APPENDICES: Here’s where we’ll show our ongoing discussions. Join in via email to nasamike@nasamike.com

A1. Scientific Dating Techniques
a) All About Ice Cores from renown Ice Core researcher, Richard Alley
From: Richard Alley <ralley@essc.psu.edu>
Date: Mon, 9 Oct 2000 15:52:22 -0400 (EDT)
To: michael.comberiate@gsfc.nasa.gov
Subject: Re: Dating Ice Cores

The long answer to your questions will be published this month or next in The Two-Mile Time Machine: Ice Cores, Abrupt Climate Change, and Our Future, Princeton University Press, by me. The shorter answer is at least partly buried in Alley, R.B. and M.L. Bender. Greenland ice cores: Frozen in time. Scientific American, 278(2): 80-85 (1998).

The shorter-yet answers are (and I'm using language more appropriate for you as a scientist than for real people in the real world, although with the examples from the real world):

Aside from theories that assume the Earth is billions of years old, what hard evidence is there in the ice cores that testifies that they have been accumulating for many millennia? I wanted to ask you about dating the ice cores from Antarctica and Greenland:

1) How far back can you go reliably?

There are indications of older ice yet, but the oldest with pretty good dating is 420,000 years give or take 10,000 years. The oldest with annual-layer counting is about 110,000 years. There is a complexity of ice flow in this. Snowfall on the top of an ice sheet is approximately balanced by flow to the coast to make icebergs or to melt at low elevation. Ice moves down and out. Because east-side ice flows east and west-side ice flows west, the ice between is being stretched and thinned. Just as a blob of pancake batter becomes a thin, wide pancake, the ice in the ice sheet is thinned by flow. A marker placed on the surface of the ice moves down as the ice beneath it thins, and over long times the ice sheet adjusts so the downward motion of the surface just balances new snowfall. Where snowfall is rapid, annual layers can be observed, but old ice has moved down close to the bed where the annual layers are disturbed by flow over bedrock bumps (think of flow over a waffle iron rather than a pancake griddle...); where snowfall is slower, the ice takes longer to be buried near the bed, but the snowdrifts at the surface are higher than the annual snowfall, so we can't recognize annual layers any more.

2) What are you using to determine the seasonal variations and can that information be ambiguous?

Summer and winter snow differ for various reasons. During the summer, the sun shines on the surface. In those places where most ice cores have been drilled, it rarely or never becomes warm enough to cause melting. However, the solar radiation pushes the temperature close to the melting point. Sublimation produces low-density layers called hoarfrost or just hoar. In the winter, the sun doesn't shine, the surface is cold, less energy is available for sublimation, and so one gets wind-packed fine-grained snow rather than hoar. Many other annual markers also exist. The isotopic composition of snow is a paleothermometer, and varies between warm summer and cold winter. (Heavy water has lower vapor pressure than light, so heavy is progressively removed as cooling causes precipitation, leaving lighter and lighter vapor that makes isotopically lighter and lighter snow as the temperature falls.) The sun makes hydrogen peroxide in the air, which doesn't stay up very long, so summer layers have hydrogen peroxide, but winter layers don't.

Big late-winter storms blow sea-salt frozen out of sea ice off the ocean and onto the ice sheet, so the sodium concentration oscillates annually. And so on--there are more.

All can be ambiguous, of course--nothing in the world is perfect. We check for consistency in several ways:

a) Is one observer reproducible? If I count layers in, say, half a mile of core, will I get the same answer if I go back the next year and repeat the exercise?
b) Are multiple observers reproducible? If Christopher Shuman, Tony Gow, Deb Meese and I all count layers in the same piece of core, without telling each other what we got, will we agree?
c) Are different methods reproducible? Do isotopic, electrical, visible, and chemical means produce the same number of years?
d) Can we match independently dated time horizons? Big volcanic eruptions spread ash around the globe. That ash can be found in ice cores, and the chemistry of the ash allows one to check that you know which volcano it came from. Icelandic eruptions are especially useful in dating Greenland cores because of proximity and good historical records. Older, before humans were writing things down, we can still check ages against dates from tree-ring and other studies. And, the ice cores show clearly that at certain times in the past, very large, very rapid, and very widespread climate changes occurred

