Paul_K: Thanks for the reply. I hope an answer to some model problems is already available in the output from ensembles of models. If we looked, could we find several parameter sets that perform better at reproducing the seasonal cycle or some other set of observables (TCR from energy balance models)? I gather from the work of Stainforth et al that a large number of parameter sets perform about as well (or badly, if you prefer) at reproducing today’s climate. Tuning parameters one by one appears to have left us in a featureless wilderness of minor local optima with no hope of finding a global optimum – except systematic exploration of all parameter space motivated by confronting the worst failures of modeling. Limitations associated with the size of grid cells could frustrate such a search, but even that problem would be more tractable if simulations could be run for several years rather than several decades or centuries.

]]>Pekka,

Your comment prompted me to re-read some comments I made on the same subject 5 years ago. It was embarrassing for me in that, on that thread, I went all round the houses to prove something which I could have done far more simply and more generally. What I wanted to show was that, under the same billiard ball Earth assumption as made by Ramanathan, the derivative w.r.t surface temp of G as defined by Ramanathan relates to the TOTAL FEEDBACK of the system, and not just the LW feedbacks. I will include a simple proof below. Given this, I still find Ramanathan’s conclusion to be unsafe, when he asserts “The analysis confirms…” The furthest he should have gone was to state that the overall feedback (including the seasonal effects of heat transfer and SW absorption change) appeared to be positive relative to Planck – under some non-trivial assumptions.

Simple proof for billiard-ball Earth follows (this assumes no error introduced by spatio-temporal averaging).

Transient net flux behaviour at TOA can be textually stated as:- change in net flux from steady-state at t= 0 is equal to the change in input flux minus the change in outgoing flux:-

ΔN(t) = ΔI(t) – ΔO(t)

.The only outgoing flux is LW. Making the common approximation that this can be expressed as a linear function of surface temperature (as does Ramanathan), we have

ΔO(t) = OLR(t) – OLR(0) = λ*ΔTs = λ*(Ts(t) – Ts(0)) (1)

where λ is the TOTAL FEEDBACK term.

differentiating (1) w.r.t. Ts, we obtain:-

d(OLR)/dTs = λ (2)

Ramanathan defines G for all-sky from the relationship:-

OLR = σTs^4 – G (3)

Differentiating (3) we obtain:-

d(OLR)/dTs = 4σTs^3 – dG /dTs (4)

Substitute (2) and we obtain:-

dG/dTs = 4σTs^3 – λ (5)

Equations (2) and (5) both indicate that the response that Ramanathan is testing must include all feedbacks. This includes the effect on the vertical temperature profile of the variation in shortwave heating via atmospheric absorption. I think it is a stretch to believe that the water vapour feedback can be isolated from these data, even though Ramanathan’s conclusion of positive feedback may turn out to be correct.

]]>Paul, thanks for your patience with comments in moderation.

Actually it is not me putting them there, it is WordPress. I use hosted WordPress and wonderful though it is, it decides for reasons I don’t understand that a comment is suspect.

All explained in Comments & Moderation

]]>]]>However, our results do not necessarily confirm the positive feedback resulting from the fixed relative humidity models for global warming, for the present results are based on annual cycle. We need additional tests with decadal time-scale data for a rigorous test. Nevertheless, the analysis confirms that water vapor has a positive feedback effect for global-scale changes on seasonal to inter-annual time scales.

Frank,

I have a comment in moderation which addresses in more detail the problem of comparing the seasonal cycle with multiannual warming. (It is in moderation, I think, because I didn’t understand SoD’s previous instruction concerning brackets. I do now. I hope that SoD will accept my apology for my repeat stupidity here, and let the comment out of moderation.)

I agree with much of what you say. In particular, I think it would be a truly valuable exercise to try to produce a model which did fully match the seasonal variation in terms of heat fluxes, radiative fluxes, atmospheric momentum and temperature. I suspect that this would reveal some shortfalls in the physics we are currently assuming in AOGCMs. However, I also think that this is a mammoth undertaking. At present, none of the GCMs have the capability to match heat and momentum fluxes at regional level. Nor can they reproduce well the amplitude of temperature variation by latitude. it would require the building of a new type of model or major modifications to an existing model.

