Do roughness changes of tropical deforestation affect surface energy balance partitioning? No, they don’t. That’s what we found when we estimated the effects from first principles.



Figure: Areal photo showing the difference between tropical rainforest (lower half) and soybeans (light green) at Tanguro ranch in Brazil. Photo by Paulo Brando (click on the link to see website with the photo and more information on the field site).

Cutting down tropical rainforests and replacing them with soybean fields alters how the land surface functions, and this affects the atmosphere.  Rainforests have a heterogeneous canopy that absorbs sunlight very well and is aerodynamically rough, and they have deep-reaching root systems that allow them to draw water from deep within the soil, especially during the dry season when water input by precipitation is limited.  When trees are cut down and replaced by soybean fields, these physical aspects of the land surface are changed, thus impacting how the absorbed solar energy is partitioned at the surface, and how this energy is transferred into the overlying atmosphere.  Tropical deforestation is one of the many aspects of global change that has been dealt with over the last decades, evaluated with observations and climate models, so what else can add new insights?  And what can these insights be used for?

We recently looked at deforestation in a different way, just published in GRL.  Instead of using semi-empirical formulations of the energy partitioning at the land surface or a climate model, we used our thermodynamic approach to surface energy balance partitioning.  The big problem in estimating the surface energy balance are the turbulent fluxes, as turbulence is inherently a highly complex process with the motion of many whirls of air being involved.  In our approach, we constrain these fluxes by the thermodynamic limit of maximum power.  This additional constraint allows us to treat the surface energy balance partitioning in an analytic way (i.e., with paper and pencil) without the use of semi-empirical turbulence parameterizations that are commonly used in land surface and climate models.  This actually simplifies the formulation of turbulent fluxes and bases them only on physical constraints.

We then used observations from Tanguro Ranch (see Figure above), located in the Brazilian part of Amazonia.  Our institute is involved in maintaining eddy covariance stations over a tropical rainforest and over a nearby soybean field at this ranch. We used absorbed solar radiation and the ground heat flux from these observations as inputs to our thermodynamic approach and predicted the partitioning into net longwave radiation and turbulent fluxes.  We then partitioned the turbulent fluxes further into the sensible and latent heat flux using equilibrium partitioning, which are simple fractions related to the psychrometric constant and the slope of the saturation vapor pressure curve (terms that are used, e.g, by Priestley&Taylor 1972’s equilibrium evaporation, or by Penman 1948, although the approach dates back to the Austrian meteorologist Wilhelm Schmidt, who developed this partitioning back in 1915).


Figure. Surface energy partitioning of solar radiation (red line) into net longwave radiation and turbulent fluxes (light blue line).  Taken from Fig. 1 of Conte et al. 2019.

We found that our approach worked very well in predicting turbulent fluxes over both sites, rainforest and soybean field (see Figure).  A small offset in the predicted fluxes we could attribute to a change in longwave downwelling radiation due to seasonal changes in atmospheric water vapor content.  Also, the partitioning into sensible and latent heat using the equilibrium partitioning worked very well, except for the soybean site in the dry season where apparently soil water availability severely reduced the latent heat flux.  The magnitude of turbulent fluxes however, as predicted from the maximum power limit, remained the same.

What I found most interesting about this result is that the drastic difference in surface roughness played essentially no role for the turbulent fluxes.  This is not to say that surface roughness does not play a role — it surely affected the vertical wind profile, although we did not evaluate this.  But that we were able to explain surface energy balance partitioning without considering changes in roughness suggests that turbulent fluxes over land, particularly at the diurnal scale, are predominantly driven by surface heating and free convection.  And since this is basically what our thermodynamic approach constrains (the magnitude of free convection), this is likely the reason why our approach worked so well.

What are our next steps?  First, one could use this approach already to estimate impacts of deforestation on evapotranspiration and the water cycle by combining this approach with a soil water balance calculation.  One can also use our approach as a baseline reference for surface energy balance partitioning.  As one of the reviewers pointed out, our approach probably predicts surface energy balance partitioning better than many climate models, something that we want to look into in more detail.  On the development side, we work on linking this approach with estimating surface and air temperatures (see also my previous post here).  And as Hisashi Ozawa, one of the reviewers, pointed out that the thermodynamic formulation is still quite crude as it does not distinguish between dry and moist convection and the dissipative losses by evaporation into unsaturated air.

So there is certainly more to be learned from thermodynamics about land-atmosphere interactions.  Yet, this does not need to rely on incomprehensible complex numerical simulation models, but can be done in a simple, transparent, and analytical way.

