Source Apportionment Mixing Models for High Resolution Trace Element Ice Core Records over the Past 350 years at Siple Dome, Antarctica

High Resolution trace element concentration records from polar ice cores are valuable in identifying paleoclimate trends. However, traditional methods of allocating the elemental concentrations between sources often assume that the analytical measurement errors and the elemental ratio uncertainties of each source are negligible. To address the impacts of each of these uncertainties, a Monte Carlo analysis was used to quantify the sensitivity of mixing models to uncertainties in the input variables. Simulations of 100,000 trails were based on an idealized concentration dataset of average concentrations at Siple Dome, Antarctica, with uncertainty applied to the analytical measurements and/or the source elemental ratios. The results indicated that for both two component and three component mixing models, uncertainty in source allocations decreased with increasing number of elements measured; most notably with the inclusion of unique endmember elements for each source (e.g., sodium and manganese are unique endmembers for seasalt and crustal dust sources, respectively). Further, the Monte Carlo analysis allowed for quantification of error bars on annual average source concentration records over time. Using the two component mixing model, we found that Siple Dome is a marine dominated area (89% seasalt, 11% crustal dust by mass) with seasalts varying approximately 65%, and crustal dust concentrations approximately doubling over the last 350 years. Using the three component mixing model, separating sources that are either not well characterized or do not have unique elemental ratios proved to be problematic.