Guest Post by Willis Eschenbach
I got to thinking again about the question of evaporation and rainfall. I wrote about it here a few years ago. Short version—when the earth’s surface gets warmer, we get more evaporation and thus more rainfall. Since what comes down must go up, we can use the Tropical Rainfall Measuring Mission (TRMM) satellite rainfall data to calculate the corresponding rainfall-related evaporation.
From that TRMM data, we can also calculate how much the evaporation changes with additional warming. Figure 1 shows the trends in evaporative cooling with respect to temperature, in units of W/m2 of additional evaporative cooling per degree of additional warming
Figure 1. Amount of additional evaporative cooling per additional degree of temperature. Red areas have the greatest rainfall and thus the greatest evaporative cooling. The area of greatest cooling is the Inter-Tropical Convergence Zone (ITCZ) just above the Equator. Note that this includes three additional years of CERES and TRMM data compared to my earlier analysis.
As you can see above, the change in evaporative cooling per degree C ranges from a drop in evaporation of -70 watts per square metre per degree of surface warming (W/m2 per °C), all the way up to an increase in cooling of well over one hundred W/m2 per °C of surface warming.
Now, I’d gotten that far in my previous analysis, but I was stymied by the incomplete coverage. At the time I said:
As noted above, the TRMM data covers about two-thirds of the surface area of the Earth. From appearances, unlike in the tropics, the correlation of evaporation and temperature is negative in the unsurveyed areas of both the northern and southern extratropics. The grey line at about 30°N/S shows where the relationship goes negative. This is no surprise. In the extratropics, rain is associated with cold fronts instead of being associated with thermally driven tropical thunderstorms. As a result, although the overall average change in cooling shown in Fig. 6 is 11.7 W/m2 per degree of warming, I suspect this be largely offset once we have precipitation data for the currently unsurveyed areas.
So my quick guess at the time was that overall the value might be around zero. However, this question continued to bother me. So I started thinking about how I could estimate the trends in the areas not covered by the TRMM data. I began by looking at the average evaporative cooling by degree of latitude. Figure 2 shows that result.
Figure 2. Latitudinal averages of evaporative cooling (W/m2 per °C), in 1° wide latitude bands.
This was quite encouraging. I had previously assumed that as we went towards the poles, the trend would continue to go more negative. But both in the northern hemisphere (positive latitude) and the southern hemisphere (negative latitude), the trend is heading back towards zero as we go towards the poles. This would indicate that the values nearer to the poles might be around zero.
Next, I took a different look at the data. Figure 3 shows a scatterplot of the evaporative cooling trends versus the average surface temperature. It shows on a gridcell-by-gridcell basis the relationship between the evaporative cooling trend for that gridcell versus the long-term temperature of that gridcell. There are 28,800 gridcells shown in Figure 3.
Figure 3. Scatterplot of the evaporative cooling trends versus the average surface temperatures, 40° North to 40° South.
Now, this is quite revealing. It shows that at the cold end of the temperature scale, the evaporative cooling trend is quite small. Where the surface temperature is 0°C to 5°C, the average trend is -0.05 W/m2 per °C. For a surface temperature of 5°C to 10°C, the average trend is -0.6 W/m2 per °C.
SO … I think we can reasonably estimate that the average trend in the unmeasured areas of the globe shown in Figure 1 is on the order of -0.5°C. Recalling that the area from 40°N to 40°S is about 2/3 of the globe, and that the average for that area is 10.7 W/m2 per °C, that means that the global average is (1/3) * -0.5 + (2/3) * 10.7 = 7.0 W/m2 per °C.
Now, if this estimate is high and the actual value in the white unsurveyed areas in Figure 1 is say -3.0 W/m2 per °C, that would give a value of 6.1 W/m2 per °C. And if the estimate is low and the actual average value in the unsurveyed areas is 3.0 W/m2 per °C, that gives us an average of 8.1 W/m2 per °C.
So it appears that a likely global value for the trend in the evaporative cooling for each °C of additional warming is on the order of 7 ± 1 W/m2 per °C.
Next, note that in the tropics the evaporative cooling trend goes up rapidly with temperature. The average in the tropics is 16.7 W/m2 per °C, with some areas cooling at over 100 W/m2 per °C.. Since the tropics is the area where the most energy enters the system, this provides a strong mechanism to prevent overheating.
Next, it is important to understand that this strong cooling effect is not applied at random. Instead, in general it occurs preferentially where the surface is warmer than the surrounding areas. As a result of the targeted application of the cooling, it has a larger effect than if it were applied willy-nilly.
Finally, in IPCC terms this would be classed as a feedback. That is to say, if the surface is warmed (by increased forcing or any other cause), when it warms the increased number of thermally-driven thunderstorms act to increase the evaporation and cool the surface back down. However … as far as I can find it is not included in the IPCC analysis of all feedbacks.
