Table Of Contents

This Page

The vertical velocity field in the 1/30th and 1/10th-of-a-deg OFES models

1/30th-of-a-deg OFES snapshot

We present here the vertical velocity field w, the ratio R of the relative vorticity ζ to the Coriolis parameter and the Okubo-Weiss parameter (OW) for the the snapshot of Jan. 1, 2001 in the 1/30th-of-a-deg OFES snapshot.

We choose a larger domain (Fig. 1) that in this note as it now includes instances of eddies strong enough that R reaches, inside them, a significant value (Fig. 2). If we define submesoscale features as R being large as in Thomas et al. (2007), then there are instances where a submesoscale feature occupies the inside of eddies. In this larger domain, the OW parameter has a more convential pattern with large negative values inside eddies and large positive values around them (Fig. 3).

Yet, from the point of view of explaining the large nutrient events of Johnson et al. (2010), vertical velocity w is more central than an index describing submesoscale features. In Fig. 4 is plotted w along one latitude that crosses several mesoscale eddies and submesoscale features based on R and OW. As seen previously in this note, there are events of large vertical velocities (reaching 50 m/day) with a submesoscale structure trapped in the upper 100 m. Below it, however, the w filed loses its submesoscale characteristic and displays a structure with a broader horizontal and vertical scale. Fig. 5 reveals how the extrema in w at 50 m in the upper layer is distributed horizontally in that snapshot. Interestingly, the two well-developped submesoscale structure in w, near 136°W-19°N and near 145°W-18°N, are located along the southeast side of two anticyclones (Fig. 1). According to Thomas et al. (2007), one process to generate submesoscale eddies is a downfront wind; applied to a field of eddies bewlo which southwestward Trades, that means submesoscale structures should develop on the southeast side of anticyclones and northwest side of cyclones. It can be a coincidence as the snapshot does not show if the submesoscale features have just been formed and as we expect, once formed, the features to rotate with the eddy. More work is needed.

Fig. 6 shows the ratio R at 50 m. If w is the important quantity for the exchange of tracers between the deep and upper ocean, R does not appear to be the adequate index as the two might be out of phase so that there are many instances where one is large while the other is low and inversely, making the scattered plot between the two quantities structureless (not shown).

As the large nutrient events were observed at a deeper depth, the magnitude of w and R are shown also at 150 m (Figs. 7 and 8). This depth is below the surface mixed layer (SML) and, although R have still the same structure –albeit with less submesoscale structure– as at 50 m, w is dominated bya wave field, the phase lines of which are mostly oriented in the east-west direction. This wave field has a very large vertical scale (Fig. 10) and the signal dominates at deeper depth (Fig. 9). To know what type of wave is and if this wave can play a role in the large nutrient events, we need to analyse a time series. We can do this only with the 1/10th-of-a-deg OFES model.

../../../../../../_images/SSH_30_OFES_Jan12001.png

Figure 1: Sea surface height (SSH) on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/R_30_OFES_Jan12001_surf.png

Figure 2: Surface R on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/OW_30_OFES_Jan12001_surf.png

Figure 3: Surface OW on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/w_30_OFES_Jan12001_19N1.png

Figure 4: w along 19°N on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/abs_w_30_OFES_Jan12001_50m.png

Figure 5: Absolute w at 50 m on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/abs_R_30_OFES_Jan12001_50m.png

Figure 6: Absolute R at 50 m on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/abs_w_30_OFES_Jan12001_150m.png

Figure 7: Absolute w at 150 m on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/abs_R_30_OFES_Jan12001_150m.png

Figure 8: Absolute R at 150 m on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/w_30_OFES_Jan12001_3200m.png

Figure 9: w at 3200 m on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

../../../../../../_images/w_30_OFES_Jan12001_147W.png

Figure 10: w along 147°W on Jan. 1, 2001 in the 1/30th-of-a-deg OFES model.

1/10th-of-a-deg OFES model

We present here the same figures but for Jan. 3, 2001 in the 1/10th-of-a-deg OFES model (Figs. 11 to 19) before analysing the time series of the wave field that dominates the vertical velocity w at depth.