(We see simultaneous changes in Greenland ice in local snowfall, sea-salt concentrations blown from the ocean, dust concentrations with isotopic, chemical and mineralogical characteristics showing a Chinese origin, and swamp-gas methane monitoring the global area of wetlands). We can then find these events in climate records from other parts of the world, and compare ages with theirs based on counting annual layers in trees, lakes or
estuaries (only in a few favorable ones, but they do exist), and on radiometric dates.

Doing this shows that some cores can be dated more accurately than others. For central Greenland, we are good to 1 year in 100 for the last 11,500 years (so if I tell you a given section contains 100 years, it might be 99 or 100 or 101, or maybe 98 or 102, but rarely will be worse than that). Before 11,500, the climate was colder, the cold air delivered less snowfall, stronger winds made bigger drifts, and our errors are somewhat larger--maybe a few percent. Still deeper, the flow of the ice makes the layers thinner and harder to recognize. We have counted more than 110,000 layers, and the ages look to be as accurate or more accurate than any other technique down to about 40,000 years. There is older ice below that, but the flow over the waffle iron of the bed has mixed older and younger layers so we can't puzzle out the ordering.

Note that some young-Earth types are fond of pointing out that WWII airplanes that landed on the Greenland ice sheet were buried a few hundred feet in a few decades, so the few thousand feet of ice in the ice sheet can represent only a few centuries. This is blithering idiocy, because it fails to recognize that the ice is flowing, the layers are thinning with the flow, and so most of the age is below the planes rather than above. (It is ultimately the same mistake made by the "snow accumulation on Greenland and Antarctica will cause the world to become top-heavy and roll over" folks--flowing ice proves them wrong.)

3) When the ice turns to slush at some depth, is there any way to continue dating it?
4) At what depths does it turn to slush or does that vary from place to place?

Actually, it doesn't turn to slush in most places. The upper part of the ice sheet is cold, with surface temperatures tens of degrees C below freezing. The downward motion of the surface snow (because of the spreading of the ice beneath, making room for more snowfall) takes that cold surface ice downward, so that for some distance down the temperature is quite cold. In central Greenland, the ice a mile down is actually colder than the ice above it, because the deeper ice was deposited during the cold ice age and hasn't had time to warm up yet. (Heat flow is diffusive, and so the distance heat
propagates in time t increases as the square root of t; hence, big quantities of ice buried a mile down take a long time to change temperature.)
Closer to the bed, the vertical velocity is lower, so conductive heat flow is more important, and the ice warms downward. In some places such as central Greenland, the ice is below freezing all the way to the bed. In other places such as Vostok in central East Antarctica, the ice reaches the melting point at the bottom (and there is a lake beneath the ice at Vostok), but the ice is below freezing at all places above the interface. In a very few places, there is enough deformation in the ice that the "friction" (viscous dissipation) warms a significant thickness to the melting point, and one might think of the result as "slush". Very rapid deformation regions have not been cored, because of the difficulty of figuring out exactly the elevation from which the ice came, and thus how much temperature change to blame on elevation change, so we only deal with cold ice.

4a) Doesn't the tremendous pressure of a mile or more of ice cause melting at the ground interface?

The pressure-dependence of the melting point of ice is pretty small.  Depending on air-saturation state, you get a degree C drop in melting point

for 750 to 1000 m of ice, so typically melting points at the base of the ice sheet are about -3 C.  Melting is widespread under ice sheets, but the melt

rate is pathetically tiny, because you need the latent heat to give the melting even after you are up to the melting point.  A typical geothermal flux melts

about 3 mm/yr.  As in most places most of the heat from geothermal plus frictional heats is conducted up into the ice, not much melts.  Fast melt

in a few special places with volcanoes, but almost certainly doesn't mean much (the huge jokulhlaups from under Icelandic glaciers during volcanic

eruptions have surprisingly small effect on the flow of the glaciers).

