- High-frequency (small-scale) events of upwelled isopycnal surfaces are observed in the two WHOTS time series (see
*this note*): - It is shown that a) neither the internal wave field nor the tides are responsible for these events and b) 5-day snapshot output, as in the ARGO time series, is sufficient to capture most of them. The ARGO-like time series can, however, miss an event or underestimate its amplitude.
- The nitrate contours in the three subtropical ARGO data are correlated with density contours with a correlation coefficient of about 0.70-.75 (see
*this note*). Contrast this result with the 0.23 correlation coefficient of the ARGO data near station PAPA. - Finite-size Lyapunov exponents (FSLE) were computed for the first WHOTS deployment. No significant correlation is found between the presence of filaments and the events of shallow isopycnal surfaces. Only in 50% of the cases (4 out of 8), we could say the event occurs near a filament. Also there is a discrepancy between the number of events and the number of filaments that weakens the hypothesis that the two are linked.
- The exact location and amplitude of the filaments, estimated from FSLE, is quite sensitive to the SSH product. Given the noisy aspect of a FSLE time series, it is difficult to have reliable conclusion concerning the relationship between FSLE and the shallow events.

- It is possible that the AVISO SSH product is not good enough so that the location of the filaments as presently estimated is incorrect.
- Am I proving that the small-scale events are more likely inside eddies (or away from filament)?
- Can the events be produced (also) by submesoscale
**vorticies**and not filaments (as observed by Chavanne*et al*. 2010), not resolved by AVISO SSH, not predictable from FSLE calculation using AVISO SSH, yet strong and intense enough to upwell the isopycnal surfaces. If true, we should have a skewness in the distribution of the depths of isopycnal surfaces (associated with small-scale anomalies only), with more shallow than deep surfaces.

- Characterize the scale and frequency of the events in the ARGO data. Do these characteristics fit the characteristics of the WHOTS events? Especially, K. Johnson speaks of a frequency of one event per month. This frequency is larger than that in the WHOTS data. Maps and timing of these events in relation with their scale.
- Use lag correlation of SSH data to estimate the zonal speed of coherent structures and the zonal scale of the small-scale events.
- Sensitivity of the FSLE calculation: Use the 1/30th-deg. OFES simulation, assume the velocity is stationary and compute the FSLE with different SSH products.
- If the FSLE calculation is considered good: Study the relationship between FSLE and density anomlies in ARGO data. Is there a correlation and if yes of what sort?
- Assume the events are due to submesoscale vorticies: Use the ADCP data taken during WHOTS to estimate their relative vorticity (use SSH data and lag correlation to estimate the zonal speed of the events and assume that the vorticies are rotationally symmetric)
- About the hypothesis of submesoscale vorticies: Do we see them in the 1/30th-deg. OFES simulation?