Warning: This writeup was done to help organize
my thoughts prior to the presentation.
The actual presentation may have been quite different!
Heat Balance (Cronin and McPhaden (1996))
Salinity Balance (Cronin and McPhaden)
Momentum Balance (Cronin, McPhaden, Weisberg)
UPPER OCEAN TEMPERATURE, SALINITY, AND VELOCITY VARIABILITY IN THE WESTERN EQUATORIAL PACIFIC WARM POOL.
The outline of this talk will be as follows: I'll begin with a short introduction about the Western Pacific Warm Pool and PMEL's involvement in the Coupled Ocean Atmosphere Response Experiment.
Then as advertised in the title of the talk, I'll briefly describe key results from our analyses of the heat, salinity and momentum balances in the western equatorial Pacific Warm Pool.
I think it's important to remind ourselves here that this is not your usual type of science talk. The purposes of these 5-year planning talks are to:
1) Show you what I'm thinking about and give you an appreciation for the upper ocean variability in the Western Equatorial Pacific Warm Pool.
2) Find common ground with other investigators in this division, The common ground includes not only the obvious climate issues, but also issues like how to measure air-sea heat, freshwater, and momentum fluxes from moored buoys, data issues, and the use of the TAO buoys in process studies,.... to name a few....
3) Discuss future plans. The analyses which I'll be discussing are in various stages of completion. The heat balance has been accepted with minor revisions at JGR. While the salinity and momentum analyses are in the more preliminary stages. Mike, Bob Weisberg of Univ. of South Florida and I have just recently got 3-years of funding to do TOGA-COARE enhanced monitoring array analyses. So what I'll be describing is our 3-year plan rather than a 5-year plan.
This figure shows the time-longitude plots of the equatorial zonal winds, SST, and thermocline depth. The Warm Pool is the body of water with SST's in excess of 28 C. Generally, the Warm Pool is west of the date line, however during El Ninos, the Warm SST's can extend out towards the eastern Pacific.
The extremely warm SST has a strong influence on the overlying atmosphere. Deep atmospheric convection is generally confined to regions with SST greater than 28C. Likewise, the ocean can respond to variability in the atmosphere. As shown in the left panel, the winds in the warm pool region are generally weaker than in the eastern Pacific, and are modulated by these 30 to 60-day westerly wind bursts which are associated with deep atmospheric convection and are surface expressions of the Madden-Julian Oscillation.
Billy has shown how the westerly bursts set off downwelling Kelvin waves which propagate into the Eastern Pacific. As I'll be showing in this presentation, in the Western Pacific, the combination of local and remote forcing due to these wind bursts result in very complicated upper ocean variability.
A major goal of the TOGA Coupled Ocean Atmosphere Response Experiment (COARE) was to describe and understand the principal processes responsible for the coupling of the ocean and the atmosphere in the western Pacific warm pool system.
During the Intensive Observation Period, between November 1992 and March 1993, there were 7 aircraft, 12 ship, and 16 buoys taking measurements within the IFA, a patch of ocean centered at 2S/156E. This was imbedded within a larger array of TAO and ADCP moorings, surface drifters, and island stations. My analyses have used data from the enhanced monitoring array of TAO and ADCP moorings, and the PROTEUS buoy at 156E/0N.
The PROTEUS BUOY measured wind speed and direction, air temperature, relative humidity, shortwave radiation, rainrate, sea surface temperature and salinity and subsurface temperature down to 4.. m, subsurface salinity in the upper 2.. m, and subsurface velocity in the upper .. m.
So I've shown how the TAO moorings can be loaded up with extra instrumentation.
In COARE, the TAO array was enhanced with additional ATLAS which were also loaded up with extra instrumentation. Several PIs of various nations and institutions were responsible for different components of the enhanced monitoring array. Roger Lukas, Joel Picaut and Mike McPhaden were responsible for SEACAT temperature and salinity measurements on various moorings, Mike had Rain Gauges on these 6 moorings, and shortwave radiation on the PROTEUS moorings. Kunio Kutsuwada had ADCPs next to these 3, Bob Weisberg had ADCPs at these 3 sites.
As you can see, there is a lot of data and it bookkeeping for the data availability has been difficult. I've found this postage stamp figure to be helpful. The full postage stamp length represents 3 years of data from jan 1992 through Dec 1994. A version of this figure for TAO data is now included in the TAO workstation, and Nancy Soreide is now creating similar types of time lines for all TAO data.