5) Is there any independent way to correlate the ice core dating, beyond the historical records of volcanic events?
6) What assumptions are involved in these dating schemes?

Some of these are given above. Tree rings, lake varves, marine basin varves, and different radiometric techniques including radiocarbon, uranium-thorium, and cosmogenic techniques all obtain an age of about 11,500 years for an abrupt warming-wetting-drop in winds, and the ice core shows that much of the world experienced a warming-wetting-drop in winds at that time, so we have the whole range of dating techniques agreeing. Clearly, each of the dating techniques involves assumptions, but the biggie ultimately is that the world isn't deliberately trying to fool us. Because the assumptions are so different in the different techniques in different regions, it is very difficult to imagine a way that all of them would be fooled in exactly the same way to give almost exactly the same wrong answer. (Making Greenland storms bigger so that a storm looks like a whole year would not affect the decay constant of uranium in a Pacific atoll coral, for example.)

Richard B. Alley, Evan Pugh Professor
Environment Institute and Department of Geosciences
The Pennsylvania State University
204A Deike Building
University Park, PA 16802, USA
ph. 814-863-1700
fax 814-865-3191
email ralley@essc.psu.edu

Richard,

Thanks for the detailed answers, tell me if I misunderstood anything in the following:

1) You are able to actually count "seasonal" variations in some ice cores of almost 500 000 cycles.
2) You can correlate the ice core data with various other independent dating schemes for tree rings, lake varves, marine basin varves, and different radiometric techniques including radiocarbon, back as far as 11,500 years ago. At that time there was a major climate change worldwide.
3) You are assuming that the seasonal changes you observe in the cores have been consistently spaced at yearly intervals throughout this 11,500 years, ie that the current weather patterns have been more or less consistent over this time. (Can you outlaw the possibility that there would be more than one apparent seasonal variation in a given year?)
4) The data in cores beyond this verifiable time are ambiguous but they would seem to indicate that the ice is much older that 11,500 years, even though it's difficult to know how to interpret this data.

Mike
From: Richard Alley <ralley@essc.psu.edu>
Date: Mon, 9 Oct 2000 18:21:57 -0400 (EDT)
To: michael.comberiate@gsfc.nasa.gov
Subject: more
Longest annually resolved record is about 110,000 years. The older part of that record includes significant uncertainties (there are annual layers, but reproducibility for counting is worse than 10% errors, so we can date more accurately using other techniques); the longest annually resolved record that is more accurate or at least as accurate as anything else we've seen is the top 40,000 years of that 110,000 years from Greenland.
>2)You can correlate the ice core data with various other independent dating schemes for tree rings, lake varves, marine basin varves, and different radiometric techniques including radiocarbon, back as far as 11,500 years ago. At that time there was a major climate change worldwide.

Actually, we can run correlations further back than that. The correlation at 11,500 years is especially good, and known from many, many other records. As one goes older, it gets harder to find records to correlate to.
For example, many lakes were hollowed out by glaciers during the ice age. There are no sedimentary records in those lakes from times when the ice was present or older (because the ice scooped out those sediments). The 11,500 year event is often the oldest big change in those lakes, because it comes shortly after the ice melted out. We can correlate the ice cores to other dating schemes for the whole length of the ice cores, but the
number of things to which we can correlate decreases with increasing age. So far, our ice-core record is the longest annually resolved record yet published (longer ones are likely to come, but haven't yet) so in the oldest part of our record we cannot correlate to other annually resolved records, but must use radiometric dating, orbital tuning, or other such things.

