Kosaka and Xie, take 2

This is my take 2. See here for my incautious take 1. Take 2 is not as interesting as take 1 – I no longer have an overall theme, and I don’t feel inclined to contradict the take-home message. That reduces me to quibbling and a slight feeling of unease, though that may quite possibly be because I now feel biased against this paper for giving me a hard time.

So, take their “We present a novel method of uncovering mechanisms for global temperature change by prescribing, in addition to radiative forcing, the observed history of sea surface temperature over the central to eastern tropical Pacific in a climate model”. But this isn’t desperately novel, nor is that quite accurate – what they actually do is restore SSTs towards the observed pattern in this region, by modifying the surface heat fluxes (and if you compare figure 2 and b closely, you’ll see that the restoring isn’t perfect; in that 2b, within the inner box, clearly isn’t the same as 2a in the same region. In fact its rather more different than you’d expect, which is odd. Ditto 3 a, b). Which isn’t new; here for example is a random example from 2006. In fact, later on they say The POGA experimental design has been used to study the global teleconnections of the interannual El Nino/Southern Oscillation (ENSO)11,12. Here we present a novel application of POGA… so I think they’ve mis-spoken in their abstract: the method isn’t new, only the application. You’d have thought that Nature-quality reviewing would have caught that. But I’m quibbling.

Although the surface temperature prescription is limited to only 8.2% of the global surface, our model reproduces the annual-mean global temperature remarkably well with correlation coefficient r = 0.97 for 1970–2012

k+x This is the bit I reacted somewhat badly to in my first go, claiming that it wasn’t surprising because the ENSO region is so important. Just to make sure that Captain Cockup doesn’t come to visit again, here are their experiments:

* (HIST) is forced with observed atmospheric composition changes and the solar cycle.
* (POGA-H) Pacific Ocean–Global Atmosphere (POGA) experiments, SST anomalies in the equatorial eastern Pacific are forced to the observed evolution, and the radiative forcing is identical to HIST
* (POGA-C) is like POGA-H, except radiative forcing is fixed at the 1990 value. In both cases, outside the equatorial eastern Pacific, the atmosphere and ocean are fully coupled and free to evolve.

So if you look at the lower figure, and compare the wiggles to the upper, they match pretty well – especially if you ignore the volcano years. Which does indeed suggest that ENSO is driving much of the interannual variability, but now I’m obliged to admit that comes with the trend coming from the radiative forcing (as, in retrospect, you’d expect). It also appears to imply that the volcanoes mainly affected extratropical temperatures for some reason, but that’s another matter.

A comment (which isn’t original to me): even attributing the change to ENSO doesn’t tell you if its forced or not: whilst ENSO is a natural mode of the system, its perfectly possible for Anthro forcing to act not to change the long-term-mean-state, but to push it further into a warmer regime; there has long been speculation that warming could manifest as “more El Nino and less La Nina”. OTOH, in this case we appear to be seeing more La Nina, and its somewhat hard to see how that could be forced; K+X address this, a little bit, in their “Whether the La-Nin˜a-like decadal trend is internal or forced is still unclear…” para.

Errm well there you have it: it looks OK to me.

9 thoughts on “Kosaka and Xie, take 2”

  1. Hm.

    The formula they use is

    F = (1-alpha) F* + alpha (cD/tau) (T’ – T’*),

    where T’ and T’* are observed and model-produced SST respectively, and F* is the model-produced heat flux through sea level. F is the new, forced flux. Inside the Eastern Pacific area alpha = 1, elsewhere = 0.

    A problem I see: if T’* approaches T’, then inside the area F will approach, not F*, but zero. Is that right? Shouldn’t we rather have

    F = F* + alpha (cD/tau) (T’ – T’*)?

    I must be missing something.


  2. I don’t think the paper touches on that Paul. But wasn’t the explanation more along the lines of ocean total heat content increasing apace irrespective of what the much smaller atmospheric heat content may have been doing?


  3. Martin Vermeer:

    But wasn’t the explanation more along the lines of ocean total heat content increasing apace irrespective of what the much smaller atmospheric heat content may have been doing?

    Can that be right? AIUI, the amount of excess heat from the retention of outgoing longwave radiation in the atmosphere should remain more or less constant. Its apportionment between the various sinks is what changes year-to-year. More goes into the oceans under la Nina conditions, less under el Nino. Somebody will correct me if I’m wrong, hopefully.


  4. Mal Adapted,

    I think that’s effectively what Martin was saying too, but with the distinction that there is large asymmetry in the magnitude of different reservoirs (e.g. ocean large, atmosphere small) so for total energy content what happens in the atmosphere is almost negligable.

    For me it’s not quite that simple because cloud and water vapour feedbacks, which are important factors for the magnitude of surface warming and energy imbalance, mostly respond to surface temperature change. If energy transport is occurring in such a way that limits warming of the surface we might expect a small temporary slowing of total energy accumulation compared to a “normal” period.

    For el Nino/la Nina impacts the Balmaseda et al. 2013 plot is interesting because it shows a spike in OHC with the 1997/98 el Nino and then a dip with the 1998/99/2000 la Nina. This suggests that positive cloud feedbacks to surface warming/cooling during these large ENSO events are overwhelming the differences in energy collection potential between the two states, at least in this reanalysis product. Those features aren’t obvious looking at statistical global OHC products but I’m not sure if that’s mostly because of sampling deficiencies or errors in the reanalysis model.


  5. Paul S, I think that is at least partly right. One cannot assume that the radiation imbalance at TOA is completely independent from the ENSO cycle. I doubt though that any such effect would be large (it isn’t in the Balmaseda graph taken at face value).


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s