ENSO
Thomas Toniazzo has had a more in-depth look at the tropical Pacific variability in FAMOUS than is presented in the GMD paper. His basic conclusion is that “the variability is not primarily caused by coupled ENSO dynamics”. Regressions of SST anomalies onto the Nino3/3.4 indices look much more like the surface Wind-Evaporation-SST -type modes as described in Vimont and Alexander (J.Clim 22, 2009) and Toniazzo (submitted to Clim Dyn). He further suggests that “perhaps in this model the surface coupling is too strong, with SST anomalies driven by surface fluxes with little contribution from the ocean”.
However, filtering the Pacific SST timeseries through a 2–6 (or 3–5) year bandpass, as I did for some of the analysis in the GMD paper, seems to show up some more-ENSO like events, although Thomas warns that “warm SSTS spread much too much westwards, escpecially late into the events. Events lock to spring (May), and the generation of equatorial anomalies still appears to rely on the meridionally-proagating surface mode, mainly from the southern hemisphere.”
Here are some of his plots - I’ve paraphrased his relevant comments below each one. All but the last figure used the unfiltered data; the last comes from 2–6yr bandpassed SST data. Thanks to Thomas for all this - any errors/misconceptions on this page are attributable to my reproduction, not his analysis!
Attach:fam-f1_timeseries.gif Δ
There’s no clear oscillatory behaviour, with “noise” at short timescales. Seasonal locking is irregular.
Attach:fam-f2_autocorr.gif Δ
There’s a short decorrelation time and no oscillation to a negative phase. A long positive
tail is also characteristic of tropical surface-evaporation modes.
Attach:fam-f4_seascyc.gif Δ
Northern hemisphere and southern hemisphere winters are the preferred seasons for variability (upper panel). The decay during nh winter suggests that it’s not an ENSO mode. The seasonal cycle shows too late warming in spring, and a sharp, but quickly faltering cooling in autumn (lower panel) - there’s probably a common cause for the errors in the variablity and the overall seasonal cycle.
Attach:fam-f6_leadcorr-sft-n34_JulAug.gif Δ
Lags (positive for leads) in months indicated on each panel. There are meridionally-propagating WES-type modes the winter hemisphere before the ENSO-like events. Equatorial anomalies at lead 12 months grow more slowly than the winter hemisphere anomalies. Only near the peak of the event is there any equatorial intensification, but it’s short-lived as it moves quickly westward and is dissipated in the warm pool.
Attach:fam_leadcorr-sft-n34_2-6yr-bandpass_allmonths.gif Δ
These regression plots use the 2–6year bandpass filtered data. The equatorial events look a bit more ENSO-like, with a better pattern and evolution. The anomalies are mainly stationary over the equator, and an oscillation becomes visible.