>3)You are assuming that the seasonal changes you observe in the cores have been
>consistently spaced at yearly intervals throughout this 11,500 years, ie that the current weather patterns have been more or less consistent over this time. (Can you outlaw the possibility that there would be more than one apparent seasonal variation in a given year?)

Yes, we assume that the recent signal of annual signals is repeated throughout the record. Notice, however, how many, many, many checks we have on this. We have the comparisons to historically dated volcanoes for the last couple of millennia. We have the intercomparisons--we have dated by having multiple individuals count multiple indicators multiple times. And we have the radiometric dates, the tree rings, and so on for older events including volcanic horizons as well as abrupt climate changes. For us to be badly fooled, the annual signal would have to change in such a way that all of the observers would be fooled every time they counted on all of the indicators counted (typically at least three), and this would have to occur in such a way that it didn't
change the mean values of the variables (during the last 11,500 years, there are no long and persistent changes in the mean values of climate variables indicating temperature, snowfall, etc.), didn't affect the physics of how snow turns to ice (we have gas-phase indicators of this, which requires that changes in temperature be exactly offset by changes in accumulation rate), and the changes would have to confuse the tree rings and the lake sediments and the uranium to thorium changes all over the earth in exactly the same
way.

Suppose, for example, that you assumed at some time before a few thousand years ago, there were two features each year that look like annual indicators, so that we counted twice as many years as we should for those older parts of the record. That would indicate half the snow accumulation that happened more recently. Snow accumulation is related to concentrations of cosmogenic isotopes (produced by cosmic rays, things such as 10Be or 36Cl). More snow dilutes the concentration of cosmogenic isotopes, so a twofold miss in the dating would cause a twofold miss in the accumulation rates and a twofold miss in the expected concentration of cosmogenic isotopes. During the post-ice-age times, there are no large changes in concentrations of cosmogenic isotopes, so our error could only have occurred if the cosmogenic isotopes changed at the same time and in the right way as the annual indicators; hence, you require that cosmic ray fluxes into the lower atmosphere are correlated with the annual indicators in the ice cores in just the right way
to fool us. But, one sees that the gases trapped in the ice core are slightly heavier than mean air isotopically, because the upper tens of meters of the ice sheet have interconnected pore spaces before the weight of further snowfall squeezes the deeper snow to ice with isolated bubbles. The main controls on the difference between the trapped air and the free atmosphere (which is a TINY difference, not affecting reconstruction of CO2 or CH4) is the temperature and the snowfall (warmer snow is squeezed to ice faster; faster snow accumulation buries snow farther before it has time to be squeezed to ice, and the thicker the layer with interconnected spaces, the more the gravitational fractionation changes it from free air). We see no significant changes in this over the last 11,500 years of the current post-ice-age warm time. Hence, if the snowfall changed with the change in annual indicators in perfect concert with the changing penetration of cosmic rays, you also have to change the temperature in just the right way
to offset the effect of accumulation changes on the depth of the bubble close-off depth. You begin to see how totally ludicrous this becomes-for our dating to be significantly off, we would have to change so many, many other things in just the right ways at just the right times as to be ridiculous.

The odds of this are so close to zero as to be beyond consideration. As in all science, we don't really live or die on one observation, because there is always another interpretation of one observation--we base our results on an interlocking web of hundreds of different observations, anchored on especially strong pillars such as our ice-core records.

>4) The data in cores beyond this verifiable time are ambiguous but they would seem to indicate that the ice is much older that 11,500 years, even though it's difficult to know how to interpret this data.

