Nevado Huascarán - Oxygen Isotope, NO3, Layer Thickness, and Particle Data ----------------------------------------------------------------------- World Data Service for Paleoclimatology ----------------------------------------------------------------------- NOTE: PLEASE CITE ORIGINAL REFERENCE WHEN USING THIS DATA!!!!! DESCRIPTION: General Information about the Huascarán Ice Cores All metadata and data for this study can be accessed via: https://doi.org/10.25921/swwy-sg59 ----------------------------------------------------------------------- Site Description and Analysis In July-August 1993, two ice cores to bedrock were recovered from the col between the north and south peaks of Nevado Huascarán, Peru (9øS, 77ø30'W, col elevation 6050 m) and were subsequently transported back to the cold room facility at the Byrd Polar Research Center (BPRC). Core 1 (HSC1, 160.40 m) was sectioned in the field into 2677 samples decreasing in thickness from 13 cm at the top to 3 cm at the base, which were then melted and poured into 2 or 4 oz. plastic (HDPE) bottles, and sealed with wax. Core 2 (HSC2, 166.08 m), drilled approximately 100 m from the HSC1 site, was returned frozen in 1 m sections. Ice motion vectors determined from stake movements from 1991-93 indicate that the drill sites are proximal to the divide between ice flow towards the east and west outlets of the col. Visible observations and borehole temperatures indicate that the glacier is 'polar' type, i.e., it remains frozen to the bed (Thompson et al., Science, v.269, 1995, p. 46-50). Each ice sample from HSC2 was prepared in a Class 100 clean room environment, and analyzed for major anion concentrations (Cl-, NO3-, and SO42-) on a Dionex 2010i ion chromatograph, d18O on a Finnigan Mat mass spectrometer (Craig, 1957), and for particulate concentration and size distribution using a Coulter TA-II particle counter (Thompson, OSU IPS Report 46, 1973). A complete d18O profile was also produced from the bottled samples from HSC1. Contamination during field preparation and transport of these samples precluded the development of a second complete record of particles and anion concentrations. For display purposes, variable averaging on the core depth scale was utilized to show the major large-scale events in the record without the confusion of the large annual variations superimposed upon the upper portion. Hence, for HSC2, 5-m integrated averages were calculated for between the surface and 140 meters depth and then 50-cm averages were generated between 140 and 160 meters. Between 160 and 166 meters, every sample value was plotted. A similar scheme was used for HSC1 (all values plotted for 155-160.4 m). These data are included in hs1-5m.txt and hs2-5m.txt in this data archive, and the graph can be seen in Thompson et al., 1995 (Fig. 3). Development of the time/depth relationship Tropical South American climate is marked by annual dry seasons (July-October) which were identifiable in the ice core record as elevated values in all relevant measurements. The nitrate (NO3-) record from the Huascarán ice core provided the most definitive seasonal marker, but the final time scale was constructed from a comparison of four major parameters (NO3-, d18O, dust and SO42-). Each annual maximum corresponds to the middle of the dry season, assumed to occur on the 1st of August. The rapid layer- thinning below 120 m limited annual resolution to the most recent 270 years. However, the high accumulation and strong preservation of seasonal cycles also made possible the subannual resolution of d18O variations for a period of at least 100 years (1894-1993). The accuracy of the time scale is of paramount importance in the development of relationships between ice core proxy data and tropical climate conditions. Several horizons in recent times were useful for confirming the layer counting as a reliable method, and indicate almost certain ages for the uppermost 50 years. In 1980, during the original reconnaissance expedition to Huascarán, a 10 m firn core was extracted and analyzed for d18O at BPRC (Thompson et al., JGR, v. 89d3, 1984, p. 4638-4646). Aside from minor accumulation variation and slight signal attenuation, the 1993 cores duplicated the earlier stable isotope profile over the common portion, and confirmed the layer counting to 1980 as absolute. Additionally, a magnitude 7.7 earthquake struck coastal Peru in May 1970, generating large mud flows following the collapse of a large portion of the Huascarán glacier from the north peak. The event was recognized in the ice core by a sharp two-year rise in particulates from the newly-created sediment source. A third time horizon was provided by the HSC2 36Cl profile (Synal et al., Glaciers From the Alps, Paul Scherrer Inst., 1997, p. 99-102), a substance produced by neutron activation during the explosion of atomic devices in the presence of a 35Cl source, such as sea water. An abrupt >100-fold rise in 36Cl concentration occurred at ~54 m depth, which dates (by layer counting) to 1951-53. This was in direct response to the October 31, 1952 U.S. 