Work in progress: The role of meridional advection in barrier layer formation

Work in progress: The role of WWB in barrier layer formation

This webpage is no longer in use. Please see published results:

Cronin, M. F., and M. J. McPhaden. Barrier layer formation during westerly wind bursts. J. Geophys. Res., 107, 8020, doi:10.1029/2001JC001171, 2002.


  • Oct-Nov 1992 WWB event
    This shows the relationship between the daily BLT, the warm pool/fresh pool and surface forcing for the Oct 1992 WWB. Fields shown are: Reynolds SST, SSS from moorings, 5-day gridded TAO winds, and Xie and Arkin Rains, and the daily BLT computed using only moored SEACAT data.
    Also check out the Oct-Dec 1992 period.

    I'M NOT HAPPY WITH THE TAO 5-DAY WINDS AND REYNOLDS SST. THESE FIELDS ARE TOO SMOOTH. Notice that high rainfall and low zonal winds is not sufficient to produce a thick BL since otherwise would have developed one at 156E ~Oct 15-20 1992. Instead, the zonal current and shear seem to be necessary.

  • U, T & S zonal section during formation of BL at 160.5E & 165E.
    The difference between these 5 panels is just how many days I use in the average -- 1-5 days. When viewing zonal sections, it's good to do some averaging. 5 days is standard. However, you lose some of the vertical structure. So am undecided. In the paper, I would only show 1 of these.
    Any one of these panels shows that:
    * Between 156E & 158E there is large dUdx -- and is perhaps responsible for the large dSdx and dTdx there.
    * At 154E-156E, dUdz is near zero or negative above 30 m (not surface intensified eastward flow) -- no tilting at these sites -- (and MLD is in fact very deep).
    * At 157.5E and definitely at 165E there IS surface intensified eastward flow -- Tilting can occur -- and in fact, the isohalines are tilted.
    * There is indications of a weak temperature inversion at 165E.



  • BLT vs. SST vs. SSS This shows the relation between the warm pool SST, the SSS front, and barrier layer thickness (BLT). Notice how BL is thick on the eastern edge of the freshpool, just west of the SSS front. According to analyses by Vialard and Delcluse, Picaut et al., and Henin et al., the zonal SSS gradient is caused by horizontal convergence. The convergence leads to vertical motion and subduction. Warm salty water subducted near this front can lead to the formation of a thick bl (e.g. Lukas and Lindstrom).

    The zonal extent of the fp and wp vary interannually. BL tends to be thick ~165E when the warm pool and fresh pool are limited to the western Pacific. During El Nino, Ando and McPhaden show that thick bl extend across the Pacific and are slightly less thick in the western Pacific. This variability has been discussed by Vialard and Delecluse, Henin et al., Ando and McPhaden and others.

    Note, the SSS in above figure is from DELCROIX ET AL. (2000). Check out similar figure using only moored SSS. Notice have stronger gradients.

  • BLT vs. SST vs. monthly moored SSS CI=0.2psu; thick contour is 35 psu.

    Although we know that the BL will reduce entrainment cooling, I challenge you to find a correlation between the SST and BL. SST is controlled by many factors, one of which is mixing.

    What's new here is that this is from an independent data set of previously published longterm BL observations (Ando and McPhaden). BLT are computed from moored data. Zhang and McPhaden used this data set to compute BLT for Sep 1992 - Dec 1993. Their method is slightly different, althoug the BLT produced are roughly similar to our 0, 165E values for that period. Our time series is longer and also done at other zonal sites.

    Although we have a more limited spatial (and temporal) range than the CTD analysis (Ando and McPhaden), we have better time resolution. Like the Zhang and McPhaden analysis, we can resolve local variability due to westerly wind bursts. These burst are intrinsically related to the termination of the trades and the horizontal convergence of surface flow that is responsible for the SSS front. While the Zhang and McPhaden analysis focused on local balances, we are interested in the spatial structure and the role of advection.

  • BLT vs. zonal winds vs Xie and Arkin rain rate As shown by Zhang and McPhaden, rainfall in the WP/FP tends to occur during WWB. During the period 1992-1994, Zhang and McPhaden found that WWB tended to be locally associated with erosion (rather than formation) of barrier layers.

    While this is true, cross correlation between zonal wind and BLT, and dBLT/dt are not significant. During the extended period beyond COARE, there are WWB that also resulted in very thick BLs. For example, See WWBs in late 1989, late 1992, and Spring 1994.

