What We Learned From EPIC -- a Personal Perspective

So, what did we learn from this project? I will leave the stratus region south of the equator to someone like Chris Bretherton, who can probably do a much better job of describing the discoveries there. Similarly, someone like Meghan Cronin is in a better position to talk about the oceanography. I will confine my comments to the behavior of the atmosphere in the ITCZ region.

I think that our knowledge was increased in two areas, (1) the general nature of the forcing of deep convection over tropical oceans, and (2) specifics about how the ITCZ works in the east Pacific.

(1) The convective forcing question is obviously a crucial one for regional and global numerical models. From my previous experience in TOGA COARE, I came to EPIC with a very "thermodynamic" view of this forcing process, which can be summarized as follows: "strong surface moist entropy fluxes ==> strong deep convection". I had to modify this view in the east Pacific only a surprisingly small amount. Ed Zipser had challenged my thermodynamic view during TOGA COARE by saying that thermodynamic control might work in the west Pacific with its weak SST gradients, but might not work so well with the extremely strong gradients characteristic of the east Pacific. Contrary to Ed's expectations, we found that thermodynamic forcing is alive and well in the east Pacific on scales of a few hundred kilometers and one day or greater. The best evidence is presented in two papers Raymond et al., 2003; Raymond et al., 2006. The first shows strong correlation between surface moist entropy fluxes and deep convection, and the second actually evaluates a PBL moist entropy budget on a 4 degree square centered on 10 N, 95 W. This budget shows that even in this region of convergence forced dynamically by the SST gradient a la Lindzen and Nigam, surface fluxes are much more important in determining the PBL entropy tendency than the convergence. Since the PBL moist entropy is central in determining CAPE and CIN, this result can be considered a "smoking gun" as to the crucial role of thermodynamics on relatively small space and time scales, even in a region of strong dynamical forcing.

One thing the 2003 paper showed was that under certain circumstances convective inhibition produced by subsidence near the PBL top could suppress deep convection even when surface fluxes are strong. A loose end is determining the origin and duration of such subsidence suppression, and hence its real importance in the larger scheme of things -- with EPIC data we could see the inversions, but not follow their evolution.

It has recently become clear that another factor in the control of deep convection is the relative humidity of the free troposphere. EPIC results show some evidence of this, but overall the EPIC region was quite moist during the period of the project, and the variability in humidity in the ITCZ region was small. Thus, EPIC was not the best context to test this factor. Perhaps the best results have come from the work of Chris Bretherton and his associates using precipitation and precipitable water data from satellites (see the paper by Back and Bretherton, 2006.) In this work they also confirmed the importance of surface heat (or moist entropy) fluxes in the control of deep convection. What is currently unclear is whether surface fluxes act only through their effect on the tropospheric humidity or whether they have an independent effect. Evidence so far tends to favor the latter view.

(2) We learned a lot about what causes the variability in the east Pacific ITCZ. This variability is coupled closely to the effect of surface entropy fluxes on deep convection; when the fluxes are weak, so is the ITCZ, and vice versa. So what causes the variations in the fluxes? In our 2006 paper we identified a mechanism which controls these fluxes in the cross-equatorial southwesterly PBL flow which feeds the ITCZ from the south. This is the effect of the free tropospheric meridional pressure gradient on the PBL. When the free tropospheric meridional gradient is negative, this gradient reinforces the PBL pressure gradient produced hydrostatically by the SST gradient. The result is a stronger SW flow which pushes farther north over warmer water, resulting in stronger fluxes in this region and strong deep convection. A positive free tropospheric pressure gradient likewise suppresses deep convection.

What causes these pressure gradients? We identified two factors, easterly waves (see the paper by Petersen et al. 2003. The influence of the latter on the east Pacific cannot be overestimated, though the mechanism of propagation from the west Pacific to the east Pacific is still a bit mysterious. Of course, a negative meridional pressure gradient is geostrophically associated with westerlies above the PBL, which explains why westerly winds favor deep convection in the region.

There are other mechanisms which can cause strong winds and fluxes over warm water in the region, principally orographic jets such as the Tehuantepec jet and the Papagayo jet. However, these tend to cause convection somewhat to the north of the region studied in EPIC.

EPIC solved one mystery which always bothered me -- what happens to the air flowing northward in the PBL when there is no deep convection? This flow is much steadier than the ITCZ convection itself, especially south of about 8 N. As Zhang et al. (2004) showed, the northward-flowing PBL air ascends only about 100 mb in this case and then returns to the south, making for a very shallow Hadley circulation under these conditions.

So, the bottom line: Have we figured out how to do cumulus parameterizations better from EPIC? As of yet, no. Of course, as we all know, cumulus parameterization is not the type of problem where you go out and make a few measurements, and poof! the whole thing is solved. Nevertheless, the results of EPIC and related observational work strongly support the primacy of thermodynamic forcing down to relatively small space and time scales, even in regions which are strongly forced by dynamical factors such as SST gradients. This will help shape at least my efforts in attacking the cumulus parameterization problem.

What about further observational issues in the east Pacific? We by no means learned everything we need to know about easterly waves and the growth of these disturbances into tropical cyclones -- the EPIC project simply was not designed to do this. In conjunction with Eric Maloney, Adam Sobel, John Molinari, and others, I am currently trying to develop an observational program to understand this intensification process in the east Pacific. So, yes, there is still a lot to be done in this region.