The different quantities (R, OW and w) are similar in the 1/10th-of-a-deg OFES model than in the 1/30th-of-a-deg OFES model with the exception that submesoscale features are, expectingly, much less present. There are still, however, some structures larger than in the 1/30th-of-a-deg OFES model that do have a submesoscale-like structure (Fig. 14), albeit with a larger horizontal scale and a decreased amplitude in w. This suggests that the 1/10th-of-a-deg OFES model starts to reproduce submesoscale-like processes.

Furthermore, as in 1/30th-of-a-deg OFES model, the vertical velocity field w is dominated at depth by a wave (Figs. 16 and 18), that has the same large vertical scale (Fig. 19). Time series of w show that the wave period is in the order of the model output (3 days). The amplitude of the wave field (shown here by its standard deviation in space over the domain in Fig. 21) is dominated by the annual cycle with a maximum in February.

../../../../../../_images/SSH_10_OFES_Jan32001.png

Figure 11: Sea surface height (SSH) on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/OW_10_OFES_Jan32001_surf.png

Figure 12: Surface OW on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/abs_w_10_OFES_Jan32001_19N.png

Figure 13: Absolute w along 19°N on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/abs_w_10_OFES_Jan32001_50m.png

Figure 14: Absolute w at 50 m on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/abs_R_10_OFES_Jan32001_50m.png

Figure 15: Absolute R at 50 m on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/abs_w_10_OFES_Jan32001_150m.png

Figure 16: Absolute w at 150 m on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/abs_R_10_OFES_Jan32001_150m.png

Figure 17: Absolute R at 150 m on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/w_10_OFES_Jan32001_3200m.png

Figure 18: w at 3200 m on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/w_10_OFES_Jan32001_147W.png

Figure 19: w along 147°W on Jan. 3, 2001 in the 1/10th-of-a-deg OFES model.

../../../../../../_images/w_10_OFES_2001_147W20N_3200m.png

Figure 20: w at 147°W, 20°N, at 3200 m in the 1/10th-of-a-deg OFES model during year 2001.

../../../../../../_images/wstd_10_OFES_2001_210E230E15N25N_3200m.png

Figure 21: Spatial standard deviation of w over the domain (210°E-230°E, 15°N-25°N) in the 1/10th-of-a-deg OFES model during year 2001.

Questions

  • Is the high-frequency wave that dominates the vertical velocity field w at depth a fluke of the model or a realistic feature? In the second case, what type of wave is it (a priori, it is an internal wave) and what generates it (given Fig. 21, we can speculate that the wave results from upper-ocean instabilities for which the eddy kinetic energy and eddy-eddy interactions peak in February (see, for instance, Fig. 4 in this note)?
  • Given the high-frequency of the wave (yet not that high compared to the response time scale of nutrient uptake), do we expect this wave to play a role in the injection of nutrients as observed by Johnson et al. (2010)? A priori no as we would have observed these injections at a much higher frequency than observed (once a month), except if rogue waves (that have sufficient amplitude) occur at that frequency?
  • In the vertical, what is likely to play a larger role, the w within the SML or below it? If it is the w within the SML, which criterion could we use to localize in space and time these large w events in ARGO/SSH data? If it is the w below the SML, but not the wave component, then what is the spatial and temporal structure of that w (it is masked in the models and 1-day averaged w may be needed), and how to locate it with ARGO/SSH data?

As usual, more questions than answers. If science were answering more questions than it asks, ignorance would quickly converge to zero and I would have no more mean to bring bread home.

PS: Eric suggested to look at the vertical displacement generated by the wave field at depth in order to deduce the frequency. A look at an isopycnal which is on average at 550 m shows that the vertical displacement has the same spatial and temporal structure as eddies suggesting that the wave field has much higher frequency that the 3-day output, is being aliaised in the model output and may play a minor role in the vertical displacement of isopycnal surfaces. This suggests that I should study w from the change in the vertical displacement of surfaces instead.