'Ivy' surface test of an experimental nuclear device on the Eniwetok Atoll in the Pacific Ocean (11øN, 162øE) (Carter and Moghissi, Health Physics, v. 33, 1977, p. 55-71). Finally, in both HSC1 and HSC2, the 1883 eruption of Krakatau, Indonesia (6øS, 105ø30'E) was identified by an anomalous sulfate concentration of ~400 ppb at 110 m depth, more than twice the level of any other local (within 10 m) event. A date of mid- year 1884 was thus considered to be an absolute time marker for both cores within the error of the time lag (less than one year). Time-scale determination, data averaging and time-linearization techniques Years have been counted in HSC2 back to 1720 via the multi-parameter technique. HSC1 was dated using transferred dry season horizons from HSC2, by matching subannual features in the d18O profiles. For each year, the mid-dry season (assumed on average to represent the 1st of August) has been identified as the depth corresponding to the first established 'high' point in each of the major parameters, according to a hierarchy of highest to lowest annual signal strength given by the following: NO3- > d18O ò Particle conc. > SO42- > Visible layers ò Cl- This somewhat subjective technique takes into consideration offsets in dry season peaks between various parameters by sometimes visually interpolating between maxima, but always weighing the nitrate record strongest in all but the most anomalous situations. Secondary, late-season peaks in individual records (most commonly d18O) are ignored in the precise determination of the layer boundaries, as they represent times other than the normal maximum dry period (July-September). Boundary depths are recorded assuming a resolution of 1 cm down to 115.5 m depth (1860), with 5 mm resolution below (to 1720). Thermal-year (Aug.-Jul.) layer thicknesses (hs1layer.txt and hs2layer.txt) determined from the difference between consecutive boundary depths, can then be said to have a resolution of ~2 cm and ~1 cm, respectively. However, the actual temporal accuracy is unknown, and subject to interannual variation in the timing of the dry season maxima, however that may be realized in nature. Following the final determination of the annually-resolved time scale for each of the two cores, annual averages of each major ice core parameter were calculated. For d18O, anions, and particulates, these averages were simple arithmetic mean values of the subset of individual sample measurements containing all samples with a top depth lying between the respective top and bottom boundary depths for that thermal year (annualt.txt). However, annual averages were also calculated in terms of calendar years (annualc.txt), for which the boundary depths are midpoints between the adjacent mid-dry season boundaries. Although these midpoints may approximate the Dec. 31 - Jan. 1 horizon, it is expected that these will have a far greater temporal uncertainty than the more easily recognizable dry season-to-dry season (thermal) years. However, because several of these parameters are 'additive' properties, and typically show spike-like maxima during the dry season, this approach might be better suited for some. In other words, calendar year averages will not necessitate the division of large dry season peaks into two halves (for two separate annual averages), and the uncertainty in the calendar year boundaries will be counter-balanced by the fact that values in this part of the year are low and show far less variability. For the uppermost two century averages (as given in hs2-100a.txt), the values are calculated by using the annual averages already determined from the layer counting, such that each year is weighted evenly as opposed to being differentially weighted according to varying layer thickness and sample density, which would occur through averaging all samples over each 100-year long depth interval. The raw layer thickness data was translated into ice equivalent (IED) thickness using a modelled density function. For HSC2, individual measurements of bulk density were determined by weighing each section (roughly 1 m) of ice core on an analytical balance, and dividing by a volume equal to an idealized cylinder of material according to the length of the section (determined to ~0.5 cm) and its diameter (determined to ~0.1 cm). A model polynomial expression of density as a function of core depth was generated in order to eliminate the effects of random or minor systematic errors in the density profiles. A maximum allowable density value was set at p = 0.92, and the firn/ice transition was defined as the point at which the density reaches 0.830, which is 22.5 m for HSC2. For each thermal year, a single density value (p ave) was created by averaging the density values for the boundary depths on the top and bottom of that annual layer. The ice equivalent layer thickness (l) is then calculated using the formula: l (ice eq.) = l (measured) * (p ave/p ice), where p ice = 0.92 g cm-3 By converting to ice equivalent depth, the total ice thickness at the HSC2 site becomes 158.85 m. The data files hs1layer.txt and hs2layer.txt include IED layer thicknesses from the surface (1993) to 1720 (HSC2) and 1835 (HSC1). The high accumulation on Huascarán provided a sample density (defined as the number of core segments cut per year of accumulation) sufficient to retain monthly resolution for ~100 years. To facilitate comparisons between ice core parameters and climatological datasets, the Huascarán records were transferred to a time scale with a constant interval. This 'time-linearizing' technique involved dividing each 'thermal' year (August 1 to subsequent July 31) into 12 separate bins of equal thickness (or depth), to yield a continuous 'quasi- monthly' time series. Of course, the complex interaction of varying accumulation rates within a year, sampling intervals, averaging techniques, and indefinite dry season time horizons are all potential errors when compared to real time. Dividing the annual oxygen isotope record into 12 monthly increments was justified because the annual oxygen isotopic signal preserved in the ice core is remarkably evenly proportioned between enriched (dry season) and depleted (wet season) isotopic values. The reasons for this are at least two fold: firstly, during the dry season it seldom rains at lower elevations; however, often clouds will form on the mountain tops and snow will fall on these high elevation snowfields. Secondly, it has been known for years from work in the polar regions (e.g., Hammer et al., J. Glaciol., v. 20, 1978, p. 3-26), that