Search

An eight-year cave monitoring effort in Gibraltar had the ultimate aim of reconstructing $\delta$18O in past precipitation from well-dated speleothems. We measured stable isotopes in cave waters from nine sites in New St Michael's (NSM) and Ragged Staff (RS) caves, comparing them with monthly GNIP values for Gibraltar rain. Additionally we conducted artificial tracer tests that tracked two dyes from surface to drip sites in NSM.\newline \newlineSeepage waters have a very narrow compositional range and little difference between sites when compared with the fluctuations of $\delta$18O and $\delta$2H in monthly rainfall ($\delta$18O range 12.5 \permil\). $\delta$18O varies seasonally by only tenths per mille whereas inter-annual ranges are up to 1 \permil\. The compressed seasonal variations indicate temporal mixing of rains that entered the sub-surface at different times. Comparison of exceptional variations in rainfall with subsequent seepage samples reveals a component of seepage that has been stored for at least one year. The artificial tracers indicated marked spatial mixing as water flowed from the surface to the caves. Tracers from single points on the surface above MSN formed overlapping plumes of dyed water that could be detected across 200 m horizontally. The tracer results confirm the temporal mixing and indicate residence times that are less than the three month monitoring period but exceed the initial breakthrough times of days to weeks after injection. Combined storage time indicated by stable isotopes, tracer results imply that the full residence time distribution for water in the unsaturated zone ranges from a few days to over one year. This temporally and spatially dispersive behaviour arises from the geometric properties of the fracture network that the cave seepages have passed through, so will have been broadly invariant in the past. The water isotope signal that Gibraltar speleothems capture is averaged over a period of around one year and very short events will not be recorded. \newline\newlineThe stable isotope data also show the influence of other processes than simple mixing. Seepage waters have lower values than the amount-weighted averages for rainfall, suggesting a bias towards isotopically lighter rains. Plots of $\delta$2H versus $\delta$18O show no consistent signs of modification by evaporation, despite Gibraltar's hot dry summers, so the isotopically heavier rains are completely lost to evapo-transpiration. A less easily explained feature is that the deuterium excess of seepage is higher than that of average rainfall, although lying within its overall range. This may be caused by addition of condensation water to the shallow epikarst. \newline\newlineIn the cooler climates of the Last Glacial evaporation would have been less than today and the isotopic composition of seepage may have been closer to the annual average of rainfall. Samples from speleothems often span several years even in highly-resolved records, with the implication that in cool climates the calcite $\delta$18O will reflect the average for rainfall over this time interval. In climates resembling the present it may be displaced slightly towards lighter values.\newline [ABSTRACT FROM AUTHOR]

The δ18O of past rainfall is important in palaeo-climatology because it can be simulated directly by isotope-enabled General Circulation Models, enabling data-model comparisons. We describe a methodology for reconstructing it from Last Glacial speleothems in Gibraltar. Three factors determine the δ18O of dripstone calcite: the δ18O of drip water, the cave temperature (that determines equilibrium fractionation between water and calcite), plus non-equilibrium fractionation caused by CO2 degassing and/or surface reactions. Reconstructing seepage water δ18O requires independent constraints on the second and third factors. It is also necessary that the relationship between seepage δ18O and rainfall be well-constrained. A merged record from flowstone and stalagmite (Gib10d/10e) from the same chamber of Ragged Staff Cave spans 30 – 110 ka. δ18O variations between 40 and 62 ka align closely with NGRIP's GICC05modelext timescale (1) but there is divergence before and after this. We therefore restrict our reconstruction to 40-62 ka.To constrain temperature we used Uk'37 estimates of annual average sea surface temperatures (SSTs) in core ODP-977A in the Alboran Sea, after converting its original timescale to that of NGRIP (2). We assume that in coastal Gibraltar Mean Annual Air Temperatures (MATs) equalled SSTs. Cave temperatures probably lagged behind MATs due to the thermal capacity of the bedrock, but with the possibility of a much smaller lag because of heat advection from outside by cave winds. To deal with the uncertainty this imposes, we use a 1-D heat flow model to set up two bounding scenarios for the evolution of cave temperatures. In one the full effects of heat storage and conduction are computed for 100 m depth whereas in the other the lag is zero. Both scenarios are used to determine fractionation factors (3) and time-series of 'equilibrium' values for δ18O in cave water. The effects of non-equilibrium are difficult to assess. The δ18O records for the two speleothems overlay almost exactly despite their different morphologies and implied drip discharges. We take a similar approach to 'disequilibrium' as for cave temperature, defining a bounding scenario of 'no disequilibrium' but alternatively using the 'Tremaine curve' that takes disequilibrium into account (4).This approach gives rise to a 2 x 2 set of scenarios that define an 'envelope' for δ18O in palaeo-seepage. Cave monitoring shows that average seepage δ18O is currently 0.6 to 0.75 ‰ less than rainfall, because of complete evaporation of summer rains. This bias would have been less in cooler climates, so we adjust the reconstructed envelope by amounts related to prevailing SST, to obtain estimates and uncertainties for the average δ18O of palaeo-precipitation in the western Mediterranean that can be compared directly with GCM output and with other records on the NGRIP timescale. [ABSTRACT FROM AUTHOR]

