Abstract
The influence of atmospheric circulation changes reflected in spontaneously occurring sea level pressure (SLP) anomalies upon surface air temperature (SAT) variability and trends is investigated using partial least squares (PLS) regression, a statistical method that seeks to maximally explain covariance between a predictand time series or field and a predictor field. Applying PLS regression in any one of the three variants described in this study (pointwise, PC-wise, and fieldwise), the method yields a dynamical adjustment to the observed NH SAT field that accounts for approximately 50% of the variance in monthly mean, cold season data. It is shown that PLS regression provides a more parsimonious and statistically robust dynamical adjustment than an adjustment method based on the leading principal components of the extratropical SLP field. The usefulness of dynamical adjustment is demonstrated by applying it to the attribution of cold season SAT trends in two reference intervals: 1965–2000 and 1920–2011. The adjustment is shown to reconcile much of the spatial structure and seasonal differences in the observed SAT trends. The dynamically adjusted SAT fields obtained from this analysis provide datasets capable of being analyzed for residual variability and trends associated with thermodynamic and radiative processes.
Supplemental information related to this paper is available at the Journals Online website: http://dx.doi.org/10.1175/JCLI-D-14-00111.1.s1.
Current affiliation: The Climate Corporation, Seattle, Washington.
Current affiliation: Atmospheric and Oceanic Sciences Program, Princeton University/NOAA/Geophysical Fluid Dynamics Laboratory, Princeton, New Jersey.