Biotic Pump Q&A #2: The Dual Role of Condensation

Why forests need both the precipitation mass sink and latent heat release to drive atmospheric moisture transport.

Last week I had the privilege of participating in the course “How Trees & Forests Shape Our Climate”, where I spoke about the biotic pump. Many thoughtful questions were raised, and we did not have time to explore them all. I would therefore like to address some of them here, along with questions Peter Wurmsdobler recently raised on this Substack about the atmospheric dynamics of the biotic pump. (Biotic Pump Q&A # 1 is here).

1. What is the biotic pump and how does it work?

The biotic pump is a set of physical and ecological mechanisms by which natural forests actively regulate the transport of atmospheric moisture from the ocean to land.

This atmospheric transport is essential because forests continuously lose soil water to gravity. Water on land cannot be fully recycled: part of it inevitably drains back to the ocean and must be returned through the atmosphere.

The biotic pump, therefore, is not just about how forests make rain. It is about how forests bring water back from the ocean to land—and, ultimately, how forests make rivers.

Why does the atmosphere deliver water more efficiently to a forest-covered continent than to a barren one?

Forests moisten the atmosphere through transpiration. When green leaves open their stomata to capture CO₂, large amounts of water vapor are released into the air.

Water vapor is the fuel of the atmospheric engine. When moist air rises and water vapor condenses, vapor is removed from the gas phase and precipitation forms. As precipitation removes mass from the air column, local air pressure decreases.

The resulting low-pressure zone pulls moist air from the ocean toward the forest. As this incoming air rises, condensation and precipitation continue, maintaining the pressure drop and sustaining the inflow.

In this way, forests actively draw atmospheric moisture inland.

2. Why does air rise over the forest?

This very interesting and deep question was asked by Peter Wurmsdobler.

The air circulation associated with the biotic pump can be described simply: moist air flows toward the forest near the surface, rises over it, spreads outward aloft, and descends over the ocean.

If we ask why air rises, we must also ask why it flows in at low levels, why it flows out aloft, and why it descends elsewhere. These motions cannot be explained independently. They are all components of a single circulation system constrained by the conservation of mass, energy, and momentum. The circulation must be explained as a coherent whole.

We have already established that low-level air flows toward the forest because surface pressure drops when water vapor condenses and precipitation forms. Air therefore moves from the high-pressure region over the ocean toward the lower-pressure region over the forest.

Air at upper levels can also move horizontally from high to low pressure. For this to occur, the upper-level air over the forest must be warmer than over the ocean at the same altitude. When upper-level air flows out of the forest column, a slight local pressure deficit develops aloft. As a result, air from below rises to compensate. This ascent can therefore be understood as a consequence of the warm outflow.

If the upper-level air were not warmer over the forest than over the ocean, air pressure aloft would be lower over the forest than over the ocean at all heights — not just at the surface. In such a case, to move against the pressure gradient (from low to high pressure), the air would need to possess substantial kinetic energy. A situation of this kind is observed in hurricanes, where the horizontal pressure gradient is so strong that it extends through much of the air column: pressure in the center is lower than at the periphery at nearly all levels, and the upper-level outflow is governed to a large extent by centrifugal forces.

But over the forest, wind velocities are low (most of the kinetic energy generated at low levels is dissipated by friction), and the air would not possess sufficient kinetic energy to overcome the pressure difference.

We therefore conclude that the upper-level air over the Amazon must be warmer than at the same level over the ocean. This enables outflow and ascent, even if at the surface the Amazon forest is colder than the ocean — the “cold Amazon paradox,” often highlighted by Antonio Donato Nobre, author of The Future Climate of Amazonia Report.

Where does this higher temperature come from? As low-level air moves inland from the ocean, it accumulates moisture, which then condenses in rising air and releases latent heat. Because the atmosphere over the forest contains more moisture, the upper-level air becomes warmer than over the ocean at the same height. Thus, condensation lowers air pressure at the surface and raises it aloft, enabling inflow, outflow, and ascent. Descent follows dynamically from the other three.

This brings us naturally to the next question.

3. How does the biotic pump concept differ from the prevailing theory?

How air circulation works has long concerned scientists. Figure from Max Margules (1901, p. 521) illustrating atmospheric circulation driven by differential heating.

