High-altitude, low-latitude calibration sites I
In the past few months two papers have described new Be-10 production rate calibration data from high-elevation, low-latitude sites in the tropics. One by Meredith Kelly and a bunch of co-authors:
Meredith A. Kelly, Thomas V. Lowell, Patrick J. Applegate, Fred M. Phillips, Joerg M. Schaefer, Colby A. Smith, Hanul Kim, Katherine C. Leonard, Adam M. Hudson, 2014 in press. A locally calibrated, late glacial 10Be production rate from a low-latitude, high-altitude site in the Peruvian Andes. Quaternary Geochronology.
Basically, this paper describes Be-10 concentrations in boulders on top of the “Huancané IIa” moraine of the Quelccaya ice cap in Peru (14° S, 4800 m). The moraine, being essentially formed in a giant peat bog and thus overlying, underlying, and including radiocarbon-dateable material, is extraordinarily well radiocarbon dated at 12350 cal yr BP (according to the INTCAL09 calibration).
The other is by PH Blard and co-authors:
P.-H. Blard, R. Braucher, J. Lavé, D. Bourlès, 2013. Cosmogenic 10Be production rate calibrated against 3He in the high Tropical Andes (3800–4900 m, 20–22° S). Earth and Planetary Science Letters, Volume 382, Pages 140-149
This one is less direct. The location is fairly similar (Uturuncu volcano in Bolivia, 22° S, 4800 m). However, the calibration is less direct. At this site, the data actually measured were Be-10 and He-3 concentrations in coexisting pyroxene and quartz. This yields a Be-10/He-3 production ratio. Getting a Be-10 production rate out of this relies on data from a separate paper:
P.-H. Blard, J. Lavé, F. Sylvestre, C.J. Placzek, C. Claude, V. Galy, T. Condom, B. Tibari, 2013. Cosmogenic 3He production rate in the high tropical Andes (3800 m, 20°S): Implications for the local last glacial maximum. Earth and Planetary Science Letters, Volumes 377–378, September 2013, Pages 260-275.
This paper includes He-3 measurements on boulders from an alluvial fan near Tunupa volcano (Bolivia, 20° S, 3800 m) that is independently radiocarbon dated at 15,300 cal yr BP. Thus, these authors obtain a Be-10 production rate by this reasoning: first, determine the He-3 production rate from the Tunupa site. Second, multiply by the Be-10/He-3 production ratio measured at the Uturuncu site to obtain the Be-10 production rate.
These studies are important because cosmogenic-nuclide production rates in the high tropics aren’t very well constrained. As discussed at length in our 2008 paper, most existing Be-10 calibration data are at mid- to high latitude and relatively low elevation, so estimating production rates at low latitude and high elevation requires large extrapolations in both geomagnetic cutoff rigidity (it’s a lot higher at low latitude) and atmospheric pressure (it’s a lot lower at high elevation). Existing scaling schemes handle this extrapolation very differently. Until now, there’s only been one set of high altitude/low latitude calibration data, from work by Farber and others in the Peruvian Cordillera Blanca (10° S, 4000 m). This site, again, is a direct calibration via Be-10 measurements on boulders on top of a moraine with a reasonably well constrained radiocarbon age of 13060 cal yr BP (this reflects some updates from the original paper). However, these Be-10 measurements scatter well in excess of analytical uncertainty, which has caused extensive speculation about whether the boulders with the highest Be-10 concentrations reflect the actual production rate (because the others are biased low due to surface erosion), or the boulders with the lowest Be-10 concentrations provide a better production rate estimate (because the others are biased high due to inheritance). The summary of all this is that it has not been possible to estimate Be-10 production rates in the high tropical Andes with better than ca. 15% accuracy. As there are lots of researchers who would like to date moraines in this region and relate late-glacial climate changes in the tropics to those elsewhere, this has been a problem.
So a couple of new and potentially better calibration sites in the high tropics could be useful. At the very least, we can expect that they will lead to a rapid release of pent-up exposure-age data from tropical moraines that various research groups have been sitting on for lack of ability to credibly discern which late-glacial climate oscillation they belong to. In this post, I am going to discuss whether the Kelly calibration data agree with calibration measurements elsewhere (i.e., low elevation and high latitude) which gives information on scaling scheme performance. In a subsequent post I will attempt to look at whether these two new tropical calibration data sets agree with each other.
