Data Density Effects on North Antlantic Oscillation Reconstruction: Analysis and Application

A deep understanding of the climate system and its variability is essential for the development of reliable climate predictions. Over the North Atlantic, the North Atlantic Oscillation (NAO), characterized by pressure differences between Iceland and the Azores, is the dominant mode of near-surface atmospheric circulation variability. It explains approximately half of the interannual variability in winter atmospheric pressure in the North Atlantic sector and affects jet streams, storm tracks, and surface climate conditions in surrounding areas. In order to improve understanding of the NAO, its teleconnections and longer-term patterns, multiple means of reconstruction have been employed over time, both model- and proxy-based. Due to the point-based nature of proxy data, the applicability of these reconstructions on a wider spatial scale is difficult to estimate.

We investigate the relationship between spatial data density and reconstruction accuracy through conducting a series of principal component-based NAO reconstructions from temperature and precipitation data. The amount of data available to reconstruct from is thinned through spatial hyperslabbing. Using ERA5 temperature data, the correlation between the reconstructed and observed NAO index for 1990-2020 decreases only little, from 0.80 to 0.79 and 0.78, when thinning the original amount of data (N = 78,899 data points) to 17% and 0.09%, respectively. The variability however is lowered significantly, limiting information on the strength of the NAO. The impact of data density and location is then applied to create a ranking of the utility and estimate biases in existing proxy-based NAO reconstructions and potential future ones. Using this information, we finally create a multi-proxy NAO reconstruction for the past two millennia.