Wrangel Island
Iceland
Svalbard
India
Far Eastern Russia
Little Diomede Island
Ellesmere Island
Kamchatka
Cosmogenic Isotope Research
Abstract Objectives and Paper Rationale Background Factors That Affect Cosmic Ray Production Rates Assumptions Production of Cosmogenic Isotopes Interpretation of Ages Methodology Experiments Results Conclusions ReferencesQuaternary Research
Center
Testing the sensitivity of two 36Cl age calculation programs
This is work that I did while completing my PhD at the University of Massachusetts.
For text, figures and raw data please contact Gualtieri directly. lyn4@u.washington.edu
Astract
This research highlights the differences
between two computer programs used to calculate ages from 36Cl
ratios. Specifically, samples from Far
Eastern Russia were used to show how changes in certain parameters
quantitatively affect calculated sample 36Cl age.
Experiments were run in CHLOE (NMex. Tech)
and RICH (PRIME Lab), two computer programs that are used to evaluate the
changes and test the sensitivity of the age on parameters other than production
rate of 36Cl. The two
programs assume different production rates and model parameters that affect
production rate differently, resulting in age discrepancies of up to 39%
between CHLOE and RICH. In some
experiments, the direction of the age change (increase or decrease) is opposite
in the two programs. This research
serves to link physicists, mathematical models, and computer programs to the
geologist, and to bring attention to the potential problems involved in
interpreting and reconstructing glacial advances based on 36Cl ages.
It is widely accepted that disagreement and inconsistencies in production rates
of cosmogenically produced 36Cl have the most significant effect on
age estimates. Production rate
estimates vary by up to 25% (Gosse et al.,1996a
and references therein). Other factors
that affect the production rate of 36Cl (elevation, latitude,
intensity of magnetic field) and also 36Cl ages are well-known and
have been mathematically modeled; however, these models have not been
well-tested using samples from a variety of sampling locations.Objectives
and Paper Rationale
The
main objective of this paper is to demonstrate, from a geologist’s perspective,
the potential inconsistencies in interpreting and reporting realistic 36Cl
ages. It is widely accepted that
disagreements exist amongst the cosmogenic community of physicists, chemists
and geologists regarding production rates of cosmogenic isotopes (Evans
et al.,1996; Gosse et al.,1996a; Swanson, 1996). Understanding, critiquing and discussing
these differences was one of the goals at the Workshop on Secular Variations in
the Rates of Production of Cosmogenic Nuclides On Earth, held in Santa Fe, New
Mexico in 1996. However, geologists and
others who are presently using, or thinking about using cosmogenic isotopes to
answer paleoclimatic and geomorphic questions should be aware of the
differences regarding production rates, as well as the factors that affect
production rate and 36Cl age (c.f. Klein and Gosse, 1996). It should
be clear that both programs discussed in this paper use different production
rates and user’s of these programs must be aware of this when reporting and
comparing 36Cl cosmogenic isotope ages. The dependence of certain factors on cosmogenic isotope production rate is mathematically known and has been modeled; however, these models have not been rigorously employed using “real-world” samples from a variety of geologic environments. Particularly in the Arctic, some variable parameters (snow shielding and water content of the rock and soil) may have a significant impact on age interpretation. Arctic geologic processes, such as frost heaving or frost shattering also need to be incorporated into erosion rates, shielding and sample depth. Not only are there differences in opinion regarding production rate, but differences also exist in the modeling of parameters that affect production rate. This is important for geologists to understand when using cosmogenic isotope ages themselves to try to interpret geomorphic relationships on ice-free vs. ice-covered landscapes, for example.
This research grew as we initially attempted to compare two published programs (CHLOE and RICH) that incorporate parameters affecting production rate to calculate ages from normalized radio nuclide/stable nuclide (NR/S) chlorine-36 ratios. This paper’s purpose is to demonstrate to the geologic community significant differences in available age programs used to interpret the timing of glacial events. The paper objectively evaluates and compares the two published programs, CHLOE and RICH using 28 samples from Far Eastern Russia.
Background
In 1955, Davis and
Schaeffer proposed a method for dating landforms using stable and radiogenic
nuclides formed in situ in rocks from
cosmic rays. The theory of dating the surface exposure of landforms from in situ
produced cosmogenic rays is
based on the understanding that cosmic rays induce nuclear reactions in the top
few meters of the earth’s surface, producing cosmogenic isotopes, such as 36Cl,
10Be, 26Al and 3He. Over time, the concentration
of cosmogenic isotopes increases and
can be measured in collected samples using Accelerator Mass Spectrometry (AMS).