Bulk AirSea fluxes. I don't want to spend alot of time on this, but.... In order to analyze the surface layer heat budget, we want to be able to use these surface meteorological and oceanic measurements to estimate the air-sea heat fluxes. I have been actively involved in testing, debugging and validating the official COARE flux algorithm.
[At the beginning of COARE, the air-sea flux group ran a single dataset through a variety of bulk flux algorithms. The resulting fluxes had a scatter of over 20 W/m2. Therefore Chris Fairel of the Boulder NOAA lab led a major effort to develop a bulk algorithm that could correctly estimate the fluxes of heat, momentum, and moisture across the air-sea interface in the Warm Pool region. ]
[Bulk formula in general have the form QH = CT du dT, QE = Cq du dq where CT and Cq are the exchange coefficient, du is the wind speeds relative to the surface currents, dT is the air temperature relative to the SST, and dq is the specific humidity relative to the saturated specific humidity at the sea surface. By similiarity theory, these profiles will be logarithmic within the constant flux layer and the fluxes can be written in terms of the frictional velocity, ustar, and scale temperature and scale specific humidity. These scale values, and therefore the fluxes, can be determined by solving the profiles iteratively.]
[A problem with the standard bulk aerodynamic formulae is that the relation becomes non- singular as the bulk wind speeds go to zero. In truth, the winds are never zero although they may average to zero. ]
In the COARE algorithm, a gustiness factor is added to the relative wind speed to account for the fact that the scalar average wind speed never goes to zero, although the vector averaged wind speed might.
Likewise, during these very low wind speed days, the top 2 m of the surface can become very warm and stratified. This figure is courtesy of Bob Weller at WHOI. The IMET buoy had thermisters at 50 cm intervals between 40 cm and 2.5 m. On low wind days, the SST at 40 cm can be a degree warmer than the SST at 1 m. This begs the question, What is SST? How do you compare SST from satellite, to SST from buoy to SST from ships? In the COARE bulk algorithm, this warm layer is accounted for using a simplified Price-Weller-Pinkel 1-dimensional mixed layer model. I'll discuss this model in more detail later in this talk. Additionally, because all the cooling occurs in the top few mm, a cool skin correction is included in the skin SST.
Comparisons with the ship directly measured fluxes indicate that the algorithm is good to within 10 W/m2. The intercomparisons with R/V Hakuho-Maru, R/V Natsushima and R/V Moana Wave data during periods when these ships were parked within 5 km of the buoy were also used to validate the data.
OK lets cut to the chase here. The top panel shows the heat storage above the top of the pycnocline at 0,156E from Sep-Dec 1992. This storage rate is repeated as a dashed line in each of these panels of the various terms in the heat balance. Because the Warm Pool is relatively homogenous in comparison to other regions of the world's oceans, it is generally believed that advection is negligible. Also, since the pycnocline is sometimes controlled by the halocline rather than the thermocline it is generally believed that mixing is small and thus the storage rate is controlled by variations in the net surface heat flux.
To a certain extent this appears to be true, as can be seen in the second panel. In late October there was a westerly wind burst, the net surface flux is negative due to a reduction in the Qsw and increase in the latent heat loss, and the surface layer indeed cools.
A surprizing result of this analysis, however, is that at times, for example in early October 1992, advection can be the dominant term in the heat balance in the Warm Pool.
While this goes against the common belief, perhaps it shouldn't be so surprizing. Although the temperature gradients are negligible in the warm pool in comparison to other regions, the overall variability is also much smaller. So in fact, with a strong equtorial surface jet, it is not difficult to account for much of the observed variability in terms of zonal advection.
Curiously, during COARE IOP the gradients were indeed negligible and the investigators are finding that balance is nearly one-dimensional.
This discussion of the mixed layer depth variability has assumed that the dynamics are all local -- that the only spatial gradient in the equations of motion is in the vertical direction. One- dimensional models show how surface forcing, mixing, produces a local response in temperature, salinity, velocity. 1-d models break down if horizontal gradients in temperature, salinity, velocity or pressure are important. Conventional wizdom is that 1-d models are fairly good in the warm pool since the horizontal temperature gradients are very small.
