Guest Post by Willis Eschenbach
I got to thinking about the effect of thunderstorms on the surface air temperature. So I figured I’d wander once more through the TAO buoy dataset. The data is available here. I swear, every time I perambulate through that data I get surprised, and I learn things, and this was no exception.
I decided to look at the relationship between sea surface temperature (SST), surface air temperature (SAT), and rainfall. Figure 1 is a graph showing all three of those variables from one of the TAO buoys.
Figure 1. Hourly data from the TAO buoy at 0°N, 156°E (in the Pacific warm pool north of the Solomon Islands). This shows both rainy and dry hours. Black line shows a 1:1 slope, where a 1° rise in SST is equalled by a 1° rise in SAT. Both color and size indicate rain amount. N = 9,067 observations
So … just what are we seeing here? And what might I learn from it?
I went into this to see how much thunderstorms affect surface air temperature and sea surface temperature. Now, I was surprised by the shape of this graph. The first thing I concluded is that we’re seeing two different regimes here. One is what is happening during the thunderstorms, and the other is what’s happening during the dry hours.
I also note the well-known SST limitation of just over 30°C. Only 0.4% of the sea surface temperature measurements in the total dataset are over 31°C (N = 62,507).
Next, it’s clear that thunderstorms are temperature limited. To investigate that, I looked at solely the hours which had measurable rain. Figure 2 shows just those records.
Figure 2. Rainy hours only of the hourly data from the TAO buoy at 0°N, 156°E (in the Pacific warm pool north of the Solomon Islands). Black line shows a 1:1 slope, where a 1° rise in SST is equalled by a 1° rise in SAT. Both the color and size indicate rain amount. N = 422 observations
As you can see, the thunderstorms have a clear minimum temperature. They are unable to persist with a temperature of much less than about 29.5°C. It’s also clear that the greater the rain, the greater the depression of the air temperature and the sea surface temperature. And as you’d expect, the depression in air temperature from the thunderstorm is larger than the depression in SST. Air temperatures drop up to maybe 3°C, from 28° or 29° down to 25° to 26°, whereas sea temperatures only drop up to about 1°C, from say 30.5° down to 29.5°C.
Finally, here are the records of only the dry times, the times without an active thunderstorm overhead. Figure 3 shows those hours when no rain is falling.
Figure 3. Dry hours only of the hourly data from the TAO buoy at 0°N, 156°E (in the Pacific warm pool north of the Solomon Islands). Black line shows a 1:1 slope, where a 1° rise in SST is equalled by a 1° rise in SAT. N = 8,645 observations
Note that both the SAT and the SST move in parallel much of the time (black line). I would say that the residual observations in the lower central area represent the colder air and colder ocean temperatures that remain after a thunderstorm when it has stopped raining.
CONCLUSIONS:
• Thunderstorms cause the coldest air temperatures in the record, well below the temperatures when there is no thunderstorm activity.
• Thunderstorms drive air temperatures down by up to 3°C or so, and sea surface temperatures down by up to 1°C or so.
• This ability to drive the surface temperature well below the normal temperature is the sign that the thunderstorms function as a governor, rather than as a feedback.
At least that’s how I interpret the graphs, YMMV of course. Dang TAO data … always turning up new stuff to puzzle my cranium …
My regards to you all,
w.
Governors are simply controllers with a set point and negative feedback. So, if the claim is that this is a governor, perhaps someone will explain what the controlled set point is [u]and[/u] give a physical reason for it. Starting point is not a set point although it might be. I would strongly advise studying basic control theory and terminology, please.
Willis: “With a governor, the system can be driven past its starting point. For example, when thermostat turns the air conditioner on in a house, it stays on until the heat drops past the starting point. To control a lagged system like a house or the climate, where things take time to heat up or cool down, such “overshoot” is necessary.
Now, let’s consider the thunderstorms. They are not a function of the size of the input (forcing). Instead they are a function of the temperature. In addition, they can drive the temperatures down below the starting point. For both of these reasons, I say they are functioning as a governor (e.g. a thermostat), rather than a simple linear feedback.”
Thanks for clarifying. I think the air-con analogy a fairly good one and probably clearer to most people.
Control systems such as air-con and thermostats have hysteresis, ie they tend to lock into one state or the other within a small control range. This primarily done to prevent jitter where the system would constantly be flashing on and off when the control variable (temperature) was at or near the desired value.