The abundance of in situ produced
cosmogenic isotopes is proportional to the amount of time a surface, or a rock,
has been exposed. Cosmogenic isotopes can be used to date landform surfaces
directly and quantitatively (Hallet and Putkonen, 1994) thereby making it an
exciting and robust Quaternary dating technique. Gosse et al. (1996b) conclude that despite 20%
inaccuracies, cosmogenic age estimates are accurate enough to establish rough
ages for Quaternary events, although Clark et
al. (1995) and Swanson (1996) suggest that cosmogenic isotopes should not
be used to calculate short-term climatic episodes or events on a 500-1000 year
timescale. The technique is useful from
a few thousand to a few million years, with 5-20% precision. It is best to use a radionuclide that is not
produced by terrestrial nuclear reactions, and the choice of nuclides depends
on expected surface exposure ages, mineralogy and chemistry of the
samples. The isotope used in this study
is 36Cl (T 1/2= 300,000 yr).For the purpose of this research, 36Cl was used to date the surface of glacial moraines and glaciofluvial terraces. The moraines and terraces best record the terrestrial record of glaciation in Far Eastern Russia (Fig. 2.1). Minimum ages were obtained on the dates of deglaciation throughout the middle to late Pleistocene in the Pekulney and Koryak Mountain Ranges (Gualtieri et al, 2000). The premise in using cosmogenic isotopes in this context is that during moraine construction glaciers transport and deposit erratics previously shielded from cosmic rays. Glacial erosion rates depend heavily on underlying rock lithology, but even a conservative estimate of 0.5 m/ka would be enough to “zero” any rock with inherited 36Cl due to a prior exposure history. When the ice melts, the rocks are uncovered and exposed to cosmic rays and cosmogenic isotope accumulation begins. Bierman and Turner (1995) suggest that isotopic abundances should be interpreted as minimum limiting model exposure ages or maximum limiting erosion rates.
Factors That Affect Cosmic Ray Production Rates
The production
rate for any one nuclide at any time and location is not well known because
only limited data exist (Gosse et al.,1996a). Also, production rates change due to
temporal fluctuations in the cosmic ray flux due to variations in the
paleogeomagnetic field strength, dipole axis position, solar modulation and
atmospheric pressure (Gosse et al.,1996a);
as well as altitude, latitude, depth of sample, target nucleus and the energy
of the bombarding particle (Dep, 1995) and possibly global climate change
(Plummer et al., 1997).Changes in the earth's magnetic dipole field strength are the most significant source of temporal variability in production rates (Gosse et al., 1996b). At a given magnetic latitude, only cosmic rays with energies above the magnetic cut-off energy for that latitude enter the atmosphere. The magnetic field acts as a filter, being stronger at the equator because the magnetic field lines are parallel to the surface there, whereas they are perpendicular at the poles (Dep, 1995). There is an apparent decrease in the production of 36Cl over the last 30,000 yr in the low and mid-latitudes; however, this decrease is not recorded in polar records. This decrease is due to the slow increase in the strength of the global geomagnetic dipole between 5000 – 30,000 yr (Bard, 1997). For latitudes poleward of 20°, another important factor is the position of the magnetic pole relative to the geographic pole (Klein and Gosse, 1996).
Assumptions
The following
assumptions are inherent to using in situ
produced cosmogenic isotopes as a surface exposure dating technique (Kurz and
Brook, 1994).1. The sample has been eroded sufficiently prior to deposition to “zero” it of inherited cosmogenic isotope accumulation, or the sample has never been exposed.
2. The rock underwent sudden exposure.
3. Cosmic ray flux is constant with time. If this is true, one can assume that the concentration of accumulated cosmogenic nuclides within the surficial rock is directly related to the time the surface has been exposed.
4. Production rate is known.
5. The sample has been uniformly exposed to cosmic rays (no erosion, snow, water, soil or vegetation cover), or the duration and nature of cover must be accountable.
6. There has been no loss or contamination of 36Cl during sample exposure history.
Production of Cosmogenic Isotopes
In order to use
cosmogenic isotopes effectively to solve geomorphological problems, the isotope’s
production rate must be well constrained. The production rate is largely
dependent on latitude as well as changes in the paleointensity of the earth's
geomagnetic field (Lal, 1991). Other factors that affect isotopic production
rates are discussed in the following section. Mechanisms for the production of
cosmogenic isotopes are described below. Secondary galactic rays penetrate the atmosphere and bombard the surface of the earth. Once the rays hit the surface, the flux of the more reactive nucleonic particles is attenuated within a short distance due to the increased density of rock to air. Cosmogenic nuclides are produced from three reactions: neutron and proton spallation, thermal neutron activation and negative muon capture. Chlorine-36 is produced by all three types of reactions, with target atoms of 39K, 40Ca and 35Cl. Sixteen to 80% percent of 36Cl is produced from spallation of 39K and 40Ca, 11-80% by thermal neutron activation of 35Cl, and 0-10% by negative muon capture by Ca (Zreda and Phillips, 1994).
Spallation is a nuclear reaction in which many particles are ejected from an atomic nucleus by an incident particle of sufficiently high energy. Nuclides that are produced by spallation have a constant production rate in the top 20g/cm2 then decrease exponentially with depth (Zreda and Phillips, 1994). Chlorine-36 has a more complex depth dependency because of the significant portion of it being formed by thermal neutron capture. Thermal neutron activation of 35Cl produces 36Cl with a profile that first increases with depth, reaches a maximum at 65g/cm2 and then decreases exponentially below ca. 100g/cm2 .
