Assessing Stationarity in Ice Core Record-Sea Level Pressure Relationships for Yukon Territory Ice Core Records

We assess the stationarity of the relationship between Northern Hemisphere winter (Dec-Feb) sea-level pressure (SLP) and proxy time series (major ions, accumulation, and stable isotopes) from the Eclipse (3017 m asl) and the Mt. Logan Prospector-Russell Col (PRCol; 5340 m asl) ice cores from Yukon, Canada. We develop a novel spatial calibration procedure to identify ranges of ice core values that are associated with consistent winter SLP anomaly patterns. Each ice core variable time series was ranked and divided into groups of 13 years each. We assess stationarity by splitting the 1872-2001 analysis period in half (1872-1936 and 1937-2001) and comparing the locations and magnitudes of SLP anomaly patterns during the two periods for each group of ice core values. Northern Hemisphere monthly mean SLP from the 20th Century Reanalysis dataset are used. The high accumulation rate (1.38 m a-1) at Eclipse allows us to analyze 6-month seasonal mean values (Oct-Mar and Apr-Sep), whereas annual mean values are used from PRCol where the accumulation rate is lower (0.40 m a-1). The Eclipse cold season accumulation and PRCol annual mean sodium concentrations (Na+) exhibit the strongest correlations with winter SLP anomaly patterns. In particular, the lowest and highest 20% annual Na+ values at PRCol and lowest 10% cold season accumulation values at Eclipse exhibit stationarity with consistent SLP anomaly patterns in the North Pacific for all three time periods. A weaker Aleutian Low consistently occurred in the central to eastern North Pacific for the lowest Na+ years at PRCol and lowest accumulation cold seasons at Eclipse, although these groups of years are mostly independent. A stronger Aleutian Low occurs in the North Pacific for the highest Na+ years at PRCol. A stationary SLP anomaly pattern is not observed through all three time periods for high cold season accumulation at Eclipse. Application of this calibration procedure with other traditional calibration and reconstruction methods can help identify the proxy values that are most likely to produce an accurate climate reconstruction. This study highlights an application where incorporating more SLP observations into the pre-1930 period of the 20th Century Reanalysis can improve the fidelity of climate proxy calibrations and paleoclimate reconstructions.