Quantifying Transpiration Of A Coniferous Forest Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Evapotranspiration is an important component of ecosystem water balance, which is related to a wide range of disciplines, including ecological hydrology and meteorology. The correlation between ET and ecosystem production highlights the role of water as a principal limiting resource for plant photosynthetic metabolism. Because of this multidisciplinary focus, a number of methodologies have been developed to measure evapotranspiration or its components (i.e. transpiration, soil evaporation and interception) across a spectrum of spatial scales, which include individual plants, soil samples and profiles, atmospheric surface layers and entire watersheds. Examples of measurement techniques include soil (Daamen et al., 1993) and plant weighing lysimeters (Edwards, 1986), soil water budgets (Eastham et al., 1988; Cuenca et al., 1997; Jaeger and Kessler, 1997), sap flow (Smith and Allen, 1996), plant chambers (Cienciala and Lindroth, 1995), chemical tracing (Calder et al., 1986; Kalma et al., 1998), Bowen ratio (Denmead et al., 1993), eddy covariance (Baldocchi et al., 1988) and catchment water balance (Bosch and Hewlett, 1982; Swift et al., 1988). In particular, the eddy covariance technique is useful for generating estimations at the ecosystem level. In woody stands, the eddy covariance technique may be combined with tree sap-flow measurements to separately estimate tree and ecosystem evaporation (Paço et al., 2009).

Evaporation and transpiration fluxes are separately represented in ecosystem water balance models (Reynolds et al., 2000), but they are difficult to measure. The contributions of soil evaporation and plant transpiration to total ecosystem ET are highly variable in space and time (Ferretti et al., 2003). Prediction of evapotranspiration in these environments is complicated by the heterogeneous distribution of vegetation elements, and the difficulty of measuring components of water exchange at meaningful and comparable scales. Partitioning of the total evapotranspiration flux is important in these ecosystems due to the need to understand the role of biotic and abiotic factors that influence the efficiency of pulsed rain water circulation through the transpiration pathway, a process which contributes to photosynthetic gas exchange and productivity (Williams et al., 2004).

Soil evaporation should comprise a significant fraction of the ET flux following water input (Paruelo and Sala, 1995). However, soil evaporation and plant transpiration are often not independently verified. Soil evaporation rates initially are high following the wetting event, but decline rapidly as the soil surface dries (Ritchie, 1972). Recent studies demonstrated that the isotope technique is useful for measuring soil evaporation within a forest ecosystem (Tsujimura and Tanaka, 1998; Kubota and Tsuboyama, 2004; Liu et al., 2006).

Sap-flow measurements provide mechanistic details, along fairly short temporal scales, about physiological and environmental transpiration controls at the branch and whole plant level, and represent spatial scales several orders of magnitude. They are useful and versatile for partitioning total ET fluxes, because complex terrain and spatial heterogeneity does not limit their applicability. However, scaling to the ecosystem level can be complicated by spatially and functionally heterogeneous vegetation (Schaeffer et al., 2000). In fact, these measurements are especially well suited for determining species effects and other types of variability that occur in highly heterogeneous environments (Barrett et al., 1996; Wullschleger et al., 2000; Wullschleger et al., 2001).

Eddy covariance measurements above the canopy provide estimates of evapotranspiration at the high temporal resolution necessary to examine processes, but also at much greater spatial scales than sap-flow. One weakness of the eddy covariance technique is the size and shape of the representative region that contributes to the measured flux, the flux 'footprint', is not fixed in time (Horst and Weil, 1992; Baldocchi, 1997). In addition, eddy covariance measurements are sometimes difficult to interpret during weakly turbulent periods, such as nighttime (Lee et al., 1996; Baldocchi et al., 2000; Paw U et al., 2000). Furthermore, the technique cannot directly account for advection in areas of significantly heterogeneous or complex terrain, thereby limiting its applicability in some locations (Wilson et al., 2001).

Mountains play an important role as water reservoirs for the lowlands. Gongga Mountain, which is southwest of China, is an important area for biodiversity, since it contains an intact vertical zone, ranging from Subtropical to Frigid. In order to clearly distinguish between the influences of forest evapotranspiration and its components on the region, a comparison of different techniques is necessary; this allows for independent estimation of local water vapor exchange, and tests the general applicability of extrapolating smaller scale measurements to the study of larger regions. Processes that controlling evapotranspiration and its individual components can be examined in spatio-temporal detail at a level that is unavailable when using only a single measurement technique. In this paper, we compare the daily and annual estimates of evapotranspiration obtained using each technique, and discuss the character of evapotranspiration components. The objectives of the present study were (1) investigate and quantify the evapotranspiration and its components in the alpine watershed from April - October, 2009, (2) prove the utility of up-scaling methods in transpiration measurements (3) evaluate the efficiency of the different research methods.

