Inferring dominant controls on transpiration across a hillslope transect from ecohydrological measurements and thermodynamic limits

Thursday, 25 September 2014
Maik Renner1, Sibylle K Hassler2, Theresa Blume2, Markus Weiler3, Anke Hildebrandt4, Marcus Guderle4, Stanislaus Schymanski5 and Axel Kleidon1, (1)Max-Planck-Institute for Biogeochemistry, Biospheric Theory and Modelling Group, Jena, Germany, (2)GFZ German Research Centre, Potsdam, Germany, (3)University Freiburg, Freiburg, Germany, (4)Friedrich Schiller University of Jena, Ecological Modelling, Jena, Germany, (5)ETH Zurich, Zurich, Switzerland
Abstract. The topography of hillslopes induces many heterogeneities which affect water and energy partitioning at the site scale. To disentangle the dominant controls on transpiration T we estimate atmospheric evaporative demand by the thermodynamic limit of maximum convective power and compare this with sapflow and data driven soil moisture water uptake measured at 5 locations across a hillslope transect. Most importantly, we find a strong linear relation to sap flow velocity (r2 = 0.84) at all trees across the hillslope. The slope of the linear relation is found to correlate with vegetation characteristics (tree size) and emphasizes the role of stand characteristics on T . The root water uptake estimates show higher transpiration rates (42±13% of atmospheric demand, r2 = 0.44) at the north facing slope than at the south facing slope (28±5%, r2 = 0.45). This effect of hillslope orientation is inline with the sap flow observations and shows that transpiration is lower at the south facing slope. Further soil moisture is on average lower on the south facing slope and seems to induce a long term control on tree transpiration.

We conclude that the large variability in daily T is strongly linked with variations in atmospheric demand driven by solar radiation. The apparent differences in the response to atmospheric demand, however, display the importance of specific location factors such as tree size or plant available water.

1 Introduction

Catchments are composed of contrasting hillslopes which induce micro-climatic heterogeneity and different land cover patterns. While the former controls short term variability of local water and energy partitioning by modulating absorbed solar radiation and lateral redistribution of water, the latter represents long term effects of the potentially anthropogenic modified soil-vegetation-atmosphere co-evolution. Observations of water and energy fluxes reflect both effects and there is a need to decompose the variability of short term fluctuations induced by land-atmosphere mass exchange and soil water availability from those long term more persistent patterns such as vegetation cover and plant available soil water storage. The identification of the spatial and temporal controls on T will increase the predictability of water and energy fluxes in catchments.

Here, we hypothesize that long term persistent patterns, which control the magnitude of transpiration, can be determined by their differential response to atmospheric evaporative demand. Hence, we expect that T scales with atmospheric evaporative demand, but location factors determine the strength of the response.

2 Methods

We estimate atmospheric demand by the recently established concept of maximum convective power which is based on the thermodynamic limit of the conversion of heat to convective motion(Kleidon and Renner, 2013, Kleidon et al., 2014). The physically based concept specifically takes the interaction between land atmosphere temperature gradients and heat fluxes into account, which show to have a trade-off with respect to the power which is generated by dry convection. The resulting optimum evaporation Eopt is similar to the well known equilibrium evaporation concept (Slayter and McIllroy 1961, Priestley and Taylor 1972), but is more general as it is based on absorbed solar radiation Rsn instead of net radiation:

  E  o p t   =  s  s + γ     R  s n   2 

where s is the temperature dependent slope of the saturation vapor pressure curve, and γ the psychrometric constant. Hence, only net solar radiation and temperature data is required to estimate atmospheric evaporative demand.

3 Data

We evaluate ecohydrological and meteorological measurements at 5 different sites across a well instrumented steep forested hillslope transect (north vs south facing) in the Attert catchment in Luxembourg over the vegetation period in 2012-13. For meteorogical data we use global radiation and air temperature from a nearby open field site. The five forested sites are instrumented with multiple sapflow sensors (heat pulse at 20 trees). Sap flow velocity vsap is proportional to the logarithmic ratio of the temperature differences of the lower and upper needles after the heat pulse. Here, we directly analyze vsap as proxy for tree transpiration because upscaling to tree and stand water equivalents introduces further uncertainties. Further, we use soil moisture measurements (3 profiles with regular measurement depths at 10 - 30 - 50 cm and one deeper drilling (approx 70 cm) at a selected profile) to estimate soil evapotranspiration ETsoil. To do so we measure the daytime reduction in soil moisture and assume that each soil moisture measurement is representative for the dynamics of the above soil layer. Because we cannot distinguish infiltration events from evaporative losses we filtered the data for dry days and removed days with nighttime dynamics larger than 1 Vol%. Assuming that soil evaporation is negligible at the observation depths, ETsoilcan be regarded as a measure for root water uptake and thus transpiration.

To test our hypothesis we evaluate (i) how much variability of daily average sap flow velocity vsap and soil moisture derived evapotranspiration ETsoil is explained by Eopt. We evaluate (ii) the effect of soil moisture on both T estimates and then (iii) test how much site and vegetation patterns influence the response to atmospheric evaporative demand.

