to the one identified as the YD onset.. Are any of them special? Or do they just reflect the expected distribution of large jumps in a fluctuating signal, with no special meaning required for any of them?
In fact, there is ongoing debate in the research literature about the utility and interpretation of the deuterium excess as a proxy for climate change. See, for example, Lewis et al. (2013, abstract) "... although deuterium excess is generally a faithful tracer of source temperatures as estimated by the MCIM approach, large discrepancies in the isotope-climate relationship occur around Greenland during the Last Glacial Maximum simulation, when precipitation seasonality and moisture source regions were notably different from the present. This identified sensitivity in d as a source temperature proxy suggests that quantitative climate reconstructions from deuterium excess should be treated with caution for some sites when boundary conditions are significantly different from the present day."
Wolbach et al.'s (2018a) view is the same; from their appendix "Steffensen et al. (2008) measured five different proxies for climate change and obtained five seemingly contradictory dates for the YD onset, from oldest to youngest: (1) 12,896 +- 1.5 cal BP for deuterium excess (d), a proxy of past ocean surface temperatures; (2) 12,787 +- 24 cal BP for annual layer thickness (l), a measure of the rate of snowfall; (3) 12,737 +- 8.9 cal BP for calcium ions (Ca21), a proxy for source strength and transport conditions from terrestrial sources; (4) 12,735 +- 8.9 cal BP for dust concentrations, a proxy for source strength and transport conditions from terrestrial sources; and (5) 12,712 +- 74 for d18O, a proxy for past air temperature at the site. Regarding these dating discrepancies, Steffensen et al. (2008) considers that the oldest date, 12,896 5 1.5 cal BP, is the actual date of YD onset and that the younger dates are due to lags in the response times of the other proxies. All of the above age differences affect the reliability of conclusions that use ice-core data."
Moreover, why define the YD onset from a single ice core? And why should that ice core be the NGRIP ice core? We could have much more confidence in their definition for the onset of the YD if they could show this abrupt change in d just after 10,900 BCE was replicated in the other ice cores. We could then be more confident it is a real effect, rather than just an artefact of the NGRIP ice core. And what meaning should we attach to this abrupt change? Perhaps it is just a local change that is not reflected more broadly?
Another crucial issue is the way the different ice cores are linked or aligned which determines how the GISP2 platinum signal is aligned versus the NGRIP d anomaly at ~10,900 BCE. This alignment is described in Seierstad et al. (2014). In fact we find the NGRIP and GISP2 ice cores are linked at two neighbouring chemo-stratigraphic points near the YD onset; 1) 1525.862 m (NGRIP2) = 1712.406 m (GSIP2), and 2) 1526.497 m (NGRIP2) = 1713.027 m (GSIP2). This latter point is close to the onset of the YD period according to Steffensen et al. (2008). However, the platinum signal in the GISP2 ice core occurs at 1712.56 m (the small pre-peak) and at 1712.19 m (the main peak). That is, the platinum signal is around 0.5 - 0.8 meters higher in the GISP2 ice core than expected given Seierstad's definition of the YD onset, corresponding to about 15-25 years. This is shown in Holliday et al.'s Figure 2, below.
If any of these link points are wrong, the alignment of the ice cores and the position of the platinum signal vs the NGRIP definition of the YD onset will also be wrong. To have confidence in this linking of ice cores we need the complete set of linked chemo-stratigraphic points (mainly volcanic signatures) in these ice cores around the YD onset. Unfortunately, this data doesn't appear to be published. All we have from Seierstad et al. (2014) is "For all sections, 2-3 investigators have done the matching independently with repeated inspections to test the reproducibility of the matching." and "Between 8.2 ka and 14.9 ka b2k the ECM-based NGRIP-GRIP-GISP2 match points published by Rasmussen et al. (2006) have been updated. Several investigators have re-examined this section with all available chemistry data using the Matchmaker tool. As a result of this revision, the number of match points across this section has increased by a factor of seven, and some adjustment of the synchronization has been introduced. The adjustment between the cores is typically of the order of 5 cm or less, but grows to around 20 cm in a few cases, and reaches a maximum of 33 cm in Greenland Interstadial (GI) 1."