While we have agreed that the four legs of the circulation — inflow, ascent, outflow, and descent — cannot be explained independently, we now appear to have two separate explanations: the inflow is driven by the precipitation-related surface pressure drop, while the outflow is attributed to the pressure surplus aloft associated with latent heat release. Yet two independent physical drivers cannot govern a single, dynamically constrained circulation.

Conventional thinking has long held that the temperature-driven pressure surplus aloft drives the entire system. In this view, upper-level air diverges from the atmospheric column over the forest; this divergence induces ascent and creates a surface pressure deficit, which in turn drives inflow, while descent follows from the combined inflow and outflow. Within this qualitative framework, the pressure drop associated with the precipitation mass sink appears unnecessary.

However, it has long been recognized that, for the reasons discussed below, latent heat release alone cannot drive an appreciable large-scale circulation. In the scientific literature, this limitation was summarized by Lindzen and Ngiam (1987) as follows (emphasis added):

The work of Schneider and Lindzen (1977), Schneider (1977), Stevens et al. (1977) and Stevens and Lindzen (1978) suggests that the flows generated by cumulus heating [i.e. by latent heat release — AM] do not contribute effectively to low-level convergence (at least for time scales > 1 week) because the cumulus heating peaks in the upper troposphere and the forced motions decay away from the heating maximum.

Thus, in the conventional framework, forests cannot power atmospheric moisture transport merely by enriching the atmosphere with water vapor (see also the section “The biotic pump and some confusion in atmospheric science” in Biotic Pump Miscellaneous).

Therefore, taking into account the pressure gradient associated with the precipitation mass sink is crucial. Why is it crucial? Because there is a fundamental difference between a pressure gradient aloft and one at the surface. The latter is far more efficient at generating air convergence (inflow).

Why is that so?

Our planet is rotating. In a rotating, frictionless atmosphere, an equilibrium state can form — the geostrophic balance. In this state, a horizontal pressure gradient that would otherwise accelerate air toward low pressure is balanced by the Coriolis force. Air then flows along the isobars (lines of equal pressure) rather than across them. As a result, creating a pressure surplus in the upper atmosphere, where friction is negligible, produces only weak cross-isobaric motion and little low-level convergence.

At the surface, however, geostrophic balance cannot fully establish itself. In addition to the pressure gradient and the Coriolis force, friction acts to dissipate kinetic energy. Because friction weakens the compensating balance, air can cross isobars and flow toward low pressure. A surface pressure gradient is therefore much more effective at generating convergence than a pressure gradient aloft.

The precipitation mass sink is thus the key physical mechanism that creates a surface pressure gradient and draws moist air from the ocean. Once inflow is established, the associated condensation provides sufficient warming aloft to permit outflow. In forest-driven circulation, it is the inflow — not the outflow — that is primary. (Hurricanes operate according to a similar dynamical logic.)

To conclude, by identifying and emphasizing the precipitation mass sink as the key mechanism, the biotic pump concept explains how forests regulate atmospheric moisture transport.

Let us now take a less scientific stance and consider an analogy, even if it may be somewhat far-fetched.

To conceive a child, both a father and a mother are needed. A father cannot give birth to a child in principle. A mother can, but only after her interaction with the father. If some hypothetical aliens—like those reptiles in Ugo Bardi’s saga—were to study only human males, they might conclude that humans cannot reproduce and would be forced to attribute the appearance of children to some mysterious external process. The same applies to the conventional focus on latent heat alone — it leads to the conclusion that forests cannot drive winds.

Meanwhile, if the phenomenon of childbirth is studied in its entirety, with females included and their interaction with males taken into account, the picture becomes clear: humans are indeed able to reproduce (although less and less so, as Ugo Bardi discusses in his recent book The End of Population Growth).

Likewise, only when the previously neglected precipitation mass sink is considered together with latent heat release does it become clear how forests are able to regulate atmospheric moisture transport.

Once we acknowledge the different dynamics of the upper and lower atmosphere, we can turn to the next question.

4. Can the biotic pump operate on the U.S. East Coast, downstream of the jet stream?

The actual question was:

I can see how this biotic moisture pump effect could take place on the West Coast, which is up-wind of the continent re the Jet Stream, but how would it happen on the East Coast, which is downstream re the Jet Stream?

Jet streams are narrow bands of very strong winds that flow high in the atmosphere, typically near the top of the troposphere at about 8–12 kilometers above the surface. They move predominantly from west to east and can reach speeds of 100–300 km/h (and sometimes more).