    THE BIG EVENT at 0, 165E in NOV89

  • T&S and U&V with MLD& MLDT for Nov 89 - Dec 89. The difference between MLDT and MLD is the barrier layer. This is the Famous WWB documented by McPhaden et al. 1992. During the WWB, the surface freshened by over 1 psu. Thus, while the thermocline deepened due to Ekman convergence and downwelling, the surface mixed layer remained shallow and a thick BL formed.

    The source of this fresh water is from the north and the west.

  • dTdt & dSdt with U&V and MLD& MLDT for Nov 89 - Dec 89. The west is fresher, but from the time the salty water flowed westward before the event to when the very fresh water arrived back again at 0 165e (~a couple of weeks), it would have required ~ 2 m of rainfall. This was not observed. the GPI precip shows at most ~0.6 m (30 days * 20 mm/day -- a generous estimate).

    However, this explaination does not take into consideration zonal convergence of the flow. Some of the water may have also originated further to the west. The gridded monthly SSS from shipdata (Delcroix et al. 2000) show water fresher than 34.5 west of 155E. For this water to reach 0 165E in 13 days would require a ~ 1 m/s eastward jet. This is observed. The fact that the zonal current reversed direction also indicates that this is a region of zonal convergence and thus frontogenesis. ... Which we need, since if the dSdt during Nov 22-24 was entire caused by zonal advection, the dSdx would need to be between 0.14 psu/100 km and 0.8 psu / 100 km. Rain is occurring, so in fact, the required gradient is probably less than this. However this is a big front!

    Note however that while the surface eastward jet is trapped above the mixed layer, the subsurface westward jet is below the top of the thermocline and therefore is not at the level required by the zonal subduction barrier layer formation mechanism. Rather, this looks like a surface phenomena.

    McPhaden et al. Figure 9 shows 34.4 psu water at ~3-4N. With a 30 cm/s southward current, that freshwater would reach the equator in ~13 days. Northward subsurface flow also has the effect of tilting zonal gradients into the vertical. However during the big freshening Nov 20-22, the meridional current was northward! which would be bringing Salty water to the equator rather than freshwater. Later on, southward advection at the surface and northward advection at depth probably played a role in maintaining the bl.


  • TAO gridded winds and Time-latitude sections of SSS at 155E and 165E from Delcroix et al gridded SSS fields. For meridional advection to be an important mechanism in equatorial barrier layer formation, there must be equatorward flow bringing freshwater from off-equator to the equator. This figure shows that the zonal migration of the 35 psu SSS front tends to be limited latitudinally to the equator. It is nearly always 34-34.5 psu along the ITCZ band ~ 6-8N. Consequently, as the equator becomes very salty (>35 psu) a meridional SSS gradient (front) can form with off-equator waters substantially fresher than at the equator.

    Now, how to get that freshwater to the equator? Switch the winds to westerlies and the Ekman convergence will bring this water to the equator. ... in theory. In practice, the Ekman convergence can sometimes be displaced meridionally or wobble around the equator. We won't worry about that, except that this can sometimes cause a non-symmetric v around the equator and also a v on the equator.

  • WARNING: THE GRIDDED SSS FIELDS ARE OFTEN BASED ON NO DATA. In this figure the grid boxes with no data are blanked out. Not alot left!



  • Subsurface circulation and BL formation associated with OCT92 WWB This is also another very well documented WWB, occuring during the COARE IOP. While this event is not necessarily a typical event in that it formed away from the 35 psu contour, it does show all the basic blt formation mechanisms at work: rainfall, zonal tilting, meridional tilting, (and possibly stretching).

    At the start (16-20 Oct) we see westward flow at the surface advecting warm salty water to the west. MLD ~MLDT and both are shallow.
    21-25 Oct: The WWB begins to cause eastward flow in the west, thus a dudx convergence. Downwelling at 156E and mixing causes the MLD & MLDT to deepen.
    26-30 Oct: dudx & d(rain)/dx ?? cause the dSdx to intensify.
    31 Oct - 4 Nov: dUdz & dVdz between 157.5E and 165E now tilts the dS/dx & dS/dy? front into the vertical, generating a ~30 m thick BL.

  • Conclusion. The physics are complex. Significant shears are observed above the thermocline. These shears are both caused by and cause shallow salt stratified mixed layers.

    dVdz might be most important for first WWB after period of easterlies.

    Meghan F. Cronin
    Pacific Marine Environmental Laboratory
    7600 Sand Point Way NE
    Seattle, WA 98115 USA
    Return to:
    Meghan Cronin's Home Page
    Privacy Policy | Disclaimer