4 Results and discussion

To test if sap flow is related to atmospheric demand we plot sap flow velocity vs. Eopt for three trees at each site across the hillslope transect in Figure 1. Every tree shows a strong linear relation to Eopt, independent of the respective location on the hillslope. The very high correlation (r2 = 0.84 on average) shows that the relatively large temporal variability in Eopt (standard deviation = 0.94mm∕d, mean = 1.52mm∕d) explains most variations in sap flow velocity. It is interesting to note, that global radiation has only a slightly lower r2 = 0.80, whereas air temperature r2 = 0.56 and vapor pressure deficit r2 = 0.67 have less explanatory power.

In the following we analyze the sensitivity of vsap to Eopt as: Ssap = dvsap /dEopt with units [m∕d] / [mm∕d], which is estimated by a linear regression through the origin.

Although the relations of vsap to Eopt are linear, Ssap differs between trees and sites. One reason are measurement uncertainties of vsap due to heterogeneities in sap wood and wounding which reduces heat conduction. As the sensors are being re-installed in spring each year, we can attribute the difference between years to these measurement uncertainties. The differences in Ssapbetween 2013 and 2012 is on average -0.02 but can range between -0.21 and 0.18 (estimated for 8 trees with more than 30 days of data at the same day of year) and shows no significant trend.

Apart from measurement uncertainties we can then test for the influence of location factors on Ssap. Evaluating Ssap for 2013 reveals an significant effect of the tree size (diameter at breast height [cm], DBH, Ssap = 0.0094(±0.00067)DBH). However, the strength of the tree size effect is different between the north and south facing slopes which indicates effects of other relevant vegetation parameters. For example tree composition is markedly different between the north (large dominant beech trees, DBH > 47cm, with understorey) and south facing slope (tall forest stand, DBH < 47 cm, no understorey). Hence long term growth and eventually forest management conditions are markedly different between the contrasting slopes and thus affect hillslope scale transpiration. We also tested if the regression residuals are dependent to the site average soil moisture content (θ). The majority of instrumented trees being dominant beech (Fagus) trees show no influence of soil wetness. However, young beech in the understorey of the north facing slope and two oak trees (Quercus) at the south facing slope showed a weak response to θ, indicating the effect of water limitation.

The relation of Eopt to the data-driven ETsoil confirms the link of evapotranspiration to atmospheric demand (r2 = 0.45 on average for whole profiles). We find a large variety of (root) water uptake per soil profile and measurement depth, which is likely due to the small representative scale of insitu soil moisture observations. While the water amount per soil depth is relatively similar, we find that the spatial variability increases with depth. At the north facing slopes there is significant water uptake (> 1mm/d) at 30 cm depth, whereas we find significant uptake also below 30 cm at a few profiles at the south facing sites. Despite that, we find that the on average wetter north facing slope (compare inlay histograms of location average soil moisture content in Fig. 1) has higher ETsoil rates (42±13% of Eopt) than the south facing slope (28±5%). This significant aspect related difference between the two contrasting hillslopes is also found for the sensitivity of sap flow. Although, both the daily ETsoil and vsap data are only sparsely correlated with daily variations in θ, the on average lower soil moisture values at the south facing slope indicate a long term control on transpiration trough plant plant available soil water storage.

5 Conclusions

Two different measures of the strength of transpiration across a steep hillslope transect have been used to evaluate the hypothesis that the sensitivity of transpiration to atmospheric evaporative demand reveals potential long term controls on forest transpiration. To estimate short term daily atmospheric evaporative demand we employed the concept of optimum evaporation Eopt which is derived from the thermodynamic limit of convective land-atmosphere exchange. Daily variations of sap flow velocity as well as soil moisture derived root water uptake are well explained by Eopt, showing linear relations independent of the location along the transect. Hence Eopt provides a simple, low data demanding, yet effective means to estimate short term daily atmospheric evaporative demand. Furthermore, the sensitivity to Eopt of both T estimates varied between single trees and soil profiles. Thereby, we find that the orientation of the hillslope has an significant effect on the sensitivity to Eopt and thus transpiration. Also tree size, and stand characteristics are consistently different between the slopes, which emphasizes their potential role as long term controls on transpiration.

We conclude that the combination of local scale eco-hydrological and surface energy balance measurements with thermodynamic theory on land-atmospheric exchange is a way forward to establish an energy balance in heterogeneous, complex terrain. The remaining challenges, however, are to estimate water limitation which largely controls the energy partitioning, for example, by stomatal control. Also confidently upscaling sap flow to transpiration fluxes, and to separate the different evaporation processes acting on soil moisture remain open questions.

This research contributes to the ”Catchments As Organized Systems (CAOS)” research group (FOR 1598) funded by the German Science Foundation (DFG). We thank all people involved in the measurement campaign. In particular the technical backbone Britta Kattenstroth (GFZ Potsdam) and the landowners for giving access to their land.


   Kleidon, A. and Renner, M.: Thermodynamic limits of hydrologic cycling within the Earth system: concepts, estimates and implications, Hydrol. Earth Syst. Sci., 17, 2873—2892,, 2013.

   Kleidon, A., Renner, M., and Porada, P.: Estimates of the climatological land surface energy and water balance derived from maximum convective power, Hydrology and Earth System Sciences (in press), 11, 265—306,, 2014.

   Priestley, C. and Taylor, R.: On the assessment of surface heat flux and evaporation using large-scale parameters, Monthly weather review, 100, 81—92, 1972.

   Slayter, R. O. and McIlroy, I. C.: Practical Micrometeorology, 310 pp., CSIRO, Melbourne, Australia, 1961.