Identifying coherent spatiotemporal modes in time-uncertain proxy paleoclimate records

Abstract

High-resolution sedimentary paleoclimate proxy records offer the potential to expand the detection and analysis of decadal- to centennial-scale climate variability during recent millennia, particularly within regions where traditional high-resolution proxies may be short, sparse, or absent. However, time uncertainty in these records potentially limits a straightforward objective identification of broad-scale patterns of climate variability. Here, we describe a procedure for identifying common patterns of spatiotemporal variability from time uncertain sedimentary records. This approach, which we term Monte Carlo Empirical Orthogonal Function (MCEOF) analysis, uses iterative age modeling and eigendecomposition of proxy time series to isolate common regional patterns and estimate uncertainties. As a test case, we apply this procedure to a diverse set of time-uncertain lacustrine proxy records from East Africa. We also perform a pseudoproxy experiment using climate model output to examine the ability of the method to extract shared anomalies given known signals. We discuss the advantages and disadvantages of our approach, including possible extensions of the technique.