The buildup of cosmogenic isotopes in terrestrial rocks is a function of production rates and radioactive decay (Zreda and Phillips, 1994). In surface rocks, buildup is fastest during the initial period of exposure and then decreases with time and eventually reaches a steady state, when cosmogenic production equals radioactive decay (Zreda and Phillips, 1994).
Accurate nuclide production rate estimates are generally not possible due to the lack of knowledge concerning the probabilities of formation in the different reactions (Lal, 1991). However, some commonly used production rates are 7550 (560 atoms/mole K yr for spallation of K to produce 36Cl (Zreda et al., 1994); 3050 (200 atoms/mole Ca yr for spallation of Ca to produce 36Cl (Zreda et al., 1991); 307 (24 neutrons/g rock yr for neutron and muon capture of Cl to produce 36Cl; (Nishiizumi et al., 1989). Discrepancies in production rates are mainly due to the inability to model certain factors that affect production rate as well as the weights given by authors to multiple variables. A list of the reasons for these discrepancies is found in Gosse et al. (1996a).
Interpretation of Ages
Interpreting
absolute ages from cosmogenic isotope data is not straightforward. The error of
exposure time calculation is a combination of analytical errors associated with
isotopic and chemical analyses (5-10%) as well as uncertainties in the
production rate and factors that affect production rate, such as erosion rate
and snow shielding (Zreda and Phillips, 1994). This paper deals with the errors
(mostly geological factors) that the Quaternary geologist is responsible for
when collecting and sending off samples to be analyzed. Although not the focus
of this paper, it should be noted that there are significant systematic errors
(i.e. production rates, latitude, altitude, and assumptions about the physics
of isotope production) that the user must be aware of. Recently, on the Internet, two computer programs became available that incorporate the factors affecting surface exposure age in age equations. CHLOE (CHLorine-36 Exposure Age Calculation Workbook) is an excel-based spreadsheet program produced by New Mexico Tech (Phillips and Plummer, 1996) used mainly for calculating 36Cl ages, and facilitating the calculation of ages due to erosion, irregular surface geometries and snow shielding. The December 1997 version was used for this paper. CHLOE is available on the Internet at http://griffy.nmt.edu/~mplummer/chloe/chloe.html.
RICH (Rock In-Situ Produced Cosmogenic-nuclide History) was produced by PRIME Lab (Purdue Rare Isotope Measurement Laboratory) (Dunne et al., 1996). This program can incorporate up to 100 pieces of data for each sample and can be used with multiple isotopes. The December 1997 version was used for this paper. RICH is available at http://primelab.physics.purdue.edu/.
Methodology
Field Techniques
Sample collection and preparation are critical steps in utilizing cosmogenic isotopes to obtain realistic surface exposure ages. A sampling strategy was developed after discussion of the project with many researchers currently working with cosmogenic isotopes (Paul Bierman, Univ. of Vermont; David Elmore, PRIME Lab/Purdue University; John Gosse, Univ. of Kansas; Mark Kurz, Wood’s Hole Oceanographic Institute; Fred Phillips, New Mexico Institute of Mining and Technology; Marek Zreda, Univ. of Arizona). A "Cosmogenic Checklist" (Table 2.1) was derived and these measurements and observations were used to aid in the final interpretation of exposure ages.
Multiple rock samples from stable erratics on moraine crests and terraces were first priority for age dating. One to four ~ 1 kg samples per geomorphic landform (usually a moraine or terrace) were collected. Examples of erratic size, amount of surface dip and weathering are shown in Figure 2.2 a-c. Wherever possible, samples were collected from the upper 5-10 cm of the highest and most stable rock on moraine crests. Samples were collected as far away from the edges of the rock as possible, so as to minimize the loss of 36Cl produced by thermal neutron activation. If erratics were absent, glacially scoured bedrock surfaces were sampled.
Laboratory Techniques
Rocks were first cleaned using a wire brush to remove lichen. Samples were then crushed to “pea size” using either a mortar and pestle or a Sturtevant jaw crusher and then pulverized into a powder using a shatterbox. Complete sample procedures for elemental analysis followed the sample preparation for the Ronald B. Gilmore X-ray Analytical Facilities at the University of Massachusetts (UMASS). Two aliquots of the powdered fraction were retained. One aliquot was kept at UMASS for XRF elemental analysis, and the other aliquot was sent to the X-ray Assay Laboratories (XRAL, Toronto) for prompt gamma neutron analysis for gadolinium (Gd) and boron (B). Gadolinium and boron have a high cross section, or probability for absorbing slow neutrons during initial cosmogenic ray bombardment; therefore, the concentration of Gd and B, as well as other target atoms must be known for final data interpretation and age assignment. The final step in the physical sample preparation is to clean the powdered sample using 10% nitric acid. Laboratory guidelines follow PRIME Lab Chemistry Operations Worksheet AW0009 (Appendix B). The cleaned sample was sent to PRIME Lab at Purdue University for chemical preparation. The full procedure, performed at PRIME Lab, follows Chemistry Operations Worksheets AW0012 and AW0002 (Appendix B) and is summarized in Figure 2.3. Chloride concentrations were determined at PRIME Lab by the ion selective electrode method. Accelerator Mass Spectrometry (AMS) measurements for whole rock 36Cl analysis were done at PRIME Lab, Purdue University. For a comprehensive discussion and description of the facility see Dep (1995) and the PRIME Lab web site at http://primelab.physics.purdue.edu.