2. Materials and methods

2.1. Site description

The study area is located at the middle and south sections of the Daxue Mountain Range, and on the southeastern fringe of the Qinghai-Tibet Plateau (101°30′-102°15′E and 29°20′-30°20′N). The altitude of the highest peak, the summit of the Hengduan Mountain, is 7556m a.s.l. The climate of Gongga Mountain belongs to the transition band between China's eastern monsoon subtropics and the frigid area of the Tibetan Plateau (Fig. 1). The eastern slope of Gongga Mountain is windward facing the monsoon with a wet climate character. The annual mean air temperature is 3.8°C, with extreme mean air temperatures of -4.3°C and 11.9°C, during January and July, respectively. The average annual precipitation is 1940 mm, of which 60.6% occurs from June to September (Fig. 2). Additional detailed features and information about this site are available in the literature (Cheng and Luo, 2004; Luo et al., 2005; Titov, 2007; He and Tang, 2008). The research site belongs to a subtropical mountain, humid monsoon climate. It has an intact vertical zone from the Subtropical Zone to Frigid Zone with abundant biodiversity.

2.2. Sap flow measurements

We used a modified heat-pulse velocity technique to measure xylem sap-flow on fifteen Abies fabri trees at the research location from April-October, 2009. To quantify the radial profile of sap flux density, sensors were placed into the outer xylem at 20-mm depth, via holes drilled with a gauge guide, into areas of peak sap flow velocity. All probes were installed on the northern side of trees to avoid direct solar heating and were shielded with aluminum foil to minimize temperature fluctuation in the sapwood. Tree DHB and characteristics are listed in Table 1.

Sap flow sensors measure the temperature differential (ΔT) between the paired heated and unheated probes. ΔT (recorded in mv) for each sensor pair was measured at 30s intervals and 30 min averages were stored on a datalogger. To convert these data into water flux, mean sap flow for the tree was calculated according to the empirical calibration of (Granier, 1987), which requires the following equation:


where ΔTmax is the maximum temperature differential at which sap flow is zero (Granier, 1987); Js is the half hour mean sap flow density (cm3¹’cm-2¹’h-1). If all observations were not available during daylight hours, that day was eliminated for comparative analyses in this paper, but interpolated estimates were used for calculating annual transpiration.

Sap-flow measurements were scaled to stand transpiration using knowledge of species composition in the sapwood area. Stand transpiration was computed every half hour by inputting the mean sap-flow density and the species-specific sapwood area index into the following equation:

(Hatton and Vertessy, 1990) (2)


where T is the stand transpiration (mm¹’d-1), A is plot area (m2), Fi is total transpiration of sample trees in the grade i ((cm3¹’d-1)), and Sk is the sap wood area's summation of all sample trees (cm2).

2.3. Eddy covariance measurements

Eddy covariance measurements of latent and sensible heat were collected beginning November, 2008 at half-hour intervals. The eddy covariance system was installed on a walk-up tower, 30m above the ground in the sample plot. Wind velocity and virtual temperature fluctuations were measured using a three-dimensional sonic anemometer. Fluctuations in humidity were measured with an open path.

To obtain daily evaporation it was necessary to estimate missing or rejected data. Missing or low-quality half-hour latent heat flux readings (i.e. due to anomalous turbulence statistics during rain events or instrument malfunction) were estimated using the two-week average value of the same period. The evaporation was calculated by the following equation after energy balance closure:


where LE is the latent heat fluxes (W¹’m-2); T is air temperature (℃); the total evaporation was calculated by summing each half-hour reading taken over the course of one day (mm¹’d-1).

2.4. Isotope measurements

Since stream water is a mixture of throughfall and soil water, and no isotopic fractionation occurs during water uptake by plant roots, the difference inδD andδ18O between throughfall and stream water during a non-storm runoff period can be explained by the evaporation process at the forest floor during infiltration. Thus, evaporation rates from the forest floor can be estimated from the difference in δ values between throughfall and stream water during a non-storm runoff period (Liu, 2006).

The evaporation rate of the forest floor can be estimated by the Rayleigh distillation equation under equilibrium conditions. Methods used in isotope data processing and calculation were described in more detail by Liu (Liu et al., 2006). The equations for the estimation of the evaporation rate are expressed as follows (Kubota and Tsuboyama, 2004):


where f is the remaining fraction of the water body; δ and δ0 are isotopic compositions (‰) of the diminishing and initial reservoir, respectively; and α is the equilibrium fractionation factor (‰). The α for δ18O is given by the following equation


T is the temperature (K).

By assuming that δ and δ0 are the isotopic compositions of throughfall and stream water (during a non-runoff period), respectively, f can be calculated from Eq.1.


Finally, the rate of evaporation from the forest floor to evapotranspiration (ER, %) is expressed as follows:


where TF is the throughfall, P is the precipitation and R is the runoff.

Isotope samples were collected three times per month in 2009 during non-storm period for rainfall, throughfall, soil water, and stream water. Rainfall samples were collected using a collector constructed from a stainless steel funnel connected to a polyethylene bottle. Ten collectors were placed on the forest floor to collect the throughfall. The steam sample was collected at the outlets of the catchment. All the samples were measured by mass spectrometer (Delta-S, Theomoqt) at the Institute of Mountain Hazards and Environment, CAS. Results of 18O and D were expressed in per mil unit δ-notation according to the Vienna Standard Mean Ocean Water (VSMOW) standard. The precisions forδ18O and δD were 0.1‰ and 1‰, respectively.