Again, we have annual-layer counting to 40,000 years or so that is more accurate than any other dating technique, annual layers to 110,000 years but with more uncertainty owing to lower ice-age snowfall rates, stronger winds making higher snowdrifts, and ice-flow thinning of layers making them harder to see and count. Other ice cores go back as far as 420,000 years. Dating of the older ice is based on ice-flow modeling, on correlation to other records dated radiometrically, and on orbital tuning. (If you don't know that one, I can explain.) Agreement from these three different techniques causes us to consider it unambiguous. To those in the field, a U-Th date on a coral with all of the appropriate cautions is as reliable as an annual-layer count, or even more so. But annual-layer counts are easier for laypersons to believe. Let's face it--changing annual indicators is at least as easy as changing the physics enough to change the decay rates of all sorts of different isotopes, so that they all agree but all are wrong (10Be and 26Al and 36Cl and 234U and 40K and a number of others, all of which agree on ages of events)
>
Richard B. Alley, Evan Pugh Professor
Environment Institute and Department of Geosciences
The Pennsylvania State University
204A Deike Building
University Park, PA 16802, USA
ph. 814-863-1700; fax 814-865-3191; email ralley@essc.psu.edu
Some Questions about Dating in Death Valley, CA:

Dr Gunther Kletetchka, Geologist, and Dr Paul Houser, Hydrologist accompanied my "You Be The Scientist" Educational Outreach team on a distance learning field trip to Death Valley recently. Dr Gunther believes that the earth is very old and he dates the rocks accordingly. As we stand on a hilltop overlooking a valley near the lowest point in the Western Hemisphere, he explains that the surrounding mountains have eroded into the valley before us over millions of years. The older rocks we see in the remaining mountains are 60million years old. Some of the sediment rocks washing into the valley are still here today, because they were much harder than the surrounding material, which dissolved over time in the presence of water. Burying rocks in silts will stop the erosion processes from affecting the buried material indefinitely. Hence, the buried rocks will be preserved for eons and when they are eventually unburied, they will look young. We only know they are much older because of association with similar rocks that were uncovered earlier and have been eroding. Scientists calibrate the events and the resulting affects of erosion and put together a long baseline timeline. Then they fit everything into that timeline.

I ask, what will the mountains around us look like say 10, 000 years from now, given the current situation? They say that everything we now see will be completely gone in only 10,000 years. My question then is how can we accurately talk about the past in such long stretches of time, when it would seem that all records would be completely gone, long before we had a chance to observe them?

Dr Paul Houser, Hydrologist, feels that you can core down into the sediments found in this valley and that will produce a record of those historical events even far back on the order of 60 million years. Once buried, the records are preserved and if you can read them, then you have all the evidence you need of the old age of the earth.

Nevertheless, the mountains need a mechanism for replenishing themselves so that the erosion process can continue on newer material. For this the scientists theorize that there have been episodic upheaval events, which create new mountains. Do these events also destroy the records in the valleys? Scientists surmise that bits and pieces of the various stages of this record exist scattered all over the planet, but they feel they can piece them together effectively. It's quite an assumption and it requires enormous stretches of time to be plausible at all. Yet, during such enormously long time spans, would the records themselves be totally destroyed? Even if we keep the theory of evolution over enormous time spans as a plausible explanation, we still need to address the fundamental question of origins. Where did it all start? What were the initial conditions at that time? Even exchanging "time" for "God" as the primary cause, does not exempt the diligent scientist from needing the same amount of pure faith that a creationist would need. Both must explain the origin of it all. The only difference is that some think the First Cause was impersonal and others that it was personal.

My position is that our entire universe was created with the appearance of age from the start, as if a model were made of a pre-existing angelic situation elsewhere. Angels live on planets that are in the process of continuous change. So God models one of those planets at a given point in that process and everyone recognizes it as actively changing in a typical fashion. From the point of its creation onward, our scientists can follow the processes, but they can't work backwards beyond the starting point. The evidence appears to carry the record farther back, but angels who watched the creation of this model planet know better. God had to start with something recognizable, but whether it was an egg or a chicken, a tree or a seed, a planet or some fields, it would have the appearance of age.

25 August 2001

ß back home next à

For more on this and a response to any questions, please email any comments to nasamike@nasamike.com