Accelerator Mass Spectrometry (AMS)
After target preparation, samples were ready for AMS analysis. The AMS system at PRIME Lab is based on a tandem accelerator with a spherical ionizer cesium sputter source with an eight-sample wheel (Elmore et al., 1996, Appendix A). The multi-plate gas ionization detector separates interfering isobars (36S) and counts the ions. The ions are collected on metal plates, amplified and read into the computer (Elmore et al., 1996).
Experiments
The purpose of the
experiments was to quantitatively assess how changes in parameters that have
been mathematically modeled actually affect 36Cl ages using “real
world” samples. Essentially all parameters that were variable in CHLOE and RICH
were tested. Only one parameter was
changed at a time, and all other parameters were set to the original values
(Table 2.2) which reflect the best-estimated values. The percent change in age
in CHLOE was compared to RICH for the tested parameters. Percent change was
also compared in CHLOE and RICH for a relatively “young” sample (16,800 –
20,400 yr) and a relatively “old” sample (73,500 – 95,700 yr). The direction of
change was also tested (did ages increase or decrease?). The original input data used in both programs is shown in Table 2.2 and Appendices C, D and E and the comparisons between CHLOE and RICH for all 28 samples are shown in Table 2.3. Below is a description and explanation of the experiments (Tables 2.4a and b). Documentation of modeling experiments on parameters that affect 36Cl age and/or production rate are also given.
Chlorine only
This experiment was only done in RICH in order to quantitatively establish how imprecise an age would be if obtained using minimal data and no “geologic” correction factors. Input parameters include sample latitude and chlorine information: mass of sample, measured 36Cl/Cl ratio for AMS sample, 35Cl/37Cl ratio for chloride in AMS sample, mass of chloride added by carrier, 35Cl/37Cl from AMS in chemical processing blank, and mass of chloride in carrier in chemical processing blank. Data were not entered for shielding, elevation, erosion or elemental analysis.
Maximum NR/S
The maximum normalized radio nuclide/stable nuclide ratio in the sample and as reported from PRIME Lab AMS Facility.
Minimum NR/S
The minimum normalized radio/stable nuclide ratio in the sample and as reported from PRIME Lab AMS Facility.
Latitude, degrees
Production rates of 36Cl increase with latitude due to a decrease in the strength of the earth’s magnetic field (Lal and Peters, 1967; Lal, 1991). This experiment compares the effect of latitude on the relative amount of age change between a young sample and an old sample, and also demonstrates the resulting age difference in the way CHLOE and RICH model this parameter.
Altitude, m
Production rates of 36Cl increase with elevation due to attenuation of cosmogenic rays in the atmosphere (Lal and Peters, 1967; Lal, 1991). This experiment compares the amount of change between a young sample and an old sample, and also demonstrates the resulting age difference in the way CHLOE and RICH model this parameter.
Erosion rate, mm/ka
Lal (1991), Zreda et al. (1994) and Dep et al. (1997) demonstrated that surface erosion yields an incorrect 36Cl age. Bierman et al. (1995) and Lal (1991) modeled erosion rates from the abundance of 36Cl. Ideally, each study site and lithology should have an independent control or method of interpreting the erosion rate; however, this is not always possible. In coastal areas of Far Eastern Russia, salt weathering may play an important factor in chemical erosion. The duration and type of vegetation cover on a sampling surface also affects the erosion rate throughout the sample’s history. Some of the samples collected for this study were covered with lichens, which will trap more water and expedite weathering and disaggregation. On the other hand, the presence of a weathering rind should be noted, as this may slow down the weathering process (Colman, 1981). The erosion rate used in the original input data represents a realistic range for Far Eastern Russia. The range of estimates we used here is based on field observations, including the maximum amount of surface relief of individual boulders. As shown in Table 2.5, the estimates fall within the range of independently obtained erosion rates for other arctic areas. Vegetation type, annual duration and amount of cover probably varied throughout a sample’s history and should be taken into account in terms of effects on rock erosion rate. For example, Pinus pumilla, which covered a significant portion of Far Eastern Russia throughout Pleistocene time, requires a specific snow cover in order to survive (Andreev, 1980). The erosion rate used in the original input data (1.5 mm/ka) represents a realistic range for Far Eastern Russia based on field observations, including the maximum amount of surface relief of individual boulders. As shown in Table 2.5, the estimates fall within the range of independently obtained erosion rates for other arctic areas.
Sample Depth, cm
Dep et al. (1994) and Liu et al. (1994) measured and predicted the depth dependence of 36Cl produced from low energy neutron capture and showed a pattern of production with depth. For part of its exposure history, a sample could, for example, have been buried deeper beneath the modern surface. This scenario is not entirely modeled with the erosion rate correction. For example, it is likely that during one arctic winter throughout the sample’s exposure history a significant outer portion of the sample may have spalled off due to physical weathering processes, such as frost shattering. Examples of this phenomenon are commonly observed in modern arctic landscapes and as a result, each rock sample should be evaluated individually. For example, in the Bendeleben Mountains and Brooks Range, Alaska, mineralogy and grain size are critical factors controlling erosion and boulder downwasting (Hamilton, T., personal communication, 1997).
