Oscillatory Spatiotemporal Signal Detection in Climate Studies: A Multiple-Taper Spectral Domain Approach

Publisher Summary This chapter introduces a methodology for signal detection and reconstruction of irregular spatiotemporal oscillatory signals—the multiple-taper spectrum estimation method (MTM)–singular-value decomposition (SVD) methodology. This methodology is offered as an alternative technique which avoids most of the problems encountered in traditional techniques and provides an efficient exploratory method for climate signal detection. The associated signal-detection parameter—the local fractional variance spectrum (LFV) spectrum—yields the correct null distribution for a very general class of spatiotemporal climate noise processes and the correct inferences when signals are present. The methodology allows for a faithful reconstruction of the arbitrary spatiotemporal patterns of narrowband signals immersed in spatially correlated noise. The results of the MTM–SVD approach are robust to the temporal and spatial sampling inhomogeneities that are common in actual climate data. Applied to observational climate data, the MTM–SVD analysis yields insight into secular trends, low-frequency, and high-frequency quasi-oscillatory variations in the climate system. The dominant mode of secular variation has been a long-term global warming trend associated with some anomalous atmospheric circulation patterns that show similarity to the modeled response of the climate to increased greenhouse gases.