CLIMATE
VARIABILITY
RESEARCH
Towards Nonlinear Identification of the Atmospheric Response to ENSO
The nature of the atmospheric climate signal associated with El Niño Southern Oscillation (ENSO) is an ongoing debate within the climate community, particularly the nature of the remote signal. This work constitutes a first step towards estimating the nonlinear atmospheric response to ENSO using state-of-the-art General Circulation Models (GCMs). Of particular interest is the nature of the nonlinear response over the North Pacific and North Atlantic sectors. A set of multi-decadal integrations of the Hadley centre GCM model, HADAM1, has been used to investigate this issue. The model is forced by observed SSTs and is integrated for 45 years with different initial conditions for each run. We focus our analysis mainly on the winter geopotential height at 500-mb, defined as averages over December, January and February. Simple EOF analysis indicates that the response over the North Pacific is close to being quasi-linear where the so-called Pacific North American (PNA) pattern gets 'synchronized' with the ENSO, while the response over the North Atlantic sector is more likely nonlinear.
A probabilistic approach (Hannachi and Allen, 1999) is then introduced to detect this response based on maximizing and minimizing the respective probability density functions (pdfs) of the ensemble mean and the estimated internal noise. The method demands that the ensemble mean be split into clusters according to the phase of the southern oscillation and then the signal pattern in each cluster found.
The analysis reveals that over the North Pacific in winter, La Niña appears to trigger the negative PNA pattern (Figure 1a) while during El Niño periods the response (Figure 1b) bears some similarities to a zonally stretched PNA-like pattern, but is not precisely the inverse of the response corresponding to La Niña (-PNA). Hoerling et al. (1997) observed similar behaviour. They found, using observations as well as model simulations, that the atmospheric response to extreme phases of ENSO exhibits appreciable nonlinearity. Over the North Atlantic the two response patterns are different. A dipolar structure tilted north-east/south-west is obtained during El Niño, whereas during the opposite phases a tripole response pattern emerged. Finally, investigations of the relationship between the spring atmospheric variability and the ENSO during the spring and the winter, using signal-to-noise ratios and canonical correlations, reveal that the spring atmospheric response is more related to Pacific SSTs during the previous winter, rather than to contemporaneous SSTs during spring. A possible explanation of this relationship, which is under investigation, is that the accumulated snow cover over Eurasia during winter may in fact feedback onto the atmosphere during the spring when it starts melting.
References
Hannachi, A., and M. Allen, 1999: A probabilistic approach to optimal filtering. Submitted to Tellus.
Hoerling, M. P., A. Kumar, and M. Zhong, 1997: El Nino, La Nina, and the nonlinearity of their teleconnections. J. Climate, 10, 1769-1786.
A. Hannachi
University of Oxford
han@atm.ox.ac.uk
