Causes of Uncertainties in Paleoclimatic Reconstructions Based on the Oxygen Isotope Composition of Glacier Ice on Elbrus (Western Plateau)

A study of the isotope signature of glacial ice in the Western Elbrus Plateau (the Caucasus) was made on the basis of five ice cores obtained in different years with high resolution. It was shown that the isotopic characteristics of ice are associated with the processes of accumulation and wind scouring of snow. Three ice cores were obtained in 2013 (C-1, C-2 and C-3), one in 2017 (C-4) and one more in 2018 (C-5). Core sampling was performed with a resolution of 5 cm. Isotopic analysis was done at the CERL laboratory (AARI) using a Picarro L2130-i isotope analyzer, the accuracy was 0.06 parts per thousand for delta O-18 and 0.30 parts per thousand for 82H. The values of delta O-18 and delta H-2 of the ice of the Western Plateau generally vary from -5 to-30 parts per thousand and from-18.7 to -225.8 parts per thousand, respectively, with well-defined seasonality. Comparison of the isotope record for all cores showed that the differences in accumulation for individual seasons reach 0.3 m w. eq., differences in accumulation for individual seasons averaged over 5 years is approximately 0.2 m w.eq. The absolute differences in the average seasonal values of delta O-18 associated with wind scouring and spatial redistribution of snow (deposition noise), averaged over 5 years, reached 1.38 parts per thousand. The irregularity of precipitation amount within the season and errors in core dating are an additional contribution to non-climate variance (noise of definition). The absolute difference in the average seasonal values of delta O-18 associated with this type of noise averaged over 5 years is 1.7 parts per thousand . Thus, the total uncertainty for two different types of noise can be estimated at 2.2 parts per thousand, which is about 20% of the annual seasonal amplitude of delta O-18 values of the glacier ice in the Western Plateau (the average difference between the delta O-18 values of warm and cold seasons is similar to 10-11 parts per thousand). One of the problems of linking the isotope record to the annual temperature record at the weather station was solved by using ammonium concentrations for dating the C-1 ice core and calculating the "ideal" annual variation of delta O-18 values by a cosine function of the annual amplitude. Using ammonium ion (NH4+) concentration each annual layer in C-1 ice core was divided into two parts associated to snow deposition in winter and in summer. It also showed delta O-18 values associated to change of seasons. The calculation of the cosine function showed the simplified delta O-18 values for each month of a particular year, due to which the delta O-18 values of the season boundaries in the ice core were linked to calendar months. This assimilation allowed us to compare the obtained average seasonal values of delta O-18 from the core with instrumental observations at the Klukhorskiy Pass meteorological station. The delta O-18 values of winter seasons have a weak relationship with surface temperatures, not only due to wind erosion, but also due to the high interannual variability of snow accumulation. At the same time, the average delta O-18 values of the warm seasons are significantly positive correlated with surface temperature (r = 0.7, p = 0.1), so ice core delta O-18 records can be used as a temperature proxy of the warm period.