A statistical analysis of Gulf Stream variability from 18+ years of altimetry data

Altimetry data that bracket the core of the Gulf Stream (GS) between 32–39°N and 68–72°W are analyzed from a combined data set spanning the duration of the Topex, Jason-1, and Jason-2 missions. Absolute sea surface height, SSH, referenced to a geoid model and sea surface height anomaly, SSHA, referenced to a global mean sea surface model are calculated using pass 50 data from Topex cycle-002 (04-October-1992) through Jason-2 cycle-098 (02-Mar-2011). Five different indices of GS variability: (i) ΔSSHΔSSH—the change in absolute height across the GS; (ii) ΔSSHAΔSSHA—the change in sea surface height anomaly across the GS; (iii) a barotropic velocity index; (iv) a baroclinic transport index; and (v) the position of the GS, are estimated from the 18+ year time series via seasonal time series, power spectral density functions, auto-covariance functions, and a principal component analysis (PCA). The average annual distribution of all the variables except velocity index increases from a spring/summer minimum to a fall maximum, then decreases during the winter to minimum values in spring and early summer. The average signal in barotropic velocity index reaches a maximum in summer, whereas the maximum baroclinic transport occurs in the fall. The seasonal signal in GS position is better approximated by a seasonal sinusoid than the dynamical variables, with minimum (southerly) positions in spring and maximum (northerly) positions in the fall. Variations in height difference across the current are in phase with changes in the width of the current, leading to smaller seasonal variations in the velocity which are out of phase with the other variables. None of the seasonal time series except GS position resemble a pure sinusoid; they more closely resemble the signal of a nonlinear oscillator.The three dominant periods for ΔSSHΔSSH are 15–21 months, 7 months, and 4 years, with an integral time scale of 32 days. The dominant periods for ΔSSHAΔSSHA are 9 months, 21 months, and 5–6 months, with an integral time scale of 38 days. The spectral and PCA analyses of both ΔSSHΔSSH and ΔSSHAΔSSHA are “pink”, indicating an energetic broad-band spectra. The dominant PCA modes explain less than 30% of the variability, indicative of highly nonlinear processes. The dominant periods for the barotropic velocity index are 9 months, 21 months, and 5 months, with an integral time scale of 14 days. The dominant periods for the baroclinic transport index are 11 months, 2–3 months, and monthly, with an integral time scale of 24 days. The dominant PCA modes of the velocity and transport indices explain 52% and 48% of the variability of their respective time series. The dominant periods for GS position are 9, 11, 13 months, and 1.5–6 years, with an integral time scale of 35 days. The first GS position PCA mode explains 50% of position variability. All of the time series of GS variability and the correlation between the North Atlantic Oscillation and GS position exhibit non-stationary behavior.