Shielding data
This parameter represents the amount of horizon shielding on the sample. The notation 0,360,0 indicates that from 0-360° on the horizon the angle between the sample and the horizon is 0°. In this case, there would be no shielding effect. The notation 0,360,45 indicates that from 0-360°on the horizon the angle between the sample and the horizon is 45°. More complex shielding geometries are optional in both CHLOE and RICH and were used in calculating the most realistic ages for individual samples. In the experiments, extreme values are used in order to demonstrate the quantitative affects of shielding on age. Dep (1995) and Dunne et al., (1998) mathematically modeled the effects of past geometric shielding.
Snow shielding, days/depth
This parameter represents the amount of snow cover shielding the sample. Fifty-five year modern snow depth and duration data (Historical Soviet Daily Snow Cover Studies) from two World Meteorological Stations: Anadyr (64.8°N, 177.6°E) and Markova (64.7°N, 170.4°E) were used (Appendix F). The number of days per year with > 50 cm of snow on the ground was used to determine snow shielding. The amount of snow is converted to its water equivalent assuming a snow density of 0.4g/cm3. For further information regarding correction factors, see documentation in CHLOE (http://griffy.nmt.edu/~mplummer/chloe/chloe.html). In Tables 2.4a and b the notation 0/50 translates to: for 0 days of the year there is 50 cm of snow covering the sample. A variety of snow shielding scenarios was used in order to assess the sensitivity of the snow cover parameter on age. For some of the tests, for example, 210/300 (210 days per year of 300 cm of snow) glacial or periglacial conditions were simulated. The persistence of year-round snow on some boulders may “protect” the boulders from mechanical erosion; however, the role of the wind on the vast tundra and in the narrow mountain passes, where some of the samples were collected, may play an important role in removing the snow cover. The absence of lichens and persistence of wind should be incorporated into paleosnow cover estimates. Dep (1995) modeled the effect of snow, in the form of 20 cm of water, on a concrete block and on granite and determined that there is an initial build-up of the thermal neutron flux below the surface, but at depth (40 g/cm2) neutron-capture production rates are nearly doubled.
Dip angle for surface, degrees
Dunne et al. (1998) modeled the effect of sampling surface/rock surface dip on 36Cl age. A dipping surface will be shielded from cosmic rays and therefore 36Cl production will be less than on a horizontal surface. It is likely that samples on the crests of moraines in Far Eastern Russia may have experienced changes in the dip of their surface throughout their exposure history due to the effects of frost heave, slumping and gelifluction, which can happen on slopes as low as 1°.
Rock Density, g/cm3
Previous studies modeling the effects of rock density on 36Cl age have not been done; however the relationship between rock density and the apparent vertical attenuation length (which affects production rate) are known (Kurz and Brook, 1994). Rock density also affects erosion rate.
Rock Water Content, %
This is given as volume. Dep et al. (1994) modeled the thermal neutron flux and 36Cl production in granite for rock water contents of 1%, 2% and 4%. The 36Cl production rate increases with increasing rock water content (Dep et al., 1994).
Neutron Leakage Factor
This variable reflects the location on the rock surface of the sampling surface where the sample was taken (edge vs. center). A value of 1 would indicate that the sample came from the middle of a rock sampling surface and correction for this effect would not be necessary. Values less than 1 indicate that the sample came from the edge of a boulder. If the sample came from the edge of the boulder neutrons produced by thermal neutron capture could have “leaked” out, and a correction must be made for this. This experiment was only done in CHLOE. For documentation regarding the mathematical modeling of this parameter, see CHLOE (http//griffy.nmt.edu/~mplummer/chloe/chloe.html).
Elemental Experiments
Assuming the neutron absorption cross-sections of Mughaghab and Garber (1973), the following elements have the seven highest total macroscopic thermal neutron absorption cross sections, or probability to absorb 36Cl neutrons during production. The amount of each of the following seven elements was doubled (denoted by 200) and halved (denoted by 50) in order to test the sensitivity of ages on varying whole rock elemental analyses.
Gadolinium, % -The amount of gadolinium was doubled and halved.
Manganese, % -The amount of manganese was doubled and halved.
Iron, % -The amount of iron was doubled and halved.
Silicon, % -The amount of silicon was doubled and halved.
Calcium, % -The amount of calcium was doubled and halved. Calcium is a target nuclide for the production of 36Cl.
Potassium, % -The amount of potassium was doubled and halved. Potassium is a target nuclide for the production of 36Cl.
Boron, % -The amount of boron was doubled and halved.
Results
A compilation of
the results of the experiments is shown in Tables 2.4a-b and in Figures 2.4 and
2.5. Below are summarized some of the larger and surprising discrepancies
between the two programs. 1. In all cases, CHLOE ages are older than RICH ages. The average age discrepancy, for the Far Eastern Russia database is 17%, but can be as high as 39% (for a sample 10,000 –17,000 yr) for the Far Eastern Russia samples. The smallest discrepancy between CHLOE and RICH ages was 4% (Table 2.3).