Hysteresis is obtained by adding some +ve feedback. This must be bounded by a stronger negative feedback to ensure the system is overall stable. This corresponds to the self-maintaining actions of a storm. As you have explained in previous threads, once they start they reinforce themselves even if the initial conditions are not longer present. Thus they can drop local temperatures below the initial conditions.
For this kind of system to work there needs to be an energy input that works in the opposite direction to the active state of the governor: a simple air-con won’t to anything to make the house warming in winter. In this case it will fail to regulate.
Now obviously the sun is always “on” in tropics, though it is variable due to change in position through the year and there are other heat losses such as evaporation and direct radiation to the sky that are substantial. There is also massive heat loss with persistent ocean currents evacuating heat to higher latitudes.
Now the tropics never freeze even during glacial periods of a ice age. It would be interesting to see whether there is any data showing whether it is still above 29 degC. Were tropical SST still above 25.4 degrees in 1812-1815 ?
But the basic idea of storms as localised air-con seems a sound analogy.
Now as Willis points out the control variable is temperature, it has no response to what climate scientists like to call “forcings”. If there is an additional energy input (eg GHG) or a slight drop in insolation due to volcanoes, solar variation or whatever, the system should be perfectly stable and stay within the same control range. It will just pump more or less heat out of the region as required.
I think this is one of Willis’ key points too. A simple linear negative feedback would settle at a slightly higher value if there was more input and lower if less input energy.
So far I’ve had to use the term ‘locally’ twice but the basic idea holds the road. Tropical air-con governor seems good.
“Were tropical SST still above 29 degrees in 1812-1815 ?”
Oops, that SAT , for SST that would be 25.4 degC.
[Fixed. -w.]
Willis, I don’t see this supported by the graph(s)
It’s also clear that the greater the rain, the greater the depression of the air temperature and the sea surface temperature.
This would require a before and after measurement wouldn’t it?
Otherwise interesting, and thanks again. It looks like the water likes to be 4-5°C warmer than the air for the most rainfall, if water is >29.5°C, assuming I get the unlabeled volume/color scheme at all.
Next question is where does tropical air-con dump its variable heat output required to maintain stable surface temps?
Short answer, mid troposphere. From there it gets radiated into space and convected polewards.
Viewed from a global energy budget perspective such radiation to space is a negative feedback process. Less heat into tropics causes less heat out (other direction, ie negative) , more heat in more heat out to space.
That, I think is the sense in which IPCC regard it a negative feedback. This is what leads to the expectation of a tropical troposphere “hotspot”, the “fingerprint” of AGW.
Sadly, our main suspect was apparently wearing gloves and did not leave any fingerprints 😉
The second destination is higher latitudes / polar regions. There has been a considerable change in Arctic ice cover in the last 35 years (notably in excess of gains down south).
Now that is a whole other field of study and is probably the last bastion of the global warming hypothesis. I won’t open that can of worms here but the link to tropical air-con is interesting.
So in conclusion, I think Willis is correct with his tropical air-con governor as far as tropics are concerned.
On the face of it I don’t see this as being incompatible with that mechanism being a simpler negative feedback on a global scale.
FrankK says:
March 28, 2013 at 4:18 am
I’ve always maintained (whether right or wrong) that the loss temperature is the mechanism for creating thunderstorms with heat energy converted into electrical energy (lightening) and that lightening is not due to the rather naïve “rubbing together of raindrops” etc.
For example lightening can also occur above volcanoes where the air is cooled as it rises without any rainfall. So this theory would seem at odds with Willis’s thunderstorm controlling mechanism if I have understood him correctly although lightening is a manifestation of dissipating the heat energy in my book.
———————————————————————————————–
e.g.
http://www.tumblr.com/tagged/volcano%20lightning
Again, Willis finda a correlation and calls it a conclusion. What great science. This would never pass peer review.
Willis
“This ability to drive the surface temperature well below the normal temperature is the sign that the thunderstorms function as a governor, rather than as a feedback.”
Reads well, but I’ll be damned if I can parse it. The ability of thunderstorms to cool water below the temperature where evaporation triggers the storms indicates something? Quite possibly, but I’m pretty sure I don’t really understand its importance.
Question: My impression is that the Red Sea and Persian Gulf get distinctly warmer than 32C every Summer. Assuming that’s true, have you given any thought as to why thunderstorm moderated Sea Surface Temperature limitation doesn’t seem to work there?
