Intraseasonal variability of wind waves in the western South Atlantic: the role of cyclones and the Pacific South-American pattern

Extratropical cyclones are known to generate extreme significant wave height (swh) values in the western South Atlantic (wSA), which are highly influenced by intraseasonal scales. This work aims to investigate the importance of intraseasonal time scales (30–180 days) in the regional wave climate and its atmospheric forcing. The variability is explained by analyzing the storm track modulation due to westerlies winds. These winds present time-scales and spatial patterns compatible with the intraseasonal component of the Pacific South–American (PSA) patterns. The analysis are made using ECMWF’s ERA5 from 5 1979 to 2019 and a database of extratropical cyclones based on the same reanalysis. Empirical orthogonal function (EOF) analysis of the 10m zonal wind and swh are used to assess the westerlies and waves regime in the wSA. The EOF1 of u10 presented a core centred at 45◦W and 40◦S, while the EOF2 is represented by two cores organized into a see-saw pattern with a center between 30◦S–40◦S and another to the south of 40◦S. Composites of cyclone genesis and track densities, and swh fields were calculated based on the phases of both EOFs. In short, EOF phases presenting cores with a positive (negative) u10 10 anomaly provides a favorable (unfavorable) environment for cyclone genesis and track densities and, therefore, positive (negative) swh anomalies. The modulation of the cyclones track are significant for extreme values of the swh. The spatial patterns of the EOFs of u10 are physically and statistically consistent with 200 hPa and 850 hPa geopotential height signals from the Pacific, indicating the importance of the remote influence of the PSA patterns over the wSA.

. EOFs of filtered anomalies of (a,b) 10-m zonal wind (u10) and (c,d) significant wave height (swh) calculated on the wSA. The percentages represent the explained variance of each mode.  The values in the PC correlation matrix of the filtered u10, v10 and swh (Table 1) show correlation coefficients greater than 0.50 between the EOF1 of u10 and swh and also for EOF2, while the correlation between the EOF1 (EOF2) of u10 and EOF2 (EOF1) of swh present correlation values of 0.40 (0.21). Correlations values lower than 0.4 were not analyzed, which was the case of the correlation coefficients between EOFs of v10 and swh. The correlation values (highlighted in the table) indicate a   . Wavelet power spectrum of the EOF1 of swh anomaly (a) and u10 anomaly (b), the black contours represent the 95% confidence level. Global wavelet spectrum of swh (c) and u10 (d) for the anomaly fields (black) and filtered anomaly fields (red). Dashed lines represent the 95% confidence level considering a red noise based on a univariate lag-1 autoregressive model.

Wave and wind spatio-temporal variability
In the following section, we built composites of the wave and wind fields based on the phases of EOF modes of u10 and swh, in order to better understand the intraseasonal relationship between the variability of swh, cyclone genesis and track densities.

Cyclone genesis and track densities
The genesis and track densities differences between u10 EOFs phases (A minus B) are shown in Fig. 6, where regions with significant values are indicated by black dots. The track and genesis density for u10 EOF1 can be found on Appendix B ( Fig. B1), where it is possible to evaluate the spatial patterns of density in order to understand the differences discussed in this section. The density differences computed for swh EOFs revealed similar patterns for u10, and, for clearance, they are 175 presented on Appendix B (Fig. B2). The swh related fields are slightly weaker, showing a weaker response of swh EOFs phases, which is expected once this field is also influenced by remote forcing (i.e., swell). In Fig. 6, the spatial pattern of the   In the composites of phases A and B of the EOF2 of u10 (Fig. 9c,d), 95 and 63 HWEs were identified, respectively. Phase A is associated with a larger HWE number and also presents cyclones that occur mainly to the southeast, but also to the Here, the trough in the wind vectors is present in phase A, while a regional cyclonic wind pattern is centered near 25 • S and 42 • W in phase B.