I also believe however that if the objective is solely to gain insight into radiative feedbacks from the seasonal data, then there is a halfway house which needs radiative code, but which avoids the need for phenomenological modelling. This requires characterisation from the observational data (i.e. prescription), for a specified number of zones, of the data required to calculate and match the local OLR. Heat fluxes (including atmospheric meridonial heat fluxes) can be prescribed rather than calculated. This is less likely to reveal any serious problems with the physics, but once a match had been achieved, LW feedbacks regionally and in aggregate could be assessed by perturbation from the local mean behaviour. This is still a large task, but one which would be somewhat easier than, and a natural precursor to, building and matching a new phenomelogical model.

]]>SoD,

Thank you for your post of July 17 9:43pm. It is always good when someone plays back what they have understood from a comment. You have grasped the main first point I was trying to make about the spatio-temporal averaging problem as it applies to interpretation of seasonal variation.

In addition to this averaging problem there are other significant confounding factors.

The seasonal variation in radiative flux and temperature are not solely controlled by the wave of insolation that passes over the planet during its annual orbit. There are at least three other known “Mexican waves” to be considered:-

a. An orbitally-driven annual and semi-annual momentum flux which causes oscillation in atmospheric circulation; this adds and subtracts energy as well as redistributing sensible heat in the mid- and upper- troposphere.

b. An albedo wave driven by recurrent seasonal changes in cloud, snow and ice distribution.

c. The cycle of shortwave absorption into the atmosphere relative to surface absorption. SW absorption is in phase with solar insolation, and on average over the annual cycle the atmosphere is more transparent than absorbing. Consequently, on average, the atmosphere is more heated by surface fluxes from below than by atmospheric absorption from above. From year to year, there is little net change in system energy, barring external forcing of the system. So it may be a reasonable assumption that the radiative fluxes are in long-term balance at TOA, and that total energy fluxes (including convection) are in balance at the surface ON AVERAGE. However, during the annual cycle, this assumption is completely invalid. Both locally and globally, there is a net gain and a net loss of system energy during the seasonal cycle. There is therefore no physical requirement that the SW absorption at the surface is balanced by upward energy fluxes from the surface; this is especially true over the oceans because of their heat capacity. The OLR at any given locale is therefore not driven just by the surface temperature at that locale, but is also driven by the variation in atmospheric heating over the seasonal cycle – which modifies the vertical temperature profile.

Each of these three waves changes **the local relationship ** between surface temperature and OLR emission at TOA. The above seasonal changes are not effects which can be plausibly neglected. Yet, none of these three effects are present when we consider the question of radiative flux changes or feedbacks associated with interannual or decadal warming. The key question then becomes:- *To what extent does the relationship between OLR and surface temperature evident during the annual cycle represent the relationship we might expect to see during a multiannual warming trend?* The problem is that we do not know the answer to this question, but the true answer may be “very little or not at all”.

If I understand correctly, Figure 3 from Dessler, J Climate (2013) provides evidence that feedbacks vary dramatically with latitude (and IMO probably terrain: ocean, coastal, continental).

Shindell’s paper attempting to discredit energy balance models asserts that climate models show that aerosol forcing, which is regional, is more effective at changing Ts than the global forcing from GHGs.

]]>OK, let’s continue this discussion here:

https://scienceofdoom.com/2011/09/22/measuring-climate-sensitivity-part-one/

]]>SoD,

*“I can’t tell if you are still totally confused or still trying to evangelize everyone to your point of view – whatever that point of view is – and having responded to 100s of your comments in the past I definitely don’t want you to explain (again).”*

Well, it’s certainly possible I could still be confused, but I may not be either. I can’t remember what threads some of these things were (I guess) discussed on. Maybe a post of yours that deals entirely with the notion of so-called ‘no-feedback’ would be worthwhile. Or if there already is one, perhaps that would be the best place to have this discussion, as I agree it’s really largely off the topic of this thread.

]]>