In other words, there is agreement between 2-3 researchers that the tie points between NGRIP and GISP2 have been identified correctly, to within 33 cm, which is about 10 years. But is there no room for error at all? We simply cannot know. Given the importance of this information, it is surprising it has not been published. Not even in any supplementary materials. But apparently there is no need for any doubt. Instead, we are required to trust Seierstad et al. (2014).
Fortunately, Seierstad et al. (2014) do provide an example of this matching exercise, but not around the YD onset. Instead, it is for the period 58.56 - 59.67 kyr (before 2000 AD) and shown above. The vertical grey bars show match points located according to common spikes in ammonium ions (orange), sulphate ions (green) and electrical conductivity measurements (ECM, light blue) between the GISP2 (top) and NGRIP (bottom) plots. The ECM measurements essentially measure the total ionic content of the ice.
This matching exercise for this period looks to be quite robust. The various peaks and trends are well matched overall. But notice the noise and resolution of the GISP2 ammonium and sulphate data. I estimate the uncertainty here is ~ 20 cm, or 5-7 years. But what about the section around the onset of the YD period? We have to trust Seierstad et al. (2014) that their matching is no less accurate in this region, despite it being the onset of the YD period where we can expect, due to catastrophic processes, this matching process and ice layer counting to be more difficult.
Now let's consider Svensson et.al. (2020). They published a comparison between the delta 18O response of Greenland ice cores on the GICC05 chronology. It shows changes in the NGRIP ice core delta 18O response (green) are poorly aligned with the other Greenland ice cores (GSIP2, GRIP and NEEM) around the YD onset.
All these ice cores have been aligned by comparing their chemo-stratigraphy in the same way as for Seierstad et al. (2014) for the Greenland ice cores. The black spike on the above plot is the position of the GISP2 platinum signal. Other than for the NGRIP ice core (green), we see the platinum signal correlates quite well with changes in temperature in the other ice cores (GISP2, GRIP and NEEM). Yet it is the NGRIP ice core which is used to define the YD onset via the deuterium excess, d. This seems to be somewhat inconsistent, and suggests the use of the NGRIP ice core as a kind of 'master' might not be appropriate.
To conclude; in order to have more confidence in the definition of the onset of the YD defined by Steffensen et al. (2008) relative to the GISP2 platinum spike we need to see i) how the deuterium excess, d, compares between the different ice cores, and ii) how the different ice cores have been linked chemo-stratigraphically near the YD onset (i.e. a plot similar to the example above is needed). But none of this information is published. Moreover, how much confidence do we have in the deuterium excess for locating the YD climate onset?
Regarding the matching process, the platinum signal is perfect for aligning all the ice cores. This is because it is a unique and well-defined signal that is expected to indicate a precise moment in time (to within 20 years) at the onset of the YD period. As soon as Petaev et al. (2013) discovered this signal in the GISP2 ice core it should have been sought in all the other ice cores. But this has not happened. Until this platinum signal is located in all the Greenland ice cores, there is at least some doubt in the YD onset defined by Steffensen et al. (2008) and its position relative to the GISP2 platinum spike.
Essentially, there could be errors in the following;
i) The use of deuterium excess as a signal for the onset of YD cooling.
ii) Interpretation of the discrete jump in the NGRIP deuterium excess as a special event rather than just an expected fluctuation that does not have any special meaning.
iii) The use of the NGRIP ice core as the 'master' record for the YD climate onset.
iv) Tie points that link the GISP2 and NGRIP ice cores could be incorrectly matched at the YD onset.
