A generally accepted basic principle in relation to the use of the noble gas thermometer in groundwater flow systems is that high-frequency noble gas climatic signals are lost due to the effect of dispersion. This loss of signal, combined with 14C dating issues, makes it only suited to identify major climatic events such as the Last Glacial Maximum (LGM). Consequently, the identification of significant noble gas temperature (NGT) cooling (≥ 5 °C) with respect to present time has systematically been associated with the occurrence of the LGM even when reasonable water age controls were unavailable. It has also become apparent at a number of studied sites that modern NGTs estimated through standard models [M. Stute, P. Schlosser, Principles and applications of the noble gas paleothermometer, in: P.K. Swart, K.C. Lohmann, J.A. McKenzie, S. Savin, (Eds), Climate change in continental isotopic records, Geophysical monograph 78, AGU (1993) 89-100.; W. Aeschbach-Hertig, F. Peeters, U. Beyerle, R. Kipfer, Paleotemperature reconstruction from noble gases in ground water taking into account equilibration with entrapped air, Nature 405(6790) (2000) 1040-1044.] are unable to reproduce ground temperatures at the interface with the unsaturated zone, a basic requirement for proper paleoclimate reconstruction through noble gases. Instead, a systematic bias to low NGTs in recharge areas is observed. The Carrizo aquifer, in which the LGM was previously identified [M. Stute, P. Schlosser, J.F. Clark, W.S. Broecker, Paleotemperatures in the Southwestern United States derived from noble gases in ground water, Science 256(5059) (1992) 1000-1001.] and which presents an NGT bias of over 4 °C, is an ideal setting to analyze and revise basic principles and assumptions in relation with the use of the noble gas thermometer. Here, we present a new noble gas data set (49 measurements) collected at 20 different locations in the Carrizo aquifer. This new data set together with previously published data (20 measurements) was used to calibrate a 3-D groundwater flow and 4He transport model in which simulations of groundwater age were subsequently carried out. These account for mixing processes due to advection, dispersion, diffusion, and cross-formational flow. We first show that samples previously attributed to the LGM belong in fact to the middle Holocene. Through a step-by-step approach we then proceed to carry out a comparative analysis of both the impact of dispersion on high frequency climatic signals and assumptions underlying competing NGT models. Our combined analysis indicates that groundwater flow systems, at least those with similar characteristics to that of the Carrizo, do have the ability to preserve short term (100–200 yrs) climatic fluctuations archived by noble gases. It also shows that abrupt climate shifts during the mid-late Holocene which are associated with significant NGT changes (≥ 5 °C) do not reflect equally important changes in the mean annual atmospheric temperature (MAAT). Instead, these reflect the combined effect of atmospheric temperature changes, seasonality of recharge and, above all, significant variations of the water table depth which result from shifts between humid and arid regimes. Together with NGTs, our excess air record plays a critical role in identifying such abrupt climate changes. Specifically, the Carrizo combined data set indicates an abrupt shift from a cool, humid regime to a warmer, arid one at ∼ 1 kyrs BP. A major Holocene (∼ 6 kyrs BP) NGT change of 7.7 °C with respect to present now identified is mostly the result of a dramatic water table drop which occurred during the ∼ 1 kyrs BP transition period. Current NGTs in the Carrizo recharge area do not appear to be recording atmospheric changes. Rather, these are recording ground conditions reflecting mostly the impact of heat flow in the area. We also show that observed systematic offsets in NGT recharge areas can be reconciled through NGT estimation models which account for a noble gas partial pressure increase in the unsaturated zone, potentially due to O 2 depletion.
Read full abstract