1) Heat from the Sun evaporates the Sea surface.
2) Vapor is forced upward.
3) Vapor cools by giving up heat to space at 5 to 10 miles up.
4) Water falls.
This is a perfect “hot spot” governor!
As we enter a new era of a “quiet Sun”, we will need less of the “hot spot” effects. Could this lead to a overall reduction in rain? Could the “quiet Sun” lead to droughts in -> +-30 – 40 Lat. around the world. Colder doesn’t necessarily imply wetter: it implies higher relative humidity easier.
Willis
This all seems really interesting – loads of data. Just a couple of questions:
1) Would it be worth separating night time from day time temperatures. I knows it’s the tropics but would you not expect two different regimes to exist. Might explain why you get a larger variation in SAT over short SST interval in your last plot.
2) The other point, does your study assume that thunderstorms form “in-situ”?
Brad in fairness he does sign off with a caveat – you should read the full article.
“thunderstorms function as a governor, rather than as a feedback.”
and from Greg goodman”:
“Viewed from a global energy budget perspective such radiation to space is a negative feedback process. Less heat into tropics causes less heat out (other direction, ie negative) , more heat in more heat out to space. That, I think is the sense in which IPCC regard it a negative feedback. This is what leads to the expectation of a tropical troposphere “hotspot”, the “fingerprint” of AGW. Sadly, our main suspect was apparently wearing gloves and did not leave any fingerprints.”
Is it possible that the number of thunderstorms and area of the thunderstorm belt can change (to dump anothroprogenic warming). The so called hot spot does not need to exist(govenor controlled), Maybe the satellites pick up lightening activity and that could act as a marker for frequency and area. The data set in any case would need to go back to the 70s, which it doesn’t.
Again, excellent science Willis, you are a genius.
@Willis
Can you recolor figure 3 as a function of (time – time of last measurable rain) with orange being the high color and a rainbow for shorter times? That should make the comma tail stand out as depression due to rain.
Brad says:
March 28, 2013 at 5:44 am
Again, Willis finda a correlation and calls it a conclusion. What great science. This would never pass peer review.
________________________
Brad, Willis stated the obvious, but used a bit of data to illustrate his point. Many of your “peer reviewed” papers get nowhere near reality, yet are used by the greedy and powerful to shape (and rob) our world.
If you think what he says is wrong- here’s your chance- prove him wrong!
Willis
Insightful.
1) Time of Day
You previously showed a repetitive cloud cycle building up in the afternoon in the tropics.
Is there a corresponding time of day dependent variation in thunderstorms?
2) Daily vs cyclonic
Is there a distinction between daily thunderstorms (small) and
cyclonic thunderstorms (large)?
While preparing an EIS for Rabaul, I noticed that cyclones are rare within +/- 7 degrees and very rare +/- 4 degrees of the equator. (Solomon Islands appears to range from ~6.4 N to 10.5 S?)
Thus there may be a very strong latitudinal variation in strong thunderstorms from occasional cyclonic activity as a function of the distance from the equator due to the Hadley cells.
Happy hunting.
I often wonder about the fact that evaporating water is an endothermic process, which is unusual in nature, and how this affects the climate in general. Of course that is why we sweat, the process cools our bodies. If the atmosphere holds more water at higher temperatures, then how is it affected by the increased endothermic effect of evaporation? How is this factored in to climate models? And where do thunderstorms, which are endothermic, fit in?
‘Blood, sweat and tears’ in human endeavour could translate to ‘solar, sweat, and clouds’ in the ‘climate’.
Willis, perhaps you have found the signature of rain. I note the lower hook (Fig 3) persists in the non-rain hours. Based on your hypothesis, wouldn’t that suggest sometimes there is rain during those non-rain hours?
Willis Eschenbach says: “With simple linear negative feedback, a system can never be driven past its starting point. To do so, the feedback factor would have to be less than -1 … and that is unstable.”
Well it’s a negative feedback that ain’t ‘simple’! When you introduce a LAG in a negative feedback it can indeed drive past it’s starting point. Just the kinetics of the convection alone represent a lag in the feedback by intuitive examination – suddenly stop the heat and moisture differences that initiated the convection in the first place and it does not suddenly stop anymore than rain drops stop on their way down.
Brad says: “Again, Willis finda a correlation and calls it a conclusion. What great science. This would never pass peer review.”