Cheng et al. (2020) analyzed a large suite of U-Th dated speleothems, and again using
change-point analysis, determined an age of 12,870 +/- 30 yr [BP 1950, U-Th] (or 12,920 yr
[b2k, U-Th]) (see also Section 11). Consequently, even in relatively well-dated ice-core and
speleothem records not dependent on calibration, there is a range of about 25 years (or more
when including uncertainties) in the estimated age of the onset of the YD/GS-1 using sources
other than conventionally calibrated 14C dates.
Let's look at the Cheng et al. (2020) result in detail. They use the Seso speleothem, from a cave in Spain, along with the Greenland NGRIP ice core to define the onset of the YD. Looking at the above plot from Cheng et al. (2020), which compares the Seso speleothem and NGRIP records for delta 18O, we see they are fairly consistent across the YD period.
The problem with this approach, though, is that it is inconsistent with Steffensen et al. (2008) who used the deuterium excess, d, profile in the NGRIP core only to define the YD onset, not the delta 18O record. Apparently, according to Cheng et al. (2020), other measures are useful (as also suggested by Wolbach et al. (2018a)). Perhaps the deuterium excess is over-rated?
Looking in more detail at the above plot, Cheng et al. (2020) notice a large positive fluctuation in the Seso speleothem record coincides with a large positive fluctuation in the NGRIP record, and use this to highlight the consistency of their results. But we can clearly see the large positive fluctuation in the Seso speleothem record occurs against a reasonably stable temperature backdrop. In fact, the speleothem record appears to be stable for ~ 50 years after this large fluctuation, which contradicts their claim that the large positive fluctuation is near the onset of the YD period. Their location for the onset of the YD period in the Seso record is determined by a ramp-fitting procedure, but probably this procedure has been unduly influenced by the large positive fluctuation. This reminds us that it is always good practice to check the results of automatic (brainless) methods against our own good sense.
In any case, the speleothem record has an uncertainty of 20-40 years while the Greenland ice core records have an uncertainty of ~ 140 years at the YD onset, so all this discussion is moot. There is no way these records can be accurately compared at the YD onset. Nor does Cheng et al. have sufficient resolution to dispute the possibility of a brief YD impact winter. Thus Cheng et al. (2020) cannot be used to refute the YDIH.
Recently, Reinig et al. (2021) determined the age
of the Laacher See eruption using multiparameter radiocarbon age calibration (i.e., wigglematching
to the Swiss Late Glacial Master Radiocarbon [SWILM-14C] datasets) to 13,006 +/- 9
cal yr BP [1950] and placed the onset of the YD/GS-1 at 12,801 +/- 12 cal yr BP [1950], or
12,851 cal yr [b2k]. The existence of multiple generations of chronologies for the ice cores, and
more than one reference age (i.e., 1950 CE vs. 2000 CE “b2k”) creates amplitude for making
mistakes. See, for example, the plotting anomalies in Wolbach et al. (2018a) first noted by
Holliday et al. (2020), and which were not fully addressed by Wolbach et al. (2020), or
Sweatman’s (2021 p 3) statement that the beginning of the YD/GS-1 in the GISP2 core was at
“10,890 BP.” This latter instance is likely a typo, i.e., it probably should have read “12,890 yr
BP” (presumably yr BP 1950),
correct - I made a typo
but if the typo was in the calendar-age designation, i.e., if it should have
read “10,890 BCE”, then that would give a plausible age of 12,839 yr [BP 1950,
Meese/Sowers]. Later, Sweatman (2021 p 19) states, “No YDB site has yet been found to be
obviously inconsistent with a synchronous event circa 10,785 +/- 50 cal BP (2 sd).” Again, this
is problematical: the suffix “cal BP” implies a calibrated radiocarbon age, in which case
“10,785+/- 50 cal BP” should probably be read as “12,785 cal BP”,
correct, I made another typo
but it could also be the case
that what intended was “10,785 +/- 50 BCE”, which gives a plausible age of 12,734 yr BP.