OK. I’ll start by inferring a reference (sea level, high latitude) production rate from the Kelly data given the following input data and assumptions. One, I disregard their samples Q-47 and Q-48. Two, I assume zero erosion for pedestal samples and 4.3e-4 cm/yr erosion for all other samples (this is based on averaging data from their Table 3). Three, I use the MATLAB code in the online exposure age calculators. Four, I use the NCEP reanalysis atmosphere approximation in that code. This procedure is similar to but not exactly the same as the scheme used in their paper. This can be replicated by entering this data block in this page.
Also there is standalone MATLAB code to do this at this address. The file ‘wrap_P10_fit_Quelccaya.m’ does this calibration. You’ll also need all the code from version 2.2 of the online exposure age calculator.
The results are as follows. These are basically the same as the results in their paper. Remember, what we are calibrating here is the reference (SLHL) Be-10 production rate due to spallation only.
|Scaling scheme||Reference Be-10||Percentage||Reduced|
|for spallation||production rate (atoms/g/yr)||uncertainty||chi-squared|
This is interesting by comparison with production rates inferred from mid- and high-latitude, low-elevation sites. The “St” scaling scheme, that is, the non-time-dependent Lal scheme, yields SLHL reference production rates from the Kelly/Quelccaya (“Kellccaya” ?) site that are the same as those inferred from high-latitude sites. The various calibration data sets summarized here give reference production rates between 3.77 and 4.07 that are basically indistinguishable given their uncertainties. The value of 3.85 from this site agrees. Thus, the Lal scaling scheme with the NCEP atmosphere successfully reconciles high-lat, low-elv and low-lat, high-elv calibration data. In contrast, other scaling schemes can’t accomplish this. Here is a figure that makes this clearer, in which I have calculated reference spallogenic Be-10 production rates from all of these calibration data sets using the same code (also see this page):
For the St scaling scheme, the high-latitude calibration data (blue) are indistinguishable given uncertainties, and also indistinguishable from the Kelly low-latitude, high-elevation data (red). For the Du, De, and Li schemes, the high-lat/low-elv data are indistinguishable from each other (which must be the case, because no scaling scheme predicts large scaling differences among these data), but they are not even close to the Kelly data. There are two differences here. First, these schemes are based on neutron monitor measurements that imply a larger elevation dependence of the production rate than do the data that Lal used. Second, these schemes include a time-variable production rate that reflects geomagnetic field variations. This comparison indicates that one or the other of these things are not working. The most likely reason is that the altitude dependence of the neutron-monitor-based scaling schemes is wrong. The reason this is the most likely reason is clearly explained by Nat Lifton in a recently published and important paper that explains the physics of why these data likely overestimate the altitude dependence. The time-dependence is not the most likely problem, because the “Lm” scaling scheme, which has the same elevation dependence as the St scheme with a time-dependence added, performs much better (although not as well as the non-time-dependent scheme) in reconciling the low-lat/high-elv and high-lat/low-elv data. Overall, what I take from the Kelly calibration data and the new Lifton paper is that the neutron-monitor-based scaling schemes, specifically the Du, De, and Li implementations of these schemes in Balco et al., 2008, have an inappropriate altitude dependence. There is no evidence that they perform better than the Lal-based schemes — in fact, all evidence that I am aware of that is based on geological data shows that they perform worse — so therefore I see no good reason to use them for exposure-dating applications.
This is an interesting result because the non-time-dependent Lal scheme specifically disregards a number of basic physical processes, in particular the fact that the Earth’s magnetic field varies over time, that we know should affect cosmogenic-nuclide production rates. However, when applied to geological data, it always performs better than more complex schemes that were designed to include these processes. This was true for the calibration data set of Balco et al., 2008, it’s true for the comparison of new high-altitude and low-altitude data here, and it’s true for the as-yet-unpublished results of the new CRONUS-Earth geological calibration work. In part this is a consequence of the fact that the calibration data that are available are not well suited to testing paleomagnetic-variation schemes, but it continues to be irksome.
This is too long. The next post attempts to look at whether the various high tropical calibration sites agree with each other. Also not discussed yet, but potentially important, are data from an older Quelccaya moraine in the Kelly paper that can’t be reconciled with the main calibration data set from the Huancané IIa moraines.