Evaluation of remote signal 230
Up to this point, we have shown that the EOFs of u10 within the intraseasonal band are significant for the variability of cyclones and waves in the wSA, especially regarding the extreme wave climate. The remaining question is: Is this variability pattern a part of a larger organized system? Over the wSA, the spatial patterns revealed by the EOFs of u10 are similar to features of the PSA found using EOFs of the geopotential height (Irving and Simmonds, 2016). There is little agreement to what EOFs of the geopotential height represent to the sub and extratropical environment between the South Pacific and South Atlantic -some 235 associate it to SAM and PSA (Ding et al., 2012), others relate it to an eastward propagating wave train that may be connected to MJO influence on the western South Pacific (Paegle et al., 2000;Liebmann et al., 2004;Irving and Simmonds, 2016). However, we will not consider here the effect of MJO, since its effects are observed only during austral summer (e.g., Liebmann et al., 2004;Rodrigues and Woollings, 2017). Following the discussion and results presented by O' Kane et al. (2017), except on the summer, the subtropical jet works as a barrier to waves propagating from the tropics (e.g., Hoskins and Ambrizzi, 1993;240 Ambrizzi and Hoskins, 1997), reducing the influence of MJO in the SH extratropics. Restricting the analysis to a unique season would reduce the sample size and would eliminate seasons where the cyclones and wave climate are more severe in the region (e.g., Pianca et al., 2010;Gramcianinov et al., 2020c).
PSA modes evaluated by EOFs in the South Pacific and South Atlantic domains require multiple EOF modes (geopotential height, for instance) to depict the wave train signal in the atmosphere that extends from the central South Pacific to the South 245 Atlantic (O'Kane et al., 2017;Irving and Simmonds, 2016). In contrast with the usual approach of using the geopotential height fields, the evaluation of the signal that propagates from the South Pacific into the wSA is made using Hovmöllers diagrams.
These diagrams do not separate the variability in different modes, but provide valuable insights into the associated variability that are difficult to interpret using the EOF approach.
The PSA patterns occur across the South Pacific and South Atlantic domains, but the analysis of EOF in such a large domain 250 would interfere and smooth the variability signal observed in the wSA. To evaluate the spatial distribution of the PSA patterns, usually observed in the Z200 and Z850 fields, we used composites during the different phases of the EOF of u10. 4.2 Impacts on extreme waves regional climate The composite analysis described above provides a general view of the processes, but in order to have a more in-depth idea of 300 what is happening in a specific region and include the cyclones dataset into the analysis, we evaluate the composites of highwave events (HWE) (Fig. 9). The selected region of this analysis lies purposely outside of areas with significant differences, which shows that the area of influence of the EOFs of u10 are not restricted to the significant areas represented in Figs. 7 and 8.
During EOF1 of u10, the similarity in the number of HWE is consistent with the fact the cyclone tracks associated with the 305 EOF1 of u10 are mainly concentrated to the south of 40 • S, which makes this area less likely to be affected by a given phase.
In phase A we see high swh values concentrated over SBB region due to the increase in genesis in LA PLATA (Fig. 7), but for cyclones with small spatial range,i.e., short tracks (Fig. 6a). On the other hand, phase B presents an enhanced cyclonic activity between 40 • S and 55 • S, associated with the increase in genesis in the ARG region (Fig. 6d). This cyclone behavior justify the spreaded swh pattern in phase B composites (Fig. 9b), once its contribution to the wave field in the region is also linked to its 310 variable spatial scale (1500 -2000 km radius) and swell propagation. In fact the shww percentages in the composites show that the remote effect in the total wave field is larger in phase B than in A.
The EOF2 of u10 presents a larger number of HWEs during phase A and this is a consequence of the higher genesis and track densities between 35 • S-45 • S. In phase B, more cyclones are generated southward of 45 • S, which lead to lower and less intense HWEs. However, a small region over the SE Brazilian coast (30 • S) presents more genesis (negative values in 315 Fig. 6d), indicating a wind forcing source close to SBB. The cyclones generated in SE-BR are usually weaker and short-time systems, which can generate HWE but not as much as the LA PLATA region (Gramcianinov et al., 2020c). Therefore, the swh composites are still presenting relatively high sea-wave percentages, although the swh are lower than the ones presented in the composite of phase A. One can note that the proximity of active cyclogenesis region influences directly the percentage of sea-wave relative to the total swh. However, the percentage never crossed the 60% value, revealing a high influence of remote 320 forcing in the SE Brazilian coast wave extremes. Gramcianinov et al. (2020c) evaluated extratropical cyclones environments in the wSA and found three situations where extreme waves are generated: (1) west/southwestward of the cyclone center, behind the cold front; (2) north/northwestward of the cyclone center, ahead of the cold front, and; (3) eastward of the cyclone center, along the warm front. The most commonly observed case in the composites is the situation 1, which is identifiable by the amount of cyclone centers placed to the east of 325 the reference point (SBB) and the trough in the mean wind field. Situations 2 and 3 are more difficult to identify but are likely to be associated with the cyclone centers over the ocean occurring to the south and southwest of SBB, and also over land.