Triggering of El Niño by westerly wind events in a coupled general circulation model

Abstract

Two ten-members ensemble experiments using a coupled ocean-atmosphere general circulation model are performed to study the dynamical response to a strong westerly wind event (WWE) when the tropical Pacific has initial conditions favourable to the development of a warm event. In the reference ensemble (CREF), no wind perturbation is introduced, whereas a strong westerly wind event anomaly is introduced in boreal winter over the western Pacific in the perturbed ensemble (CWWE). Our results demonstrate that an intense WWE is capable of establishing the conditions under which a strong El Niño event can occur. First, it generates a strong downwelling Kelvin wave that generates a positive sea surface temperature (SST) anomaly in the central-eastern Pacific amplified through a coupled ocean-atmosphere interaction. This anomaly can be as large as 2.5°C 60 days after the WWE. Secondly, this WWE also initiates an eastward displacement of the warm-pool that promotes the occurrence of subsequent WWEs in the following months. These events reinforce the initial warming through the generation of additional Kelvin waves and generate intense surface jets at the eastern edge of the warm-pool that act to further shift warm waters eastward. The use of a ten-members ensemble however reveals substantial differences in the coupled response to a WWE. Whereas four members of CWWE ensemble develop into intense El Niño warming as described above, four others display a moderate warming and two remains in neutral conditions. This diversity between the members appears to be due to the internal atmospheric variability during and following the inserted WWE. In the four moderate warm cases, the warm-pool is initially shifted eastward following the inserted WWE, but the subsequent weak WWE activity (when compared to the strong warming cases) prevents to further shift the warm-pool eastwards. The seasonal strengthening of trade winds in June–July can therefore act to shift warm waters back into the western Pacific, reducing the central-eastern Pacific warming. This strong sensitivity of the coupled response to WWEs may therefore limit the predictability of El Niño events, as the high frequency wind variability over the warm pool region remains largely unpredictable even at short time lead.

Appendix 1

1.1 The Mann–Whitney test

The Mann–Whitney test is a non-parametric procedure, which is powerful to test the hypothesis that two sample populations (X and Y) have the same mean of distribution against the hypothesis that they differ. The test is derived by examining the probability distribution of a linear combination of the ranks of the population under the null hypothesis that all the values are sampled from the same continuous distribution.

The procedure ranks the population’s values from smallest to largest, assigning the rank 1 to the smallest observation, 2 to the next largest, and so on up to rank n, the number of elements in the two populations. Then, the Mann–Whitney statistics for X and Y are defined as follows

$$ \begin{aligned} & U(X) = n_x n_y + n_x (n_x + 1)/2 - W_x \\ & U(Y) = n_x n_y + n_y (n_y + 1)/2 - W_y \\ \end{aligned} $$

where n _x and n _y are the number of elements in X and Y, respectively, and W _x and W _y are the rank sums for X and Y. The test statistic Z, which closely follows a normal distribution for sample sizes exceeding ten elements, is defined as follows

$$ Z = \frac{{U{\left( x \right)} - (n_{x} n_{y} )/2}} {{{\sqrt {(n_{x} n_{y} (n + 1))/12} }}} $$

When this probability level is sufficiently small, we reject the null hypothesis and conclude that the two sample populations do not come from the same distribution.