the isotopic input signal from individual snow events is much noisier than the record preserved in the snow pack. The mass exchange by diffusion via the vapor phase in the porous snow usually obliterates the high-frequency variability in d18O within a few years, depending on the temperature and the thickness of the individual layers. Furthermore, sublimation enrichment (Grootes et al., JGR, v.94d1, 1989, p. 1187-1194) of near-surface late-winter snowfall may also contribute to an increased representation by 18O- enriched snowfall. Thus, the dry season d18O signal is distributed over a greater percentage of each annual increment than might be expected from the annual precipitation patterns observed at the lower altitude meteorological stations. Within each year, each measured d18O value was funneled into the monthly bin corresponding to the sample depth. A large majority of years did not fortuitously have exactly 12 measurements, so adjustments were required. Those years with fewer samples required the insertion of simple averages of the two measurements on either side of each previously empty bin. In contrast, years with sample density (x) greater than 12 normally produced (x - 12) values that were averages of two measurements, up to x = 24. Only a small number of years had more than 24 samples and required averages of three or more values. For HSC2, a total of 1633 samples covered the 1201 months that were time-linearized. In summary, 702 (58.4%) of the 1201 months represented an unmodified value transferred to the time scale, 403 (33.6%) represented an average of two values within a bin, 39 (3.2%) represented three values, 2 (0.2%) represented four values, and the remaining 55 (4.6%) were interpolated values. The identical procedure was performed for the bottled core, HSC1, and confirmed that this procedure truly captured the temporal fluctuations in the Huascarán d18O record. For the 1198 months (the top three months of accumulation were lost in the drilling) covering the top 100 years of HSC1, only 1195 samples were cut; 887 months (74.0%) were single measurements, 151 (12.6%) were averages of two samples, 2 (0.1%) were averages of three, and 158 (13.2%) were interpolated values. Although HSC1 was sampled more coarsely, the major subannual and interannual variations were duplicated (see PAGES, Quat. Sci. Rev., v. 19, 2000, p.22), yielding particularly good correlations over the last 40 years. Because HSC1 lacks true monthly resolution before 1940, HSC2 should be considered to be the superior temporally-resolved record. The quasi-monthly values for the three major parameters shown in PAGES Fig. 12 are given in all-mo.txt, along with the monthly d18O values for HSC1. The lower portion of the Huascarán cores were dated by cross-correlating the ice- oxygen isotope record to the inverted CaCO3-oxygen isotope record from marine core SU81-18, off Portugal (Bard et al., Nature, v. 328, 1987, p. 791). The relationship is shown on Fig. 5 (Thompson et al., 1995). Originally, a two-parameter age-depth model (Thompson et al., Ann. Glaciol., v. 14, 1990, p. 288-297) was used to date the early Holocene portion of HSC2, using the well-known Younger Dryas (YD) period as a target date, strongly recorded in most terrestrial and marine paleoclimate indicators in and around the North Atlantic basin at ~11.7-13.0 kyr. BP, or at ~10-11 kyr. BP 14C. Both HS ice cores record the event as a partial reversal of isotopic values to glacial conditions during the transition. The lower resolution of the HS cores and perhaps the large distance between the Peruvian Andes and the far northern Atlantic region (where the event is strongest) results in a signal that differs not only in magnitude but also in shape, as it has much less of a square-wave appearance (as in GISP2) but is rather a more gradual 'v-shape' oscillation. Therefore, it would appear more realistic to assign a center age of 12.3-12.4 kyr. BP to the depth corresponding to the lowest value, as opposed to assigning beginning and ending dates for the Southern Hemisphere response of this event. In HSC1, this depth is 158.11 m, and in HSC2, it occurs at 164.07 m. The resultant 100-year averages of three climate indicators from HSC2 (using the two-parameter model for age control) were shown in Fig. 7 (Thompson et al., 1995) for the entire Holocene period. Subsequently, a similar age modelling of the HSC1 record gave different results, showing an offset of about 2000 years in the mid-Holocene. Sowers et al. (JGR, v. 94, 1989, p. 5137) have demonstrated that air bubbles trapped in ice caps during the firnification process record changes in the isotopic composition of the O2 component of the global atmosphere (d18Oatm). Whereas it may be argued that seemingly simultaneous isotopic fluctuations in archived precipitation records in various places around the world could be subject to a certain amount of response time lag, the isotopic composition of air has been shown to be well mixed to the point of being globally homogenous on these time scales. Therefore, periods such as the Glacial/Interglacial transition (~7.0-15.0 kyr BP), which is characterized by a large change in d18Oatm, can be easily matched between records, independent of their respective locations or the distance separating them. Measurements of d18Oatm on HSC2 done by Sowers and Bender (unpublished results) since 1995 yielded several useful time horizons within this time period. These ages have led to a revised Holocene chronology for HSC2 that closely matches the two-parameter model results for HSC1. The layer-counted portion of the record (to AD 1720) was utilized as before. The analogous 100-year averages for the three main climate indicators from HSC2 were published in the 2000 PAGES volume of QSR (Fig. 11, p. 27). The data reside in the hs2-100a.txt file.