2. There is no biasing in the age discrepancy between CHLOE and RICH when comparing young and old samples. In other words, on older samples, CHLOE and RICH produce ages that differ by the same percentage as on younger samples. When comparing younger ages (Kuveveem River Valley, Cape Dionysia, Nygchekveem River Valley, Nahodka Valley, Lake Mainitz, Lake Rocamaha) CHLOE yields ages older by 6-39% with an average age discrepancy of 15%. When comparing older ages (Tanyurer River Valley) CHLOE yields ages older by 16-27% with an average age discrepancy of 21% (Table 2.3).
3. In both programs, entering extreme values for elevation and latitude changes the age by >50%, as expected (Tables 2.4a,b, Fig. 2.4 and 2.5).
4. In CHLOE, entering an extreme value (89°) for dip angle of surface, changes the age by >50% (Tables 2.4a,b, Fig. 2.4 and 2.5).
5. Halving the potassium concentration in a young sample in CHLOE increases the age by 50%; however, halving the potassium concentration of the same sample in RICH decreases the age by 0.5% (Table 2.4a and Fig. 2.4).
6. Doubling the potassium concentration in a young sample in CHLOE decreases the age by 40%; however, doubling the potassium concentration of the same sample in RICH increases the age by 0.9% (Table 2.4a and Fig. 2.4).
7. Halving the potassium concentration in an old sample in CHLOE increases the age by 22%; however, halving the potassium concentration of the same sample in RICH decreases the age by 0.5% (Table 2.4b and Fig. 2.5).
8. Doubling the potassium concentration in an old sample in CHLOE decreases the age by 26%; however, doubling the potassium concentration of the same sample in RICH increases the age by 0.9% (Table 2.4b and Fig. 2.5).
9. Halving the calcium concentration in an old sample in CHLOE increases the age by 43%; however, halving the calcium concentration of the same sample in RICH decreases the age by 0.3% (Table 2.4b and Fig. 2.5).
10. Doubling the calcium concentration in an old sample in CHLOE decreases the age by 38%; however, doubling the calcium concentration of the same sample in RICH increases the age by 0.6% (Table 2.4b and Fig. 2.5).
11. Even though RICH can incorporate up to 83 elemental analyses, it is demonstrated that even by doubling and halving the seven elements with the highest neutron cross-sections, RICH ages were changed by only 2% (Table 2.4a,b, Figs. 2.4 and 2.5). Uncertainties in production rate of up to 20% will overshadow this in a final age calculation.
12. An increase in rock density for an old sample increases the CHLOE age, whereas an increase in rock density in the same sample decreases the RICH age (Table 2.4b).
13. For a young sample, both RICH and CHLOE ages increase (Table 2.4a). Even though the amount of age change is insignificant in both cases, it is interesting to note that the programs do not agree in the direction of change.
14. Aside from the disagreement in direction of change for calcium and potassium, RICH and CHLOE also disagree in the direction of change for halving the boron concentration. In CHLOE, the age decreases with a decrease in the boron concentration; whereas in RICH the age increases with a decrease in boron concentration (Table 2.4a,b, Figs. 2.4 and 2.5).
15. Entering only information regarding the chlorine concentration and not making any "geological" corrections resulted in a RICH age 20-24% too young (Table 2.4a and b).
Even when entering extreme values for elemental compositions in both programs for both a young and old sample, the ages only changed by 2% or less (Table 2.4a,b, Figs. 2.4 and 2.5).
16. CHLOE ages changed by only 2-4% when 180 days of 100 cm snow cover was tested. This is due to the size (2 m) of the boulders. If there is year-round snow cover yet it does not actually cover the sampling surface, then correction for snow cover is insignificant. There would be a greater effect of snow cover for small samples. However, simulated glacial or periglacial conditions (210 days of 300 cm of snow) increased the young sample age by 19%. An integrated approach as to the snow conditions during the samples’ exposure history must be considered.
The Differences between CHLOE and RICH
Aside from assuming different production rates of 36Cl, a fundamental difference between CHLOE and RICH is that they utilize different models of thermal neutron production (Dunne, A., personal communication, 1997). CHLOE uses a diffusion model developed by Fred Phillips (Phillips and Plummer, 1996) whereas RICH uses a model based on Dep's (1995) work and a Monte Carlo code for neutron and photon transport (MCNP). Another difference between the two programs is the Ca and K elemental input data. In RICH, both pre and post leaching values of K and Ca are used. In CHLOE, only the post leaching values are used. A summary of the differences and input parameters from the user’s perspective is shown in Figure 2.6.
A Suggestion for Interpreting and Reporting 36Cl Ages
The most realistic age can be obtained by assigning a reasonable range of estimates to the most important local factors affecting production rate (snow shielding, erosion etc...) as well as age calculations. The age ranges can be plotted together on an age vs. parameter graph (Fig. 2.7). The most likely age should be plotted as a range encompassing a range of reasonable ages for a given set of the most variable parameters (Fig. 2.7). Also, at least two programs should be used to report 36Cl ages with all input data given.