Ten years of comprehensive cave monitoring in Gibraltar provides a unique basis forinterpreting proxy records in speleothems, and in this study we discuss some of theopportunities and challenges in constructing an interpretive framework for long proxyrecords. We present a new trace element and stable isotope record spanning 30-110 kaobtained from two neighbouring speleothems from Ragged Staff Caves, Gibraltar. The agemodel is defined by over 100 U-Th dates with continuous micromilling for stable isotopes,and LA-ICPMS used for acquiring the trace element proxy record. These provide highlyresolved critical evidence for some of the main meteorological and environmental controls onpast western Mediterranean terrestrial climate at near decadal and sub-annual resolution,respectively. Our interpretations of the monitoring data highlight geochemical and hydrologicalprocesses critical to speleothem deposition. We conclude that δ13C and δ18O in speleothemcalcite are controlled by within-cave conditions that vary seasonally and are themselvescontrolled by the temperature and water balance at the ground surface. For interpretation oftrace elements, we utilize multi-annual monitoring of drip waters. The Gibraltar record showsvariations in δ13C and δ18O on timescales from multi-decadal to multi-millennial which wecompare with independently dated records of multiple proxies from marine cores,including vegetation and sea surface temperature (SST), and the Greenland ice corerecord. The precision of the Gibraltar speleothem chronology allows for greater confidence incorrelation with other records, and we particularly note a remarkable correspondence withδ18O in Greenland in the period 40-60ka, and similarity with SST records from the AlboranSea. We believe this reveals a dominant influence of North Atlantic climate at the gateway tothe Mediterranean. However, we note that in periods which there is weaker correspondencebetween speleothem and ice core δ18O, this may indicate changes in the dominant climatecontrols in the western Mediterranean. All comparisons of proxy records show a broad similarity with the cave recordover the longest timescales, i.e. δ13C and δ18O both display higher values in thecolder, drier climates of the mid-glacial, and generally lower values in the longinterstadials of the early glacial period. Superimposed on this trend are millennial-scalevariations with amplitudes of the same order of magnitude seen over stadial-interstadialtimescales. At times these fluctuations correspond closely to the variations seen in SST,pollen, and vegetation records while some excursions appear out of phase. We usemultiproxy marine core records to explore these phase differences, as the speleothemrecord may partially reflect other factors independent of those recorded throughSST and pollen proxies. Over centuries high frequency variations similar to thoseobserved on multi-annual timescales in cave monitoring and modern speleothemoccur, showing both the small and large scale applications of our monitoring data. [ABSTRACT FROM AUTHOR]

Gibraltar caves, in common with many other systems, undergo seasonally reversing dynamic ventilation. Variation in cave air pCO2 closely controls CO2 degassing of drip waters and speleothem fabrics. Proxy capture and isotope fractionation are controlled by the mixing of advective CO2-rich ground air with background atmosphere. Extrapolating the significance of the ventilation-advective system to times past is hampered by artificial entrances which have altered airflow patterns in natural caves. Although modelling of advective airflow shows that tunnels amplify ventilation in Gibraltar caves, 'modern-style' paired laminations are observed in older speleothem, provide compelling evidence for a ventilation-advective system permeating Gibraltar karst in the past.This study focuses on the meteorological controls acting on cave air pCO2 in the modern dynamic system and identifying which of these are relevant to an interpretive framework that can be applied over longer timescales. We compare data for cave air CO2 measured simultaneously at 2h intervals at eleven locations in New St Michaels Cave (NSM, 240-270 m altitude) and Ragged Staff Cave, (RS, 0 - 70 m altitude). These detailed records reveal lateral and vertical spatial patterns of cave air pCO2 over six seasonal cycles. The most striking properties of the CO2 records are 1) annual seasonality in cave air pCO2 showing an antiphase relationship between caves in NSM and RS, and 2) that the high pCO2 seasons (November to April in NSM and April to November in RS) are disrupted by frequent and large fluctuations in pCO2 which occur on hour-day timescales.With respect to palaeoclimate reconstructions, the first order controls on pCO2 are 1) the processes that regenerate and maintain the ground air reservoir (soil organic matter production rates and downwash by rainwater recharge), and 2) the range in seasonal mean daily temperatures which creates buoyancy-driven ground air advection throughout the unsaturated zone. Rapid fluctuations in cave air pCO2 during the high pCO2 (incoming ground air) season are second-order effects which may also accumulatively control the net extent of seasonal degassing associated with calcite deposition. These are caused by changes in the ground air-atmosphere mixing ratio linked to synoptic-scale patterns of 1) diurnal temperature range and 2) ambient wind conditions. Rapid changes in the balance between ventilation and ground air advection occur when the when rock core temperature is bracketed by the diurnal range on cool summer days or warm winter days. These conditions are often associated with Atlantic frontal systems. Longer periods of persistently weaker seasonality would also impact on the net seasonal pCO2 and decadal patterns in the speleothem δ13C record. A secondary but important systematic control by wind direction on cave air pCO2 is also observed, but this is often overridden by fluctuations in daily temperature. Should seasonal temperatures become more stable, the δ13C record may include a component of variation related to the local wind-field. Whilst monitoring refines palaeoclimate reconstructions this deeper knowledge also shows that interpretive frameworks may need to be adapted for application to different palaeoclimatic regimes. [ABSTRACT FROM AUTHOR]