EnsoFbTauxSsh
Taux-SSH_feedback: coupling between Taux anomalies in the western equatorial Pacific and SSH anomalies in the eastern equatorial Pacific
Computes sea surface height anomalies (SSHA; used as a proxy for subsurface temperature) in the eastern equatorial Pacific (horizontal Niño3 average) regressed onto zonal wind stress anomalies (TauxA) in the western equatorial Pacific (horizontal Niño4 average).
TropFlux and AVISO 1993-2018 (main)
SSH: JPL-MEASURES 1993-2021, CSIRO-SSH 1993-2019, GODAS 1980-2023, ORAS5 1958-2022, SODA3.4.2 1979-2018
Taux: ERA5 1940-2022, 20CRv3 1836-2015, NCEP2 1979-2023
Niño3, Niño4
None
- seasonal cycle removed
- detrending (if applicable)
- spatial average
- seasonal cycle removed
- detrending (if applicable)
- spatial average
- SSHA regressed onto TauxA (slope)
- abs((model-ref)/ref)*100
monthly
% of error
- sea surface height (SSH)
- zonal wind stress (Taux)
The first level shows the diagnostic used to compute the metric and highlight the difference between the model and the reference.
Figure 1: scatterplot of zonal wind stress anomalies (TauxA) in the western equatorial Pacific (Niño4 averaged) and sea surface height anomalies (SSHA) in the eastern equatorial Pacific (Niño3 averaged), showing the strength of the Taux-to-SSH coupling (usually too weak). The black and blue markers show respectively the reference and the model. The metric is based on the slope of the regression and is the absolute value of the relative difference: abs((model-ref)/ref)*100.
The second level tests the hypothesis of a nonlinear relationship between SSTA<0 and SSTA>0.
Figure 2: scatterplot of zonal wind stress anomalies (TauxA) in the western equatorial Pacific (Niño4 averaged) and sea surface height anomalies (SSHA) in the eastern equatorial Pacific (Niño3 averaged), showing the possible nonlinearity in the strength of the Taux-to-SSH coupling (usually shows weak nonlinearity, with a gentler slope for TauxA<0 and steeper slope for TauxA>0 in both reference and model). The black, red and blue lines and numbers show respectively linear regression computed for all TauxA, TauxA>0 and TauxA<0, the left and right scatterplots show respectively the reference and the model.
The third level shows the remote coupling in the equatorial Pacific.
Figure 3: spatial structure sea surface height anomalies (SSHA) in the equatorial Pacific (meridional 5°S-5°N average; zonal 30° running average) regessed onto zonal wind stress anomalies (TauxA) in the western equatorial Pacific (Niño4 averaged), showing the possible nonlinearity in the strength of the Taux-to-SSH coupling (the reference shows the maximum coupling around 140°W, same position or slightly westward for TauxA<0, around 120°W for TauxA>0, with a stronger amplitude for TauxA>0; usually the models simulate a very small displacement of the maximum coupling for all TauxA, TauxA<0, TauxA>0 and simulate similar amplitude in each case). The black, red and blue lines and numbers show respectively linear regression computed for all TauxA, TauxA>0 and TauxA<0, the dashed and solid curves show respectively the reference and the model.
The fourth level shows the spatio-mean annual structure of the coupling.
Figure 4: spatio-mean annual structure of sea surface height anomalies (SSHA) in the equatorial Pacific (meridional 5°S-5°N average; zonal 30° running average) regessed onto zonal wind stress anomalies (TauxA) in the western equatorial Pacific (Niño4 averaged), showing the possible nonlinearity in the strength of the Taux-to-SSH coupling. For all TauxA, the reference shows two peaks in the coupling: spring and late-autumn, both located around 140°W. Usually the models do not simulate the spring peak and the autumn peak is too weak. For TauxA>0, the reference shows three peaks: spring, summer late-autumn, all mainly in the eastern Pacific but the spring peak is also present in the central Pacific. Usually the models do not simulate the spring peak and the summer-late-autumn is only one and too weak (here the three peaks are simulated but all are too weak, particularly during spring). For TauxA<0, the reference shows two peaks: spring and late-autumn, both in the central Pacific. Usually the models do not simulate the spring peak (or it is very weak) and the late-autumn peak is weak (here the spring peak has the right amplitude but happens too months later, the late-autumn peak is too strong, last too long and happens one or two months earlier). The first, second and third rows show respectively linear regression computed for all TauxA, TauxA>0 and TauxA<0, the left and right Hovmöllers show respectively the reference and the model.