Conclusions
1. There are significant
differences between CHLOE and RICH and probably other programs that need to be
reconciled.2. More than one program must be used in order to obtain the most realistic 36Cl age.
3. The most realistic age should be reported as a range of ages.
Even though inputting in some cases “extreme” values for certain parameters that affect 36Cl age, the overall error for age interpretation will be largely dependent on inconsistencies and disagreement of production rate. However, in certain areas, such as arctic environments, local factors, such as snow cover or erosion may far outweigh regional factors or even production rate variations. The most critical test would be to investigate scenarios of varying erosion and snow cover and to present ages as ranges. Numerical ages should not be an over-interpretation of non-linear geologic processes, such as erosion, leaching, weathering or contamination.
Implications for Future Studies
It would be useful for other researchers of the cosmogenic community to test CHLOE and RICH and other programs using their own “real world” data in order to determine where the modeling inconsistencies lie. Modeling can only go so far. Do the models work with real samples? These data are useful for everyone using exposure age dating and will help to standardize procedures for interpreting and reporting 36Cl ages (much like the 14C database) as well as standardized publishing procedures. The suggestions of Gosse et al. (1996a) as well as age uncertainties should be reported by all authors presenting new cosmogenic isotope ages. Assumptions regarding erosion rate, sample depth, horizon and snow shielding should also be carefully documented. Ages that are reported as a range of realistic values using a variety of age calculation programs and techniques will be most applicable to geologic interpretation. Otherwise, as the cosmogenic community grows, it will become impossible to compare ages from sites around the world.
We can use CHLOE and RICH and other programs to help determine unknown relationships regarding production rate factors. We can only go so far with the models until we have to return to real world data to analyze the causal relationships. Different production rates may lead to over-interpretation and drastically different ages. The worst-case scenario could be up to a 39% difference (shown in this paper) which would provide enough ambiguity so that ages could not be correlated with the proper oxygen isotope stage. For example, an event, based on 36Cl exposure age dating, could pre-date Stage 5 using one age calculation program and post-date Stage 5 using another age calculation program. The incorrect correlation of an event with pre- or post- Stage 5 could lead to problems in paleoclimatic correlation with other terrestrial records. If these problems are not reconciled it will become more difficult to correlate these low resolution terrestrial dating techniques globally with other terrestrial records, and also among high resolution marine and terrestrial records. The hope in writing this paper is to urge users of cosmogenic isotope analysis to critically test CHLOE and RICH and other programs in order to progress further towards a universal understanding regarding modeling of production rate and factors affecting production rate.
References
Andrè,
M.F. 1996. Rock Weathering rates in arctic and subarctic environments (Abisko
Mtns., Swedish Lappland). Zeitschrift für Geomorphologie 40: 499-517.Andreev, V.N. 1980. “Vegetation and Soils of Subarctic Tundra. Nauka. U.S.S.R.” Academy of Science, Siberian Sub-branch Biological Institute. Novosibirsk. [in Russian].
Bard, E. 1997. Nuclide Production by Cosmic Rays During the Last Ice Age. Science 277: 532-533.
Bierman, P. and Caffee, M. 1994. Cosmogenic Erosion Rate Estimates for Granite Landforms; Eyre Peninsula, South Australia. GSA Abstracts with Programs. 26: (7) 256.
Bierman, P. and Turner, J. 1995. 10Be and 26Al Evidence for Exceptionally Low Rates of Australian Bedrock Erosion and the Likely Existence of Pre-Pleistocene Landscapes. Quaternary Research. 44: 378-382.
Bierman, P., Gillespie, A., Caffee, M, and Elmore, D. 1995. Estimating erosion rates and exposure ages with 36Cl produced by neutron activation. Geochimica et Cosmochimica 59:3779-3798.
Clark, D., Bierman, P., and Larsen, P. 1995. Improving in situ Cosmogenic Chronometers. Quaternary Research44: 367-377.
Clement, B., Landry, B. and Yergeau, M. 1976. Dèchaussement postglaciaire de filons de quartz dans les Appalaches Quèbecoises. Geografiska Annaler58A: 111-114.
Colman, S. 1981. Rock-Weathering Rates as Functions of Time. Quaternary Research 15: 250-264.
Davis, Jr., R., and Schaeffer, O.A. 1955. Chlorine-36 in Nature. Annals of the New York Academy of Science. 62:105-122.
Dep, L., Elmore, D., Fabryka-Martin, J., Masarik, J. and Reedy, R.C. 1994. Production rate systematics on in situ-produced cosmogenic nuclides in terrestrial rocks: Monte Carlo approach of investigating 35Cl (n, gamma) 36Cl. Nuclear Instruments and Methods in Physics ResearchB92:321-325.
Dep, L. 1995. Cosmogenic Radionuclide Production in Terrestrial Rocks: Accelerator Mass Spectrometry Measurements and Monte Carlo Simulations. Unpublished Ph.D. Dissertation. Purdue University.
Dep, L., Elmore, D., Vogt, S., Sharma, P., Bourgeois, M. and Dunne, A.1997. Erosion-corrected ages of Quaternary Geomorphic Events Using Cosmogenic 36Cl in Rock. Prime Lab Report: 1-13.