Jet streams form where there are strong horizontal temperature contrasts — for example, between cold polar air and warmer midlatitude air. Again, because the Earth rotates, these temperature differences generate strong pressure gradients that, under geostrophic balance that we discussed above, produce fast winds flowing along the isobars.

Because temperature and pressure gradients are irregular rather than forming perfect circles, jet streams meander. These meanders contribute to the development and steering of mid-latitude weather systems, including cyclones and anticyclones.

However, the key point is that being upstream or downstream of the jet stream mainly describes how weather systems are steered aloft, not how moisture is supplied near the surface. Most moisture transport occurs in the lower troposphere, where winds are often quite different from those in the jet. Along the East Coast, warm Atlantic waters (especially the Gulf Stream) provide abundant moisture, and weather patterns guided by the jet regularly generate low-level onshore southerly or southeasterly flow. Likewise, along the same latitudes in Asia, there is an East Asian monsoon, with moisture delivered inland from the south and southeast. Condensation over forested land can then support continued inland moisture transport.

In one of the first biotic pump publications we explored precipitation distribution from the American and Asian East Coasts inland, see curves 4 and 5 in the image below. We can see that there is a more or less stable distribution over richer vegetation (closed symbols) and then a decline in a treeless zone (open symbols). This is naturally consistent with considerable moisture transport from the east.

Fig. 2 from Makarieva et al. 2009. Dependence of annual precipitation P (mm/year) on distance x (km) from the ocean over non-forested territories (open symbols) and forest-covered territories (closed symbols). Note that the biotic pump concerns moisture convergence (i.e., net moisture import) rather than precipitation per se. However, given the positive correlation between precipitation and river runoff, precipitation can be used as a proxy for runoff (cf. Fig. 5 in Makarieva et al., 2013, where both runoff and precipitation are shown).

Related to this discussion of how moisture propagates inland is Peter’s second question:

5. Does the biotic pump operate as a single large circulation cell?

Or does it consist of many smaller cells that, like runners in a relay race, pass moisture inland from one to another?

The regions of ascent and descent in the major atmospheric circulation cells — such as the Hadley, Ferrel, and Polar cells — extend over a thousand kilometers or more. Therefore, a large forest spanning similar distances, together with the adjacent ocean, can also function as a single circulation cell.

This does not mean that trees do not pass moisture to one another. Transpired moisture is added to the oceanic inflow and transported inland. However, it is the forest as a whole that undergoes large-scale ascent or descent.

Below is Fig. 5 from Fernández-Alvarez et al. (2025), who examined the progression of the anomalous 2023 drought in the Amazon forest.

The omega variable in the first column represents the rate of change of air pressure following the motion of an air parcel. When air rises, the surrounding pressure decreases and omega is negative; when air descends, pressure increases and omega is positive.

A positive omega anomaly, such as the one shown in the figure, indicates that descent is stronger than usual. We can see that this anomaly spreads over the forest as a whole. This suggests that the compensatory mechanisms also operate at the scale of the entire region.

The large-scale ascending and descending air motions are directly relevant to our next question:

6. How do forests cool the Earth? Why doesn’t the released heat remain in the atmosphere?

As I understand it, a major way that the biosphere helps regulate the temperature of the earth and that evaporation helps cool it is that when liquid water evaporates and changes from liquid to vapor it requires 590 kal of energy per one cubic centimeter of water and this energy/heat travels up with water vapor where it can leave the atmosphere and thus, goes away from the earth, cooling it. But what prevents this energy from being held in the toposphere closer to the earth and thus, contributing to heat?

What is important to bear in mind is that when condensation occurs and latent heat is released, this energy is not emitted directly as thermal radiation and therefore cannot be immediately radiated to space. The question of what ultimately happens to this released energy is therefore very relevant.

A useful way to understand latent heat release during condensation is to begin with evaporation. During evaporation, only the more energetic molecules in the liquid have enough kinetic energy to overcome the attractive forces between water molecules and escape into the vapor phase. This selective removal of higher-energy molecules lowers the average kinetic energy of the remaining liquid, causing the surface to cool.

While the liquid water cools, the vapor phase does not become warmer as a result of evaporation. Escaping molecules must climb out of the intermolecular potential well, converting kinetic energy into potential energy; they therefore do not inject a surplus of kinetic energy into the vapor.

