Earlier this year Steven Sherwood (UNSW) and Qiang Fu published a short note in Science suggesting that the world is, in fact, getting drier, as a result of anthropogenic climate change.
Now we know that global precipitation tends to increase with global mean temperature; this is shown not only by models (there is a very consistent relationship, for the tropics and the extratropics as well as the globe, that holds all the way from the lat glacial maximum, LGM with CO2 < 200 ppm up to global-warming scenarios with elevated CO2 at the end of the century) but also by reconstructed climates from the LGM.
However, as Sherwood and Fu pointed out, there is more to “dryness” or “wetness” than precipitation. Evaporation matters as well. They used climate models to indicate that both relative humidity (and hence vapour pressure deficit) and the ratio of precipitation to “potential” evapotranspiration (PET) are predicted to decline over much of the world. These indicators seem to increase over India and part of east Africa but to decrease nearly everywhere else.
This is truly enigmatic. No other map of anything that I’ve seen looks like these maps. The paper is silent on how PET was calculated (the Penman-Monteith equation was used; but this requires canopy and aerodynamic conductances to be specified… how was this done?) while relative humidity also is not a purely physical variable, as it depends on the rate of evapotranspiration (and therefore on vegetation properties, partly) which is where most of the humidity comes from.
Nonetheless, these authors make a first-principles argument that is hard to fault. They point out that the land is warming faster than the ocean. This, too, is quite generally true, and shown in data from the LGM as well as model simulations from the LGM to a warm future. It happens because of evapotranspiration on land is necessarily more restricted than from the ocean, so more energy goes into heating the land surface. Meanwhile, the ocean (where precipitation comes from) is warming as well, but not as fast. Therefore, the source of precipitation to the land is increasing more slowly than evapotranspiration from the land.
Now at a WCRP Grand Challenges meeting in March, Steven admitted there is an obvious problem, to people who know palaeodata. Based on the general nature of vegetation at different times, it is quite clear that the LGM (with low CO2) was “dry”, while earlier periods of the Cenozoic like the Pliocene (much studied because of CO2 concentrations being as high as they are now, or higher) and the Eocene (with at least 1000 ppm CO2 and maybe a lot more) but definitely “wet”. During the Eocene, “mixed mesophytic forests” extended to the Arctic coasts.
A possible resolution of this enigma is to consider the direct effects of CO2 concentration on the physiology of C3 plants (including trees). Several lines of evidence show that the extreme reduction of forests at the LGM was due more to low CO2 (lowering water use effeiciency, i.e. plants needed to transpire more water to fix a given amount of carbon) than to the climate. Similarly, biosphere model runs with high CO2 increase the global cover of forests.
Even the recent CO2 rise has had measurable effects. In work led by group members Anna Ukkola and Trevor Keenan at Macquarie, we’ve shown that the threshold for water limitation of vegetation cover in Australia has shifted – from 900 mm in the 1980s, to 700 mm today – and this is fully consistent with what we predict for the effects of the rise in CO2. And after precipitation variations are corrected for, all but the wettest and the driest regions have become significantly “greener” due to CO2.
CO2 gets a very brief mention, in passing, in Sherwood and Fu’s paper. But it might be the key thing they missed.