Regards to all,
w.
My Usual: If you comment please QUOTE THE EXACT WORDS YOU ARE DISCUSSING, so we can all understand your subject.
Further Reading On Evaporative Feedback:
Thanks for this thread. It makes it very clear that the science is in no way settled. It identifies significant known unknowns.
Willis: Above you discuss evaporative cooling as a feedback. Changes in evaporative cooling are critical to surface energy balance. If evaporation increases at a C-C rate of 7%/K, that is 5.6 W/m2/K.
If ECS is high, say 3.7 K/doubling, then the change in the TOA flux with rising surface temperature must be 1 W/m2/K ( 3.7 W/m2/doubling divided by 3.7 K/doubling ). So, if ECS is high, only about 1 W/m2 more heat escapes to space or is reflected back to space for each degC the planet warms.
Obviously, we can’t have evaporative cooling sending something like 5.6 W/m2/K up from the surface indefinitely and only 1 W/m2/K escape to space. So feedback in evaporative cooling is as critical to ECS as traditional feedbacks. This is important work.
Frank, please note that you are another person than me! 🙂
It seems to me you could partially answer Nick Stoke’s comment by comparing the low altitude temperature with the high altitude temperature for those same gridcells that are showing massive evaporative cooling and see how they compare to the low vs. high for low evaporative cooling.
Thanks, Peter. Nick’s basic point is that the whole system is not cooled when water evaporates at the surface. On the simplest level, this is true. However, there are other issues at play.
The main issue is … thunderstorms. Much of the tropical rain that we see in Figure 1 comes from thunderstorms. The intensity of tropical rainfall is a result of the increased evaporation right at the base of the thunderstorm. This increased evaporation is driven by the winds at the base of the storm. These winds increase the evaporation up to five-fold or more.
Now, what happens to the energy removed from the surface at the base of the thunderstorm? As Nick points out, it is released when the water condenses. However, the story doesn’t stop there. The released heat warms the air at the cloud base. This warmed air then rises up through the cumulus tower, and finally exits at the top of the tower.
So let’s look at the path taken by the heat. It starts out as thermal energy at the surface. It is converted to latent heat by the evaporation of the water. It is converted back to thermal energy by condensation at the base of the cumulus tower. It then rises through the center of that tower to exit high in the troposphere.
There are two things to note about this thermal journey.
First, the energy has moved from the surface to the upper part of the troposphere. Up there it is much freer to radiate to space. This is because there is less CO2 and much, much less water vapour than at the surface.
Second, at no point during the journey is the energy exposed to the greenhouse gases. From the surface to the base of the cloud it is transported physically as latent heat. And during its travel from the base of the cloud to the upper troposphere, the warmed air is shrouded by the cumulus tower.
At no point is the energy free to radiate to the open atmosphere.
This means that the thunderstorm is able to pipe the heat from the surface to the upper troposphere without exposing it to most of the poorly-named “greenhouse effect”.
And by avoiding the “greenhouse effect” in the trip from the surface to the upper atmosphere, it allows for much greater energy loss to space than we would see otherwise.
To make a long story short, Nick is right in the short run (surface to the lifting condensation level), but he’s left out the rest of the vertical voyage.
Best to all,
w.
I question the premise: “Short version—when the earth’s surface gets warmer, we get more evaporation and thus more rainfall. Since what comes down must go up, we can use the Tropical Rainfall Measuring Mission (TRMM) satellite rainfall data to calculate the corresponding rainfall-related evaporation.”
1) There is no need for the moisture in the atmosphere to remain constant. What must come up, does not need to come down. In other words, the amount of moisture in the air does not have to be constant.
2) Just like evaporation is a cooling process, condensation is a warming process. If you assume that there is no net gain in air moisture, then you are assuming that there is no net evaporative cooling. Rainfall seems like a poor measure of evaporation.
3) It is curious to me how close that picture looks like the El Nino pattern. I wonder if what you are measuring is only an approximation of the ENSO cycle.
Willis
“This was quite encouraging. I had previously assumed that as we went towards the poles, the trend would continue to go more negative. But both in the northern hemisphere (positive latitude) and the southern hemisphere (negative latitude), the trend is heading back towards zero as we go towards the poles. This would indicate that the values nearer to the poles might be around zero”
“SO … I think we can reasonably estimate that the average trend in the unmeasured areas of the globe shown in Figure 1 is on the order of -0.5°C. Recalling that the area from 40°N to 40°S is about 2/3 of the globe, and that the average for that area is 10.7 W/m2 per °C, that means that the global average is (1/3) * -0.5 + (2/3) * 10.7 = 7.0 W/m2 per °C.”