If you think surface and air temperature are basically the same thing, think again. Or read our new paper.

In meteorology, air temperature measurements are typically taken 2m above the surface.  It is a routine measurement at weather stations, and this temperature is the basis for analyzing trends, such as global warming.  The temperature of the surface is not so often measured, but it can be inferred by satellites from how much radiation is being emitted by the surface.  Being only 2m apart, one may think that the temperatures basically reflect more or less the same, given their close proximity.

We actually found out that this is not the case: surface temperature responds much more strongly to a lack of water than air temperature.  This finding was just published in our article in the journal Geophysical Research Letters.

Our starting point was to look at the diurnal variation in temperature in a different way.  Typically, variations are plotted against time.  But we know that the strongest forcing of the diurnal cycle is the Sun.  It defines the night (no sun), and solar noon (strongest sunlight).  So what we did is that we plotted temperature variations against solar radiation, rather than time (see Figure).  What we then found is an interesting hysteresis, showing an almost linear increase in air temperature in the morning to noon, but then the temperature stayed almost the same over the course of the afternoon.  We already found this characteristic pattern in earlier studies (Renner et al 2016, Renner et al 2019) in Luxembourg, but did not quite know what to do with it.


Figure: The air temperature hysteresis measured at the rooftop weather station at our institute on 19 July 2016.

We then looked at the rate by which the temperature increased in the morning in response to an increase in solar radiation.  Mathematically this corresponds to the derivative of temperature to solar radiation, and we called this the warming rate.  We then looked at how this warming rate is affected at different levels of water availability.  For this, we used observations of the turbulent heat fluxes, and calculated the evaporative fraction, which expresses how much of the turbulent heat flux is composed of evapotranspiration.  It is a measure of water availability.  No water, and the evaporative fraction is 0.  A value of 1 is typically not found for the evaporative fraction because there needs to always be some sensible heat flux.  In any case, the more water is available, the higher the evaporative fraction is.

What we then found is that the warming rate for surface temperature decreases quite substantially with water availability.  This is what one would expect.  It is the phenomenon known as evaporative cooling.  So the more water is available, the less the temperature rise during the day.  This is no surprise.  But when we looked at the warming rate of air temperature, we found that this rate is almost independent of water availability.  It showed practically no variation with evaporative fraction.  We did not expect this, and so we learned something quite interesting about land-atmosphere exchange.

The lack of response in the warming rate of air temperature has a relatively simple, physical explanation.  Hysteresis behavior, as seen in the variation in air temperature, results from storage effects, and the hysteresis in air temperature results from the heat that is stored during the day in the lower atmosphere.  The size of this heat storage is, however, not fixed, but it depends on the height to which the boundary layer has grown.  And it is well known that boundary layer growth depends on the sensible heat flux.  So this is the key to understand the lack of response.  When water becomes more limiting, evaporation is reduced and so is the evaporative fraction.  This reduction is associated with an increase in the sensible heat flux.  The greater sensible heat flux then results in greater boundary layer growth, so the size of the heat storage is increased, so the air temperature actually does not increase.  In other words, the lack of response in air temperature to evaporative fraction is due to the compensation of boundary layer growth.

Why this is important to know?  Well, it tells us, for instance, which temperature to look at if we want to infer evaporation from the surface.  We should look at surface temperature, not air temperature.  On the other hand, air temperature includes information on the boundary layer, which I think is pretty cool.  It also relates to our work on the thermodynamics of land-atmosphere exchange, starting with our paper on the contrasting climate sensitivities of ocean and land surfaces (Kleidon and Renner, 2017) , where we already dealt with the boundary layer as being the buffer of the strong diurnal variation of solar radiation and that is clearly different from an open water surface (where the buffering takes place below the surface in the water, see ).  It also fits with the insight gained from our thermodynamic perspective that the magnitude of turbulent heat fluxes is set by how much power can be gained from solar radiation (Kleidon and Renner, 2018).  It fits to what we found here, that the sensible heat flux compensates the reduction in evaporation.

So our next steps from here is to look to see how general this behavior is.  So far, we only looked at one station, and Annu is currently extending this analysis to other stations.  Also, I have been working on a theoretical derivation for the warming rates that looks very promising.  So there is certainly more to come from this line of analysis of land-atmosphere interactions.

Making sense of German wind energy

I thought it might be a good idea to meet with wind energy critics to hear what they have to say. But my experience told me otherwise. Perhaps I should have been warned. They are associated with a German organization called „Vernunftkraft“, which has „sensible“ in its name. How presumptuous! It should have been a red warning flag for me.