The model is initialized in at the beginning of the record with CTD data. Surface fluxes are used as boundary conditions, and the flux profiles are calculated by assuming that mixing occurs under the following 3 conditions: 1) static instability, 2) a bulk Richardson # less than .65, 3) a gradient Richardson # less than .25. ]
So how bad is it. Climate modellers would like to be able to get the SST correct using a 1- dimensional model.
The top panel shows the 1-Dimensional Price-Weller-Pinkel model simulated SST in red vs. the observed SST in blue. Although the 2 time series diverge in October, as can be seen in the 2nd panel, the model is excellent at simulating the high frequency variability throughout the entire record. The cross correlation between the 5-day high-passed simulated SST and the observed SST is almost 0.9.
And surprizingly even at longer time scales the model is fairly good at modelling the SST tendencies. However, by missing an event, the overall SST value ends up being wrong.
The model also produces a salinity field time series. The top panel shows....
The difference between the observed SST tendency and the 1-d model SST gives an estimate of the 3-dimensional processes, model and data errors. This is shown in dashed.
We can compare these estimates of the 3-d and mixing rates to those estimated in the empirical salinity balance. Advection and the 3-d rate are generally of the same order and have similar features.
The salinity analysis however is made difficult by sharp salinity fronts which cannot be resolved with the mooring array. This figure is courtesy of C. Henin, J. Grelet and M-J. Langlade and shows the thermosalinograph along a north-south ship track through 0,156E. As you can see, with moorings at 2S, 0, and 2N, the meridional salinity gradient will be underestimated by a factor of two (at least).
Another difficulty is that the salinity sensors can have calibration drifts which are as large as the signal. The drift, due primarily to biofouling, can be seen by taking the difference between sensors on the same mooring. For example as shown here, 3m - 5m, and 5m - 10m. As you can see, this steady increase in density and resulting convectively unstable stratification at 5m is unphysical and is due to the calibration drift. By sorting through all the pairs, we hope to remove much of this drift from the salinity data. It represents however a huge amount of work. Paul Freitag will be responsible for the processing of the PMEL data.
Not all the moorings have such a bad calibration problem though. Here is the data at 0,156E. As you can see, the PMEL mooring at 0,156E is in good shape. While there are some periods of convective instability (i.e. heavy water overlying lighter water)....
Paul Freitag has made a histogram of the duration of these instability events and 99% are less than 2 hours in length.
Furthermore, looking at the diurnal cycle, most of these events occur at 4 in the morning, exactly when we expect to see the highest amount of turbulent mixing.
This is an example of the very important and healthy symbiosis between the the data collection, data processing and data analysis groups. If we want high quality data and high quality science, then it is important to have these functions done in house.
The velocity time series are very interesting too. The wind burst is also associated with an eastward acceleration of the surface waters.
This slide shows the zonal velocity along the equator at 0,147E to 0,165E. Red is eastward ... here is the EUC. Blue is westward.... here is the SEC. Notice the amount of variability. These data provide a challenge to modellers. I've shown this to Ed Harrison and other modellers and have asked "Do you see this amount of structure in your models?". And so far, they all say no.
For me, I look at these plots and wonder, what is going on? What are the forces and mechanisms responsible for this variability?
Bob Weisberg put out the ADCP mooring at 0,157.5E and .75N and .75S so as to be able to compute vertical velocity at the central site, 0,156E. This slide shows a preliminary estimate of the vertical velocity. We will be working with Bob Weisberg to analyze the dynamics occuring at this site.
again, ALL terms are important.
Temperature and salinity can have sharp vertical gradients in the near surface and significant horizontal gradients in the upper ocean. The vertical structure of the zonal velocity shows a variety of reversing jet structures.
All terms in the heat, freshwater, and momentum balances are important in the Warm Pool upper ocean variability.
The COARE bulk air-sea flux algorithm v2.5 is available.
TAO buoys can be used as platforms for a full suite of instrumentation. Embedding the process study array within the TAO array, allows the local analyses to be place within a framework of large-scale, longterm ocean variability.
Process studies improve the quality of the TAO data by providing data redundancy and additional diagnostic tests. Lessons learned during these process studies will ultimately lead to a better next generation of moorings.
Meghan F. Cronin
Pacific Marine Environmental Laboratory
7600 Sand Point Way NE
Seattle, WA 98115 USA
Meghan Cronin's Home Page
DOC | NOAA | OAR | PMEL