This situation can also lead to the adoption of overly casual approaches for aligning
chronologies. For example, Sweatman (2021, p 5) notes that to compare platinum anomalies
from near and far with those in the GISP2 core (i.e., Petaev et al., 2013a) (Figure 2), … the
GISP2 ice core chronology must first be converted into a radiocarbon timescale. This is
achieved by the GICC05 chronology. Essentially, according to the radiocarbon-aligned GICC05
chronology we should subtract around 80 years from GISP2 dates in the vicinity of the YD
cooling (Svensson et al., 2008)”. Leaving aside the question of why the calendric chronology of
the ice cores should necessarily be converted to a radiocarbon one (i.e., “inverse calibration”
which can create artifacts in the “uncalibrated” ages; Bartlein et al., 1995), there are three things
that are wrong with this idea First, the GICC05 chronology is not “radiocarbon-aligned” (Sweatman,
2021, p 5), but is based on annual layer-counting and electrical conductivity as well as continuous-flow
measurements of impurities in the GRIP and NGRIP ice cores (Rasmussen et al., 2006); no
radiocarbon dating was involved.
All correct - my mis-statements. But they have no impact on Holliday et al.'s supposed refutation.
Second, the paper by Svensson et al. (2008) focuses on the
extension of the GICC05 timescale from 42 ka to 60 ka and offers no simple prescription for
adjusting from one chronology to another nor does it prescribe the 80-year offset as quoted by
Sweatman (2021). Sweatman (2021) is unclear on the source of this offset value, perhaps from
Southon (2002). The chronologies of the GRIP and GISP2 ice cores exhibit discrepancies first
thought to arise from gradual accumulation of errors during counting of annual layers. Southon
(2002) found most of the offset centered on two periods, an 80-year discrepancy near 3300 –
3400 yr BP and a second 100-year discrepancy near the onset of the YD/GS-1. Third, the
difference in ages between the GICC05 chronology assigned to the GISP2 core, and the
“original” (Mayewski et al., 1993) or Meese/Sowers chronology is 65 yr at the beginning of GS-
1, and decreases to near zero at the end, making any simple prescription for converting from one
ice-core chronology to another unsuitable in the first place (Figure 3).
Exactly. This is where my 80-year estimate is from. It was just an estimate for their difference around the YD onset.
Similarly, Wolbach et al. (2020), while attempting to explain the discrepancy between the
published GISP2 data and the plot in their figure 3C of Wolbach et al. (2018a) argued that
“Mayewski et al. reported their data on a pre-GICC05 age scale that cannot be directly converted
to GICC05 because the ice layers were subsequently recounted. Instead, the original GISP2 age
scale must be interpolated to the GICC05 scale using the ages of depths that are common to both
scales, yielding an average difference of 10–15 ice-layer years.” Figure 3 demonstrates that this
value, like Sweatman’s 80-year offset, is not appropriate. (We also note that two different age
scales are used in figure 3C of Wolbach et al. (2018a) to plot data from the same ice core. They
are aligned at “12600 Calendar yrs B.P.” (on their plot), but differ by about 20 yrs at “13000
Calendar yrs B.P.”
In fact, Figure 3C of Wolbach et al. (2018a) which compares GISP2 data on the GICC05 timescale is correct. There is a problem, however, with Figure 4C from Wolbach et al. (2018a) which compares NGRIP data (delta 18O and ammonium ions) with GISP2 data (platinum). Indeed, Holiday et al. are correct that the NGRIP delta 18O data has been shifted by ~50 years to the right compared with the data given by Seierstad et al. (2015) and the ammonium data is shifted by ~ 15 years with respect to the data provided by Fischer et. al. (2015). These problem arise because Wolbach et al. (2018a) consider the GISP2 (Meese-Sowers) layer counts and NGRIP (GICC05) layer counts at the YD onset differ by only ~ 15 years. However, the discrepancy is actually ~ 65 years at the YD onset (as mentioned above), close to the value I quoted (~ 80 years).