We recently published a paper that evaluated wind energy generation over the last 15 years in Germany. It was published by PLOS ONE, an open access journal so that everyone can get access to it (here is the link). I was already anticipating some responses to it before it got published. After all, wind energy is a highly polarizing subject. So we decided to issue a press release, as it helps us to explain what the background of this paper was and what the results mean (the press release is here). After its release, I got contacted by a few media outlets, they published it as news, and then I got some responses, by e-mail, or by phone. It is quite a typical line of events.

It is also quite typical that after we publish something on wind energy I get contacted by wind energy critics. In the past, we have published work on limits to wind power that are much lower than what others have said. There are good, physical reasons for us to say this, and these low limits become important at large scales, although we are currently nowhere close to such limits. But I got plenty of phone calls of people who want to use these findings to discredit wind power in general. Either because there are wind farms planned near their village, or because they find nuclear power cooler, or god knows why.

What I feel quite uncomfortable with is that by making physical arguments of lower wind energy potentials, I am being put into a wind critics box. I am not a critic of wind energy. In fact, I am all for renewable energy, as it is the only way out of the global warming mess that we are currently in. This is a logical response to counter global warming. And not just that, it is the logical next step into a sustainable future if human societies want to sustain their quality of life. The pressing question for me is how we can make the transition to a renewable energy future the most effective, and this involves, to some extent, questioning resource potentials for wind energy. Questioning the need for this transition to renewable energy does not even come close to my mind.

With regards to questioning wind resource potentials, in our most recent paper, for instance, we found out that we can explain the produced wind energy fairly well using wind fields from the German weather service and turbine information. Yet, what was produced was about 30% less than what was expected from the wind fields. Some of this we could attribute to turbine aging or to shading effects within wind farms. But the largest part we could not explain. I think this is quite important to know that actual wind energy generation is at least 20% less than the resource potential.

Another aspect we found is that wind speeds in Germany declined over the time period we evaluated at rates that are similar to those reported in the global stilling literature (e.g., see Wikipedia entry here). I do not know whether this trend continued in Germany in recent years, and the reasons for global stilling are also mostly unknown, at least in my opinion. But it is a trend that certainly does not help wind energy in the future to stay productive, and this is important to know. (Actually, we only noticed this because one of the reviewers specifically asked about such trends.)

Also, we could not find statistically significant indicators for decreased efficiencies of wind energy generation due to large-scale wind speed reductions that we would have expected based on our previous work. But this is mostly due to the fact that too many variables correlated with each other, so we could not identify the installed capacity of wind turbines in a region as a significant factor that led to reduced yields. On a qualitative level, it seemed to us that one may find such negative effects already in the northern and northeastern parts of Germany, but this would need more work.

So there is certainly more questioning, more work, and more research to be done to inform the best way to shift to renewable energy.

When the two wind energy skeptics then came recently to my office, they had something entirely different in mind. First, they pointed out that by being retired physicists from industry, they were very well qualified for the topic. But then, after they sat down on the sofa in my office, they demonstrated their “qualifications” by quickly turning into preaching mode. I felt like I had Jehova‘s witnesses sitting in my office. Instead of sins, they talked about the evils of renewable energy, and instead of paradise, they talked about the glory of brown coal. I just could not believe what I heard, stunned by the lack of science and rational thinking! It was hopeless. It did not take long for me to decide that it was time for them to leave my office.

It left me with a sense of frustration and that a transition to renewable energy does not just involve rational thinking, science and technology. It also involves using science to open up human minds beyond to what they are used to and to be able to envision the future. But on that challenge, it seemed that I have failed miserably.

Starting a blog

Why do I start writing a blog? Well, I see a few reasons:

  1. From time to time I get e-mails with questions about my research or regarding my opinion. Sometimes, these are quite good and valid questions, yet I feel that it would be useful to document and/or share the answers. A blog would be a great place to do so.
  2. Sometimes, I feel like explaining why I think a certain paper from our research group is cool, and explain more broadly what it implies. There is not really space in a scientific, peer-reviewed manuscript to do so. So a blog may do well in such added information.
  3. Sometimes I come across something curious that I like to share. Just a bit of insight, not enough for a paper. But enough for a blog post.
  4. I’ll also use it to post some activities regarding my field of research, like when we organize another workshop or a block course on thermodynamics. A blog may work well for this as well.
  5. There may be other reasons too, but I cannot think of it right now.

So I’ll try this out and see how it goes. My focus here is to be objective and scientific, not opinionated. My approach, just as my research, is from physics and thermodynamics.

Any feedback is very welcome! I may need some time to get back to it though, so I apologize for any delays in advance.