During condensation, the reverse transition occurs. Vapor molecules descend into the intermolecular potential well. The associated decrease in potential energy is converted into kinetic energy through molecular collisions, warming both the condensate and the surrounding air.

In other words, latent heat release upon condensation manifests as an increase in the kinetic energy of molecular thermal motion.

Now the atmosphere must dispose of this added heat by radiating it to space. This requires greenhouse gas molecules — mainly water vapor and CO₂. Through molecular collisions, these molecules become energized and can emit infrared (thermal) photons.

Whether those photons escape to space depends on how much absorbing air lies above them. The higher in the atmosphere the emission occurs, the fewer greenhouse gas molecules remain overhead to absorb the radiation, and the greater the chance that the photon will escape. (I discussed this in our Global Cooling from Transpiration webinar and in the very first post on this blog; recently Rob Lewis and Ali Bin Shahid have also examined this effect.)

Now how the air circulation plays in. If latent heat is released aloft, where the air is thinner, and the warmed air remains there for a sufficient time, a significant fraction of the emitted radiation can leave the Earth system directly. This is how cooling occurs.

If, however, the warmed air rapidly descends and returns heat to the surface, radiation emitted there must pass through the entire troposphere. With many absorbers above, much of it will be intercepted and re-emitted downward, strengthening the greenhouse effect. In this case, no effective cooling occurs.

That is why air circulation is central to understanding the cooling effect of transpiration. I conclude with the quote from our paper devoted to this subject (emphasis added):

2.3. Dependence of global transpirational cooling on atmospheric circulation

The higher up in the air column that convection transports heat, the more pronounced global cooling it exerts. This is because the energy is radiated more directly to space from the upper atmospheric layer (cf. Figures 3E, F). In addition to the altitude, it matters how rapidly the cooled air descends. When the air rises and increases its potential energy in the gravitational field, its internal energy accordingly declines, and it cools. While evaporation cools the evaporating surface, the release of latent heat during condensation in the rising air partially offsets this decline of the internal energy of air molecules, making the rising air warmer than it would be without condensation. Radiating this extra thermal energy to space takes time. The more time spent by the air warmed by latent heat release in the upper atmosphere (above the main absorbers), the more energy is radiated unimpeded to space and the stronger the global transpirational cooling. With the characteristic radiative cooling rate of the order of 2 K day⁻¹, it takes about 15–30 days to radiate the latent heat released by tropical moist convection (Goody, 2003).

Therefore, long-distance moisture transport (including the biotic pump run by forests, Makarieva and Gorshkov, 2007) enhances global transpirational cooling: moist air travels for many days, and thousands of kilometers from the ocean to land, where it ascends and latent heat is released. The dry air warmed by latent heat makes the same long way back in the upper atmosphere, thereby radiating energy to space (Figure 4). If, on the contrary, the warmed air descends rapidly and locally, then most heat is brought back to the surface before it is radiated, and the global power of the net cooling effect can be nullified. Therefore, disruptions in the long-distance moisture transport (e.g., by deforestation) and violent local rains should warm the Earth. In smaller convective clouds up to a quarter of ascending air descends locally at a relatively high vertical velocity (Heus and Jonker, 2008; Katzwinkel et al., 2014). These effects are not taken into account when assessing the temperature effects of land cover changes (e.g., Bright et al., 2017).

Current global climate models do not correctly reproduce either the long-distance ocean-to-land moisture transport or the moisture transport over the ocean (Sohail et al., 2022). For example, the Amazon streamflow is underestimated by up to 50% (Marengo, 2006; Hagemann et al., 2011, their Figure 5). This corresponds to a 10% error in the global continental streamflow, the latter being of the same order as global continental evaporation. Similarly, global climate models do not correctly reproduce how the local diurnal cycle of convection changes upon deforestation by producing extreme low and high temperatures (Lejeune et al., 2017, their Figure 7). These are indirect indications of the models’ limited capacity to reproduce global transpirational cooling.

I am grateful to Peter Wurmsdobler, to the participants of the course “How Trees & Forests Shape Our Climate”, and to its host, Hart Hagen, for their insightiful questions and persistent interest in the biotic pump topic. Peter has recently published a popular account of the biotic pump story, which can be read on his Medium page. Further questions and comments are very welcome.

Related reading: