Evapotranspiration Components Measuring Transpiration

None of the ET measurement methods mentioned in the previous section allows differentiation between T and Es or can be applied in heterogeneous stands (natural landscapes, for instance), when a distinction between species is required. New opportunities for the estimation of T, typically the main component of ET, especially in arid environments, are offered by sap flow methods, usually based on correlated heat transport. These methods can be applied to investigate the responses of T to different treatments and also allow the estimation of water uptake from different root zones (as in Nadezhdina et al., 2007). The daily T of the canopy (or sap flow density per unit of ground area) can be obtained by up-scaling the measurements performed in an adequate number of individual plants of sap flow density (per unit of active xylem area) or sap flow per plant.

The thermal sap flow methods include (1) the heat-pulse methods (e.g. Cohen et al., 1981; Cohen et al., 1985; Swanson and Whitfield, 1981) where the velocity of convective heat transport is related with mass flow; (2) the method based on the cooling (in relation to a reference) of a single probe constantly heated, being the relative cooling related with sap flow density (Granier, 1985, 1987a and 1987b), (3) the heat-balance methods (e.g. Sakuratani, 1981; Valancogne and Nasr, 1989), where the convective term in the energy balance of a heated part of the stem is related with the volume of moving sap, with no need to measure the xylem area, (4) combinations of (2) and (3) as for instance in Cermak et al. (1973) or (5) the heat field deformation method (Nadezhdina and Cermak, 1999; Nadezhdina et al., 1998, 2007). Reviews on these methods can be found in Jones et al. (1988), Swanson (1994), Smith and Allen (1996), Valancogne and Granier (1997), Tatarinov et al., (2005) and Roberts (2007).

The sap flow methods have important advantages for measuring individual plant transpiration when compared to other methods. They are easily automated, allowing a continuous data record for long periods of time (Granier and Loustau, 1994; Smith and Allen, 1996). Continuous data series are particularly useful for the construction and validation of models. Sap flow methods have been widely applied, frequently together with the EC method for both forest studies (e.g., Loustau et al., 1996; Köstner et al., 1998a) and for orchards and vineyards (e.g., Green et al., 1989; Shackel et al., 1992; Weibel and de Vos, 1994;

Valancogne, 1995; Ferreira 1996b; Braun and Schmid, 1999), in order to determine the tree individual contribution to the total water vapor flux of the surface.

These methods have been compared to others for the water balance estimation such as as lysimeters (gravimetric measurements with plants in pots, as in Ameglio et al., 1993), eddy covariance (Ferreira et al.; 1997b, Berbigier et al.; 1996, Silvestre et al., 1999; Wilson et al., 2001) and also numerical modelling (Peramaki et al., 2001). Their performance has been validated, but they still need development and testing before becoming routine techniques for monitoring tree behavior and water consumption in the field. Due to the complexity of sap flow in woody plants, the application of heat dissipation sensors, though apparently simple, is not straight forward and still needs adaptation. For instance, the Granier method, with low cost sensors suited to large samplings used in many studies, when compared to other SF methods or with micrometeorological methods, showed good agreement in several situations (e.g. Kostner et al. 1992; Berbigier et al., 1996), but other studies showed underestimations (e.g. Lundblad et al., 2001: 50% in Pinus sylvestris and Picea abies; Wilson et al., 2001: 50 to 60% in deciduous forest, Ferreira et al., 1997a: 30% in peach orchard). The results presented in this Chapter confirm a clear underestimation for high fluxes when comparing to a relatively reliable reference, such as the EC method or weighing lysimeters. Underestimations were also observed by other researchers when using other sap flow methods (Baker and Nieber, 1989; Cohen at al., 1993; Weibel and de Vos, 1994; Khan and Ong, 1995; Tarara and Fergusen, 2001; Klnitenberg and Hans, 2004). These results indicate some uncertainty in the use of sap flow methods to quantify transpiration.

