What We Publish

Climate forcing and response to idealized changes in surface latent and sensible heat

George A. Ban-Weiss, Govindasamy Bala, Long Cao, Julia Pongratz, and Ken Caldeira · Environmental Research Letters 6, 034032 · 2011

Key finding. In the NCAR CAM3.1 model with a slab ocean, adding a uniform one watt per square metre source of surface latent heat while removing an equal one watt per square metre of sensible heat — an energy-neutral change — lowered global mean surface air temperature by 0.54 plus or minus 0.04 kelvin, mainly because the extra evaporation increased low cloud cover and so raised planetary albedo.

A bar chart with four groups along the horizontal axis — applied forcing, regressed forcing, change in surface air temperature, and change in precipitation — each showing the same four model cases as coloured bars with 95 per cent confidence intervals. Blue is the latent heat source alone, red the sensible heat sink alone, dark grey the two applied together, and a hatched bar the arithmetic sum of the two separate runs. The applied forcing group has only the blue bar at plus one watt per square metre and the red at minus one, the combined case being zero by construction. In the temperature group the blue bar is about plus 0.5 kelvin, the red about minus 1.1, and the grey and hatched bars both sit near minus 0.55. The precipitation group has the same pattern at larger magnitude, blue near plus 1.8 per cent and red near minus 2.4 per cent, with the grey and hatched bars close to minus 0.5.
Repartitioning surface heat from sensible to latent form cools the model even though the applied forcing is zero. In every group the dark grey bar for the combined case and the hatched bar summing the two separate cases agree within their uncertainties, so the latent and sensible components act nearly independently. Figure 1 from Ban-Weiss, Bala, Cao, Pongratz, and Caldeira (2011), Environmental Research Letters 6, 034032. Reproduced under CC BY 3.0. Extracted from the published PDF and resized for web display.

What question did this research address?

Changes in land use and land cover alter how surface energy is split between sensible heating, which warms the air directly, and latent heating, which carries energy away as water vapour. Forests generally evaporate more than the short vegetation replacing them, and irrigation raises evaporation too, so this partitioning — the Bowen ratio — shifts whenever land cover or land management changes.

Increasing evaporation is well known to cool the surface locally. But that cooling is not obviously a global effect, because the latent heat is released again wherever the water vapour condenses. On a planetary energy-balance argument the two cancel, and any global temperature change would have to come indirectly, through changes in clouds, water vapour, or the vertical temperature profile. Whether it does, and in which direction, had not been characterized.

Real land-cover change also alters surface albedo and roughness at the same time, which makes the latent-versus-sensible component impossible to isolate in a realistic experiment. This study therefore imposed the repartitioning directly, as an idealized heat source and sink.

What did we find?

Four simulations were run with NCAR CAM3.1 coupled to the CLM3.0 land model and a slab ocean at two degrees by 2.5 degrees resolution, with carbon dioxide held at 390 parts per million. The control was compared against a case adding latent heat and removing sensible heat together (denoted up-L down-S), a case adding one watt per square metre of latent heat alone (up-L), and a case removing one watt per square metre of sensible heat alone (down-S). The forcings were imposed directly in the model's surface-flux code, as sources and sinks applied at every time step over the whole Earth surface.

The energy-neutral repartitioning cooled the model. Global mean surface air temperature fell 0.54 plus or minus 0.04 kelvin and precipitation fell 0.41 plus or minus 0.07 per cent, even though no energy was added to or removed from the system. At the top of the atmosphere, reflected shortwave radiation rose about one watt per square metre and outgoing longwave radiation fell by about the same amount.

Low clouds are the mechanism. Separating the fast atmospheric adjustment from the slow response to surface temperature by the regression method of Gregory and colleagues gives a fast-response top-of-atmosphere imbalance of 0.49 plus or minus 0.34 watts per square metre in the cooling direction, matched almost exactly by a fast shortwave cloud forcing of minus 0.50 plus or minus 0.33 watts per square metre. The clear-sky shortwave change is near zero, so clouds, not water vapour, are doing the reflecting.

Longwave radiation shows essentially no fast response. Changes in atmospheric water vapour, in the vertical temperature profile, and in medium and high clouds appear only as slow feedbacks once surface temperature has already begun to fall, which makes them consequences of the cooling rather than its cause.

Splitting the experiment shows the latent heat term drives the cloud response. Adding latent heat alone warms the model 0.48 plus or minus 0.04 kelvin and raises precipitation 1.77 plus or minus 0.07 per cent, since it adds energy; removing sensible heat alone cools it 1.09 plus or minus 0.04 kelvin and cuts precipitation 2.36 plus or minus 0.07 per cent. But the fast increase in low cloud and in reflected sunlight is present in the latent-heat case and largely absent in the sensible-heat case.

The two components are close to additive. Summing the separate runs gives a temperature change of minus 0.61 plus or minus 0.06 kelvin against minus 0.54 plus or minus 0.04 kelvin for the combined run, and a precipitation change of minus 0.59 plus or minus 0.09 per cent against minus 0.41 plus or minus 0.07 per cent — differences that are not statistically distinguishable.

Once the forcing is diagnosed by regression rather than taken as applied, the feedback parameter is statistically indistinguishable across all three cases, at roughly 0.8 to 1.0 watts per square metre per kelvin. What differs between forcing agents is the fast response; the slow response per unit of forcing is much the same.

Why does it matter?

Evaporative cooling is usually described as a local effect that cancels globally. These results say otherwise for this model. When latent heating replaces sensible heating, surface temperatures fall globally and not merely where the evaporation happens, because more evaporation means more low cloud and more reflected sunlight.

That gives land-cover change a climate pathway separate from its carbon and albedo effects. Afforestation and irrigation lower the Bowen ratio and should carry a global cooling signal by this mechanism; anything that raises it should warm. The same argument runs in reverse for the physiological effect of rising carbon dioxide, which closes plant stomata and cuts transpiration.

Isolating one component of land-use change is the point of the idealization. A realistic deforestation experiment changes albedo, roughness, and heat partitioning at once and cannot apportion the result among them; imposing the heat repartitioning on its own can.

The magnitude should be treated as model-dependent. The forcing is uniform and global rather than patterned like real land-cover change, the ocean is a slab, and the result rests on the low-cloud response — the feature on which climate models disagree most.

Citation

George A. Ban-Weiss, Govindasamy Bala, Long Cao, Julia Pongratz, and Ken Caldeira (2011). Climate forcing and response to idealized changes in surface latent and sensible heat. Environmental Research Letters 6, 034032.

Related