Guest essay by Charles Samuels
A recent article in the WUWT, The Daily Albedo Cycle, by Willis Eschenbach caught my eye when the author postulated that daily thunderstorm development was active in controlling temperatures in the tropics.
The article shows that during the day as certain temperatures are reached, thunderstorms and cumulus clouds develop which increases the albedo and reflects sunlight back into space, thereby cooling the atmosphere and ocean. Looking for a way to extend Eschenbach’s results it seemed reasonable that data on precipitable water, which is the total amount of water in a column of air, would be an indirect measurement of the amount of cumulus and cumulonimbus clouds in the tropics.
Precipitable Water (PW) data is available from the NOAA Earth System Research Laboratory from 1949 to 2014. The data is world-wide at 2.5 degree increments of latitude and longitude. A program was written to extract the yearly average PW for selected latitudes for the period of record. A graph of the results show that in the tropics from 5° N to 5° S the amount of PW decreased from about 1959 to 1997. Since the precipitable water in this area is primarily cumulus clouds and thunderstorms, it follows that there has been a decrease in such clouds during that period.
Fig 1 Precipitable Water and Sea Level Pressure for latitude 5.0 degrees north.
There was no such reduction in PW north of 5N or south of 5S. There are probably many reasons that could account for such a reduction but a look at Sea Level Pressure (SLP) for the same period shows slightly rising SLP while PW is falling. Higher pressure over the tropics would tend to provide more stable air that is not conducive to thunderstorm development.
Note that in Fig 1 the rising curve for Seal Level Pressure coincides with downward curve for PW until about 1997 when both curves flatten out. The correlation between the two is -0.759. It is interesting, but perhaps coincidental, that the period from 1997 is the same period as the current hiatus in global warming.
Charles Samuels June 16, 2015 at 11:33 am
My friend, thank you for your perseverance. However, I fear that what you have done is the equivalent of asking your barber if you need a haircut …
For starters, she says what you have used is “model generated output”, but not “model data” … say whaaaat? That’s just modeler NewSpeak. What is “model data”, and how is it not “model generated output”? On my planet “model data” is an oxymoron.
And yes, she agreed with what I said. The computer takes the observations and gives its best guess at all the thousands of times and places where there are no observations.
Now, such computer model output is interesting. But one big problem is, depending on the amount of infilling on your particular chosen variable, you could be looking at anything from 100% observations to 100% computer guesses, and there’s no way to tell the difference. So you simply can’t just grab some variable and assume it means something.
Another problem is the aforementioned linearity of the computer models, which don’t do edges well at all.
So you can’t just grab a reanalysis dataset and expect all of its dozens of datasets to correspond to the real world equally well … or well at all.
Now, that is very interesting. It compares the NCEP-NCAR reanalysis “data’ to a dataset I’d not heard of, the NVAR dataset. This appears to be a merging of various actual measurements of precipitable water. In my opinion, such a merging of actual data is likely to be closer to reality than the output of a climate model. I’ll need to take a look at that … so many datasets, so little time …
All the best, thanks for the response,
w.