No, Willis is making conclusions on his observations of the data. Such is the food of new theories but, in this case, REAL food from REALITY – not ‘junk food’ spewed from incomplete and maybe even intentionally biased climate computer models.
I’m not surprised at what you are observing. The rates of the processes of evaporation and condensation are controlling the rates of energy transfer between the ocean and the atmosphere. Also, I think these rates are controlling the global uniformity of atmospheric concentrations of CO2. CO2 is not controlling these rates.
Steve Keohane
This would require a before and after measurement wouldn’t it?
In truth you’d need wind spead as well, even if slight, it can reduce vapour pressure thus increase evaporation which in turn lowers atmospheric temperatures, furthermore, it reduces the insulating properties of warmer air at the surface all of which affects SST. But at some point you have to look at the data, and without all this additional analyses, you must draw what seem like reasonable conclusions.
The case for thunderstorms (if forming in-situ) occurring within a small SST range is borne out by the data – it’s not conclusive proof but it is suggested and seems a reasonable conclusion. The other points about the effects of thunderstorms are also there in the data. If you have other data or studies to contradict these then present them.
What you are asking for is a comprehensive study that would require a dedicated research team. Beyond the scope of what is being attempted here.
Willis:
The TS is acting as a “super radiator” putting a huge burst of IR out into space. We know the number of TS’s, we know relative size! Can we start making a estimate of energy dissipation away form good old Earth? Certainly the cold water in the oceans is part of that moderating effect…making the problem “difficult” as you have a bifurcated balance. I.e., one part that is VERY immediate, and one part that has a “lag time”. (Could be days, could be years).
Just some toughts.
Max
Brad:
Your post at March 28, 2013 at 5:44 am says in total
I strongly suggest that you try to read the article by Willis and his subsequent comments in this thread because the kindest understanding of your post is that you have not read what Willis wrote.
Willis has a hypothesis which he has repeatedly explained, and he continues to investigate available data to determine whether it falsifies his hypothesis. His above article is yet another of his investigations of available data. To date he has failed to find any data which falsifies his hypothesis and has determined that much data supports it. His above article reports yet more supporting information.
Please note that Willis does not merely throw out his studies but interacts with proper disputations. For example, in this thread he yet again repeatedly attempts to explain the difference between what he calls a ‘governor’ and simple linear feedback for those who fail to understand: i.e. linear feedback is a response to any change such as to reduce the change but a ‘governor’ actively opposes any change such as to overwhelm it.
It seems you fail to understand that WUWT is science blog and not some propaganda blog such SkS or RC. Were you able to cite information which refutes his hypothesis or if you were able to refute his interpretation(s) of the data then I am sure Willis would be grateful, but comments such as yours are only a waste of space in the thread.
I have bothered to write this response to your post in hope that this response will dissuade other anonymous troll from wasting space in the thread as you have.
Richard
Willis, there is another factor involved in the charts and the change in temperatures. Have you ever noticed when you boil water, there is what appears to be a first boil or bubbling before the water actually boils. These are the gasses dissolved in the water escaping. All gasses in water act as a liquid until their triple point is reached. You can demonstrate this by a simple experiment. Take a beaker of water, put a thermometer in it and place in the sunlight. As the temperature reaches approximately 29 C, bubbles will form on the sides of the beaker. They will continue to form until the temperature reaches approximately 30.5 C. Carbon dioxide is one of those gases, It’s triple point where it can no longer remain a liquid is 30.3 C. Your charts show this transition point. A chemistry lab test for carbon dioxide is to bubble it through a solution of calcium hydroxide. The solution will turn cloudy as calcium carbonate is formed. If you will check your chemistry book, you will find that this reaction is endothermic and removes a significant amount of heat from the water.
Willis,
It has been claimed that less ice coverage in the Arctic causes both colder temperatures (by blocking, shifting of the jetstream) and more snow (by increased humidty leading to more precipitation). I reviewed the Arctic sea cover data for each of the sub-regions and saw that the Greenland sea area did have an historical lower ice coverage during the main winter time period, though the rest did not (again, regional not global). The precipitation, though …
Perhaps you have the data, perhaps not. Your correlation here is interesting. I wonder:
what this data would say about the Arctic seas:
1) what is the correlation between SST, open water and air temperatures, and
2) what is the correlation between these three and precipitation, in particular snow?