Guest Post by Willis Eschenbach
As I mentioned in an earlier post, I’ve started to look at the data from the TAO/TRITON buoy array in the Pacific Ocean. These are an array of moored buoys which collect hourly information on a variety of environmental variables. The results are quite interesting, because they relate directly to the subject of my previous post, “It’s Not About Feedback“.
Before I get to the buoys and what kind of diurnal cycle their temperatures undergo, let me first look at what a common extra-tropical temperature cycle looks like. I used Mathematica to get the hour-by-hour temperature records for the US Historical Climate Network (USHCN) station nearest to where I am at the moment, Santa Rosa, in the wine country north of San Francisco in California.
Figure 1. Location of the diurnal temperature records shown in Figure 2. Santa Rosa is in an interior valley at some distance from the true local maritime climate of say Bodega Bay. Thunderstorms are uncommon in that area, and summer days are often cumulus-free. The Golden Gate Bridge and a bit of San Francisco can be seen at the lower right.
So let’s look at what kind of temperatures it takes to make good wine …
Figure 2 shows how the temperature varies with the time of day around Santa Rosa.
This looks like you’d expect, or at least like I’d expected. The surface air temperature rises and falls with the sun. In addition, as the night progresses the cooling slows. It all seems very reasonable, and gives us a comparison for the surface air temperature information from the TAO/TRITON buoy array. Let me start with the location of the array of buoys:
You can see that a number of these buoys are in the “hot pool”, the area in the western South Pacific just south of the Equator. It is shown in the darkest red. Not all buoys collect the same information, but a large number of them have hourly air temperature records.
What follows are some of the preliminary results from my look at that TAO/TRITON data.
I have explained elsewhere what I have called my “thunderstorm thermostat hypothesis”. I propose that a combination of cumulus clouds and thunderstorms maintain the tropical temperature within a fairly narrow range. This is done by means of sequential thresholds which, when each is passed, marks a change into a different type of circulation.
In the morning, the sky is clear and the air is generally calm. When a critical threshold is passed, cumulus start to form. Each cloud marks the centre of a rising column of air. The surrounding air is descending to replace the rising air.
Note that this is not a negative feedback in the sense usually discussed. It is not dependent on the exact feedback from clouds and water vapor and whether it is positive or negative. Instead, it is a change between atmospheric quiescence and a defined circulation pattern containing rising air, clouds, and descending air. The net result is increasing wind, increasing evaporation, reflection of incoming energy, and surface cooling.
Particularly in the warmer regions, the temperatures continue to rise despite the emergence of the cumulus circulation regime.As the air temperature continues to rise, another threshold is passed, and a new circulation pattern emerges. This pattern is set by the thunderstorms that drive the surface air deep into the upper troposphere. Again, this is not a negative feedback, but a new form of self-organized criticality.
In the context of my hypothesis I was interested to look at the hourly air temperature data from the buoys. My first procedure as always is to look at each and every record. This is a critical step which is often omitted. Figures 4–6 show the hour-by-hour average temperature variations of the 67 buoys that collect air temperature:
Figure 4. Air temperature records from the first 24 TAO/TRITON buoys, ordered from the Western Pacific to Eastern Pacific. Each record shows the hour-by-hour average temperature over the entire record for that buoy. Records are colored from red to blue, from the warmest to the coldest. The colors are sequential, showing relative rather than absolute temperature. This group is from the western Pacific.
Here the value of examining each and every record becomes apparent. Five of the records, from the central Pacific, are strangely jagged and obviously quite unlike the others. I don’t know why these buoys are so anomalous, particularly as despite being dissimilar to the others, they are quite similar to each other. In any case, I simply took the easy path and removed them from the dataset. Figure 7 shows another view of the various records, before removing the questionable observations.
One big difference is visible immediately. The tropical oceanic records only have about a tenth of the day/night temperature swing, due to the huge thermal reservoir of the ocean and the fact that it is heated at depth. The land is warmed by the sun only at the surface, which (in addition to having no thermal mixing and lower thermal mass) leads to much greater variations in day/night temperature swings over the land.
In order to try to understand what’s going on, after removing the jagged records I converted each of the records to anomalies about their averages. Figure 8 shows the anomalies of the remaining records:
Figure 8. Temperature anomalies, all valid records. I have shown two days (repeating the average anomalies) to clarify what happens overnight. Heavy black line shows average temperatures, all records.
I kind of understand what’s going on in Figure 8. The onset of cumulus formation, shortly before noon, is quite visible. I was surprised to find that the onset of cumulus on average actually cools the air temperature. I had expected it to merely slow the warming.
The reasons for the “shoulder” where temperatures tend to level out between about 9PM and midnight is less easy to understand. I suspect that it is related to the onset of the nocturnal overturning of the upper mixed layers of the ocean, which (in my experience at least) doesn’t start until a few hours after dark. But that is conjecture about the shoulder, I welcome alternate physical explanations.
I was surprised to see that despite the large difference in local average temperature, the daily swings were quite similar in size.
I find it significant that the afternoon peaks of the cooler areas (blue) are higher than those of warmer areas. I interpret this as an indication that the afternoon peaks are knocked down by strong afternoon thunderstorm action in the warmer regions.
We can get more insight into the patterns by splitting them into the warm, medium, and cool records. First, Figure 9 shows the averages of the warmer buoys.
There are a couple points of note. The onset of the cumulus just before noon is very visible, and does more than slow the warming. It actually cools the air temperature significantly. The “shoulder” in the curves after dark are also quite evident and strong.
Next, Figure 10 shows the midrange temperature buoys, along with the average of the warmer buoys (red line) for comparison:
My interpretation is that when the temperature is not as hot, the thunderstorms are more successful in keeping down the peak afternoon warmth. The effect of the cumulus onset, however, is quite similar, as are the same evening “shoulders” in the curves.
Finally, Figure 11 shows the cooler buoys. Again I have included the average of the warmer buoy records for comparison:
The average of these cooler buoys, along with some of the individual cooler records, is starting to resemble the Santa Rosa record shown in Figure 2, losing the shoulders on both sides of the afternoon temperature peak. I interpret this as the result of weak cumulus generation and infrequent thunderstorm formation in the cooler areas.
That’s what I’ve found so far. I have no big conclusions out of all of this, other than that overall it provides clear evidence of the homeostatic mechanisms which I described in my last post. As such , it provides support for my underlying claim, that the tropical temperature is regulated by the interplay of cumulus and thunderstorm clouds.
There’s more to look at in the records. There are unanswered questions in what I show above. Why is the time of cumulus onset about the same, from the coolest to the warmest regions? Heck, I don’t know, the investigation of climate homeostasis is not far advanced, mostly the question never even gets asked, much less investigated. So I’m mostly swimming alone in the dark here, and swimming upstream against scientific orthodoxy to boot. I have also not yet split out warmer days from cooler days, to see what difference that makes in the onset time of the morning cumulus regime. Always more to do, and never enough time.
Regards to all,