I was looking at figure 2 and if the curve was a Mesokurtic curve I could go along with your estimate of 30 to 90 degrees but I see a Polymonial curve and this fits the global air circulation at 30 60 90 degrees and my guess based looking at figure 2 is 15 watt m 2 at 60 degrees.
0 & 60 degrees is low pressure.
30 & 90 degrees is high pressure and deserts.
0 & 45 degrees and wet forests.
Interesting is the peak of the curve is not at the equator and the step in the curve? Any one have thoughts on them?
WE: “Nick, you are correct … but the issue at hand is the temperature at the surface. Your claim is as wrong as saying that sweating doesn’t work to cool us down, because the water evaporated from our skin will condense and release its heat somewhere else …”
Surely this comparison is faulty. Sweating cools us down precisely because when the evaporated sweat condenses it condenses elsewhere on the surface, or later, or over a wider area of the surface. If we look at the total system there is no “cooling” of the surface by evaporation. There is an unbroken cycle of cooling by evaporation and corresponding warming by condensation. The cooling is local and temporary and does not exist for the whole system. (You can say that some of the energy from evaporation leaks into space but that’s a different issue from sweating, i.e. local versus whole system.)
I should have said “corresponding warming by condensation and precipitation” of course. When the rainfall hits the ground, it warms it.
You are missing the Earth’s heat engine.
Increase convection and you accelerate the motion of energy to the LOCATION where heat escapes earth system via radiation.
Take your car merrily driving along at a constant. Speed and T. Now punch it, sending considerable energy to your car. If you do not accelerate the hydrological system in your car, which is converting conducted heat to the place where it radiates, curiously called the radiator, the car will overheat. Fortunaly the motion of water and air from the fan in your car quickens it’s pace, thus shortening the residence time of heat in the system and your car stays the same T despite greater input because the response of water and air has REDUCED THE RESIDENCE time of heat in your car’s engine.
The earth does the same thing. The Earth’s radiator is up high in the atmosphere. Accelerating the hydrological cycle expedites the movement of energy in air and water to Earth’s radiator, thus shortening the residence time of energy in Earth’s system, NOT just at the surface.
This is, without questuon, a negative feedback to any increase in the energy of Earth’s system. I appreciate Willis’s attempt to quantify it.
Willis, your figure 2 seems to have invented evaporative warming – i.e. cooling less than zero. Surely that can never be the case?
Also where the rain falls and where the rain evaporated from are two very different places once non-convective weather is concerned, or there would never be much rain in Kansas. The frontal weather and orographic precipitation are normally caused by lifting of humid air until the water vapor condenses out and precipitates. The humidity may have come from hundreds of miles away so the frontal rain in Kansas has no relationship to surface heating in Kansas nor necessarily any relationship with Kansas latitude. Consider a pulse of humid air from the Gulf of Mexico carried North in a frontal system dropping rain, I think that there are relationships there but they are a lot more complex once you move away from convective weather over the oceans.
Willis, the great thing about your analyses is that they would make, together, a fine climate science for “Dummies”. I say this with the greatest respect. All your readers scientists, engineers, poets, social scientists (whether they wish to or not) can understand your simple exposition and logic. You have a gift. I have no doubt you would have a best seller if you produced a “CliSci by Colours” “(I thought” by numbers” in allusion to the popular “paint by… numbers” but that may scare many off). It would have the added bonus of revealing the wondrous stuff of satellite instruments and the beauty of the scientific method- “… Hey, I can understand this stuff…”
Things left out. Thunderstorms are violent especially “towering” types. Are they well researched?
Some of the precipitation may be/usually is hail which may or may not make it to the surface before melting. Lots of energy transfers.
Joe and Joe at Weather Bell talk about the “Texas perma-drouth” and how poorly the models do forecasting long range weather more less climate.
Willis, have you ever considered subtracting these two graphs? The predicted and observed tropospheric hotspot. Looking at the equator, it seems plain that there is tropospheric cooling in the observed data, while the climate models predict warming. I see this as strong evidence that water feedback is negative.
A similar effect is visible comparing northern and southern hemispheres, with cooling in the south consistent with increased oceans, and warming in the north consistent with increased land. The graphs are from David Evans’s 2008 work.
Willis, have you ever considered subtracting these two graph’s? The one of the left is model projection based on the assumption of positive water feedback. The one of the right is observed. It would seem to me that the difference between these could be used to calculate actual water feedback. Looking at the observed warming on the equator and southern hemisphere as compared to the northern hemisphere, it does appear consistent with your graph above.
http://jonova.s3.amazonaws.com/graphs/hot-spot/hot-spot-model-predicted.gif