Dunne, A., Elmore, D. and Dep, L. 1996. A Server-based Code for In-Situ-Produced Nuclides that Incorporates Irregular Geometries. Radiocarbon38: 25.
Dunne, A., Elmore, D. and Muzikar, P. (in press). Scaling of cosmogenic nuclide production rates for geometric shielding and attenuation at depth on sloped surfaces. Geomorphology.
Elmore, D., Lipshutz, M., Ma, X., Miller, T., Mueller, K., Perry, M., Rickey, F., Sharma, P., Simms, P. and Vogt, S. 1996. Status and Plans for the PRIME Lab AMS Facility. Abstracts of the 7th International AMS Conference. Tucson. May 20-24. p. 27.
Evans, J.M., Stone, J.O.H., Fifield, L.K. and Cresswell, R.G. 1996. Cosmogenic Chlorine-36 Production Rates from Potassium. Radiocarbon38: 29.
Gosse, J.C., Reedy, R.C., Harrington, C.D. and Poths, J. 1996a. Overview of the Workshop on Secular Variations in Production Rates of Cosmogenic Nuclides on Earth. Radiocarbon 38:135-147.
Gosse, J.C., Reedy, R., Harrington, C.D., Poths, J. and Gillespie, A.R. 1996b. Terrestrial Cosmogenic Nuclide Dating: Tested and Ready for Action. EOS 77: 275- 276.
Gualtieri, L., Glushkova, O. and Brigham-Grette, J. 2000. Evidence for restricted ice extent during the last glacial maximum in the Koryak Mountains of Chukotka, far eastern Russia. GSA Bulletin 112: 1106-1118.
Halflidason, H. 1983. The Marine Geology of Eyjafjördur, North Iceland: Sedimentological, petrographical and stratigraphical studies. Unpublished Mphil. Thesis. University of Edinburgh.
Hallet, B. and Putkonen, J. 1994. Surface dating of dynamic landforms: young boulders on aging moraines. Science265: 937-940.
Klein, J. and Gosse, J. 1996. Terrestrial Factors that Influence Production Rates. Radiocarbon 38: 170.
Kurz, M.D. and Brook, E.J. 1994. Surface Exposure Dating with Cosmogenic Nuclides. In Dating in Exposed and Surface Contexts. Edited by C. Beck. University of New Mexico Press. Albuquerque. pp. 139-159.
Lal, D. and Peters, B. 1967. Cosmic Ray Produced Radioactivity on Earth. In Encyclopedia of Physics. Edited by S. Fluegge. Springer Verlag, Berlin. pp 551-612.
Lal, D. 1991. Cosmic ray labeling of erosion surfaces: In situ nuclide production rates and erosion rates. Earth and Planetary Science Letters 104: 424-439.
Liu, B., Phillips, F.M., Fabryka-Martin, J.T., Fowler, M.M. and Stone, W.D. 1994. Cosmogenic 36Cl accumulation in unstable landforms I: Effect of the thermal neutron distribution. Water Resources Research 30:3115-3125.
Mughaghab, S.F. and Garber, D.I. 1973. Neutron Cross Sections, vol. 1, Resonance Parameters. Brookhaven National Laboratory.
Nishiizumi, K., Winterer, E.L., Pohl, C.P., Klein, J., Middleton, R., Lal, D. and Arnold, J. 1989. Cosmic Ray Production Rates of 10Be and 26Al in Quartz from Glacially Polished Rocks. Journal of Geophysical Research 94:907-916.
Phillips, F.M. and Plummer, M.A. 1996. CHLOE: A Program for Interpreting In-Situ Cosmogenic Nuclide Data for Surface Exposure Dating and Erosion Studies. Radiocarbon 38:98.
Plummer, M., Phillips, F.M., Fabryka-Martin, J., Turin, H.J., Wigrand, P.E. and Sharma, P. (1997). Chlorine-36 in Fossil Rat Urine: An Archive of Cosmogenic Nuclide Deposition During the Past 40,000 years. Science277:538-541.
Smith, D.I. 1972. The solution of limestone in an arctic environment. In Polar Geomorphology. Edited by D.E. Sugden. Inst. Br. Geographers. Publ. no. 4. pp.187-200.
Swanson, T. 1996. Determination of 36Cl Production rates from the Deglaciation History of Whidbey and Fidalgo Islands, Washington. Radiocarbon 38:172.
Zreda, M.G., Phillips, F.M., Elmore, D., Kubik, P.W., Sharma, P. and Dorn, R.I. 1991. Cosmogenic Chlorine-36 Production Rates in Terrestrial Rocks. Earth and Planetary Letters 105:94-109.
Zreda, M.G. and Phillips, F.M. 1994. Surface Exposure Dating by Cosmogenic Chlorine-36 Accumulation. In Dating in Exposed and Surface Contexts. Edited by C. Beck. University of New Mexico Press. Albuquerque. pp.161-183
Zreda, M.G., Phillips, F.M. and Elmore, D. 1994. Cosmogenic 36Cl Accumulation in Unstable Landforms: II. Monte Carlo Simulations and Experimental Observations on Eroding Glacial Moraines. Water Resources Research 30:3127-3136.