Because the research presented in next Section deals with the so-called Granier method applied in several stands, we are giving a brief description of the basic principle, with some indications on ways to overcome some of the problems identified - mainly the recorded underestimation.

The Granier method uses two probes with temperature sensors, which are radially inserted into the trunk. One of the probes, in an upper position along the vertical axis, is heated while the lower one is kept at the undisturbed stem temperature (reference temperature). The distance between probes prevents the heat applied in the upper probe from influencing the lower one. The temperature difference between the two probes is related to the flux index (k) by the following equation:

where AT is the temperature difference between probes and ATmax the maximum value for AT, occurring for null flux situations.

Sap flux density (u) is determined using k (Equation [4], based on calibrations to obtain the empirical parameters a and P for the equation:

This leads to a relationship initially assumed independent from species (Granier et al., 1990; Valancogne and Granier, 1997) written as (Granier, 1985):

where u is expressed in m3 m-2 s-1. Sap flux F [m3 s-1] is determined from previous equations and from conducting area of the cross section, A [m2] (Granier, 1987b): F = u A.

The ET estimate underestimations observed, mainly for sap high flows, motivated an analysis of heat dissipation process. A finite difference simulation model was used as a tool for simulating the heat field around the linear heater. With this model it was possible to visualise the heat fields for varying flux densities and to estimate the sensitivity to varying parameters such as wood thermal properties, heater power dissipation and distance between probes. In addition it contributed to the analysis of the algorithms used in converting sensor AT measurements to sap flow estimates (Thomsen and Ferreira, oral communication unpublished). The results obtained, while confirming the influence of varying wood properties on the parameters of the empirical equation [5 a], could not explain the important underestimation obtained. Therefore an analysis of the saturation effects at very high flows still needs further investigation. This emphasizes the need to relate the SF data with data from a reliable independent method.

In addition, the influence of natural gradients (vertical thermal gradients in absence of heating) and the impact of trunk insulation were analysed. When the measurements were made close to the soil surface, it was not always possible to eliminate a peak by the middle of the morning followed by an apparent relative underestimation at midday when compared with the transpiration patterns obtained with other independent method. Experiments in chambers and open fields showed that the underestimation was linked to a positive temperature gradient (AT) between the two probes in absence of heating. Conversely a negative gradient leads to an overestimation of the sap flow density. In absence of any analytical solution, a pragmatic one was to use the difference between a heated AT sensor (two probes) and a non-heated one in order to obtain the corrected thermal gradient (Ferreira and Zitscher, 1996). When measurement is impossible in the same plant, an extrapolation based on physical parameters is suitable (orientation, irrigation events, meteorological variables).

The influence of the radial profile (variability of stem sap flow density in radial direction) on the interpretation of single point measurements at a specified depth was also considered. Corrections were made either directly, according to the method presented in Ferreira et al. (1998), or indirectly via equivalent statistical adjustment.

All the aspects mentioned were taken into account in the case studies presented in this chapter, with adaptations for each experimental situation. Besides, some lack of precision on hourly measurements (when assuming SF ~ T) can be related to the time lag between the water flow at the measurement point and the canopy transpiration (Senock and Ham, 1993; Loustau et al., 1996). This applies even to daily measurements as shown by David et al. (1997), when analysing the seasonal trend of time lag between sap flow and modelled T, for an Eucalyptus globulus stand, during a period of progressive drought.

With the Granier method, which presents operational and cost advantages, it is possible to sample multiple trees and to make long term inexpensive measurements. Provided minimum precautions and an underestimation correction based on temporary comparison with independent methods, it is possible to get reliable values of transpiration at the field level. The relationships between the values obtained from SF and EC methods (or SF and weighing lysimeters with individual plants) were used to refine T estimations. However, eddy covariance (EC) measures ET, not just T, so for comparison, soil evaporation (Es), when more than negligible, was measured and deduced from ET.

Was this article helpful?

0 0

Post a comment