Another week, another team of scientists pointing out that climate models predict too much warming. Last week it was because modelers assumed too much aerosol cooling. This week it’s because they assume clouds do stuff that real clouds don’t do. Like Joni Mitchell, modelers have looked at clouds from both sides now, but mainly from one side: that as the atmosphere warms they will act like a thickening blanket and trap more heat, a positive feedback effect amplifying the surface warming. It was a convenient assumption back in the early days when there wasn’t much data to go on. But now with more data available it’s been discovered that when cloud tops warm they get shinier and reflect away sunlight, which is a negative feedback effect. So, as the lead author told the press, “Our work shows that the increase in climate sensitivity from the last generation of climate models should be taken with a huge grain of salt.”
To put the matter more formally, the authors of the study noted that the tops of clouds, even in the tropics, are in a very cold part of the atmosphere where they get shrouded in ice crystals. The ice crystals act as a thermal layer and keep heat in, which is indeed a positive feedback on surface warming. And models have assumed cloud tops get icier as the amount of water vapor increases in a warming atmosphere. But as the air warms something else happens: The cloud tops become liquid water, and the droplets act as mirrors, reflecting sunlight away and cooling the air underneath. Models simulate some of that conversion, but they also assume the liquid water rains out quickly so the cooling effect is short-lived. Whereas newly-available data has shown that the cloud tops in liquid phase hang around longer than models allow for, so their real cooling effect is stronger. When the team programmed that effect into a climate model, the result was a strong new negative feedback effect, especially in the Southern Hemisphere.
Having shown all these things, the modelers then point out that there are lots of unknowns in the climate system, and while fixing this one problem in the models would lead to lower warming predictions, fixing other problems might make have the opposite effect. Moreover, there are lots of models that fit the present climate equally well but that predict very different futures. So, they say, scientists need to get cracking and test their model components against real world data to figure out which ones make the most sense.
If you thought that basic step had already been taken 20 years ago when we first started being told not to ask any questions because the science was settled, then it’s time to get your head out of the clouds.