This makes it look like the platinum spike and the onset of cooling in NGRIP occur roughly simultaneously when in fact the cooling in NGRIP occurs first by ~ 15-25 years (see Holliday et al.'s plot above). But as we already know, NGRIP and several other Greenland ice cores are inconsistent at this point, which makes precise definition of the YD onset quite uncertain, as already discussed (see the plot above from Svensson et al. (2020).
The specific chronology assigned to a particular core does influence comparisons among
cores but does not alter the relative position of samples within a core. Thus, the “Platinum spike”
samples of Petaev et al. (2013a), at 1,712.125 to 1,712.250 m (82.2 ppt) and 1,712.250 to
1,712.375 m (27.6 ppt) in the GISP2 core, when placed on the GICC05 timescale still lie the
better part of a meter above the level in the core dated to the onset of the YD/GS-1 (12,896 yr
[b2k, GICC05]; 1,713.00 to 1,713.20 m in the GISP2 core), and therefore must post-date it
(Figure 2). If we take the midpoint depths of these samples, 1,712.3125 m and 1712.1875 m,
respectively, then these samples date to 12,874.6 yr [b2k, GICC05] and 12,871.3 yr [b2k,
GICC05], 20 years after the YD/GS-1 beginning.
This is all discussed above and is correct. However, as we saw, there could be several problems with this comparison.
Cheng et al. (2020, figures S3, S8) further
demonstrate that the Petaev et al. (2013a) “Pt-spike” occurs after the onset of the YD/GS-1 (but
note that they present ages relative to 1950 CE).
As discussed above, Cheng et al. (2020) cannot be used to refute the YDIH. There is no way to reliably compare the timescales of the Seso speleothem and Greenland ice cores with sufficient precision, and the Seso record lacks the necessary resolution to refute a brief impact winter.
A further complication in dating samples at the YDB (and throughout the LGIT) arises
from the presence of “age plateaus” in the radiocarbon calibration curve (Bradley, 2015;
Sarnthein et al., 2020). The age plateaus mark intervals when atmospheric 14C temporarily
increased, which could be related to increased production, but around the time of the YDC, is
likely due to changes in atmosphere-ocean 14C exchange (ocean ventilation) and in oceanic and
atmospheric circulation (Stuiver et al., 1991). The age plateau in the latter half of the YD/GS-1
is comparatively well known, in which 1000 years of calendar time (12.4 - 11.4 ka) is
compressed into 400 years of radiocarbon time (10.4 - 10.0 ka 14C yr BP). The recent IntCal20
curve (Reimer et al., 2020) provides details on a second plateau at the transition between the BA/
GI-1 and YD/GS-1, in which 350 years of calendar time (13.10 - 12.75 ka) is compressed into
200 years of radiocarbon time (11.1 - 10.9 14C yr BP). The age compression has the effect of
making a range of calendar ages appear to be more tightly clustered in radiocarbon time than
they really are, thereby contributing to a false sense of synchroneity.
But this does not refute the YDIH.
SUMMARY
To conclude, Holliday et al. do not provide any arguments that refute the YDIH at all. The only evidence they provide that potentially contradicts the YDIH is the timing of the platinum spike in the GISP2 ice core versus the timing of the YD onset defined by Steffensen et al. (2008) using the deuterium excess from the NGRIP ice core. But this discrepancy can itself be questioned from multiple angles (see above). On the other hand, we see from Wolbach et al. (2018a) that data from the GRIP ice core indicates an extraordinary biomass burning event did actually occur near the YD onset. Moreover, platinum might be a better marker for linking the chronologies of the different ice cores near the YD. And finally, apparently simultaneous (within dating accuracy) flooding near the YD onset from several distant ice sheet margins (Laurentide, Fennoscandian and Greenland) points towards a catastrophic event.
Comments
Post a Comment