2.5. Hydrological conditions measurements

The observation system of the Gongga Alpine Ecosystem Observation and Experiment Station was established in 1988. The meteorological station was established in the watershed to measure climatic factors. Stand climate data included air temperature, soil temperature, humidity, precipitation, and solar radiation.

3. Results

3.1. Stand transpiration determined by sap flow techniques

The sap flow method is widely applied (Wilson et al., 2001), but often provides unsatisfactory results (Williams et al., 2004). Dominant components of natural vegetation (i.e. species, functional types) are heterogeneously distributed within ecosystems and among landscape units, and often have unique responses to wetting events (Williams and Ehleringer, 2000). Techniques that scale sap flow from individual trees to stand ET flux must account for the heterogeneously of sampling plots.

We used sap flow techniques to monitor and calculate the stand ET flux during the period of April-October, 2009. The daily mean ET flux in different months was shown in Figure 3. The ET flux was highest in April and September (Fig. 3).

Commonly, transpiration is higher during the May to August (Paço et al., 2009); we found that transpiration determined by the sap flow method during this period was lower than others month. Correlation analysis found that varieties of primary factor in different months (Table 2, 3) are the main reason of the phenomena. In the April, September, and October, the temperature has a significant positive correlation relationship with the sap flow (the correlation coefficient is 0.829) and the ET varies rapidly with the temperature (Table 2). From May to August, the primary determining factor is relative humidity (the correlation coefficient is -0.839) (Table 3), which has a significant negative correlation with the ET that decreases ET flux in these months.

Cross-correlation analysis between the temperature and sap flow showed a maximum correlation at a time lag of 0 min during April, September, and October (Fig. 4). Since we measured the sap flow at half an hour intervals, this study indicates that the time lag between temperature and sap flow is less than half an hour.

3.2. Soil Evaporation determined by stable isotope techniques

Rainwater, throughfall, and stream water were collected from April-October, 2009 for stable isotopic analysis at different catchments in the research area. Stream water during the non-storm runoff period was considered to reflect the effect of evaporation from the forest floor. Evaporation rates from the forest floor were estimated using isotope composition values in stream water and the total throughfall. The result is shown in Table 4. We observed that value of f in different month ranges from 92.06% to 96.50%, indicating that more than 3.5% of the total throughfall was evaporated from the forest floor from April-October, 2009. Using the estimated growing season evapotranspiration, the evaporation rate (ER) was estimated to range from 2.76% to 8.82%; and the mean daily soil evaporation in different months ranged from 0.12 mm to 0.34 mm. These measurements suggest that soil evaporation flux is highest in April, and soil evaporation is only a small portion of total forest ecosystem evapotranspiration.

3.3. Evapotranspiration in the research area

3.3.1. Evapotranspiration as determined by eddy covariance

We also measured the evapotranspiration using an eddy covariance system installed on a tower. Closure of the energy balance (Rn-G=LE+H) depends upon the validity of the eddy covariance measurements and the ability to quantify the available energy within the flux source area (Wilson et al., 2001). Many factors easily could account for the apparent lack of energy closure observed. Despite these deficiencies in energy balance closure, we were concluded that the eddy covariance measurements could accurately estimated ET. We accepted the energy balance closure results for the research area over the one month measurement period (Fig. 5).

The eddy covariance and sap flow estimates showed differences in their mean daily fluctuations. The highest ET measured by eddy covariance was in August and the lowest was in April. This trend was different from the one obtained using the sap flow and isotope techniques (Fig. 6).

3.3.2. Evapotranspiration determined by the P-M model

We calculated daily ET based upon the Penman-Monteith equation, and the daily mean ET over the seven month study period is shown in Figure 7. As observed previously, ET was lowest from the May to August. The variability trend was similar to that calculated using sap flow method. The mean daily CV calculated using the P-M equation was larger than that calculated using other methods. This result suggests that daily fluctuations are more obvious when using the P-M method compared to other techniques and that the results calculated using the P-M equation is not unfounded.

4. Discussion

Partitioning total ET into its components at the ecosystem level or larger requires integrating several measurement techniques. Often, these techniques require measurements that differ greatly in spatial or temporal scale (Grelle et al., 1997; Wilson et al., 2001). The total evapotranspiration description contains vegetation transpiration, soil evaporation, and rainfall evaporation (that is intercepted by the forest canopy). In this study, we measured rainfall interception as a separate component. The sum of these components (vegetation transpiration, soil evaporation, and evaporation of rainfall intercepted) was calculated for comparison with other methods.

We combined these methods (i.e. eddy covariance, the P-M equation, and the sum of the components) to determine the total evapotranspiration and its components (Fig. 8, 9). The results of eddy covariance method and the summation of transpiration components method showed a strong positive correlation. The variation trend can be described by a single peak curve with the largest variability occurring in August. The highest transpiration components sum was greater than the eddy covariance result, especially during July and August. Several reasons could explain these phenomena. On the one hand, evaporation of intercepted rainfall increased with elevations in precipitation, thereby causing the results of two methods to vary significantly (Fig. 9). On the other hand, the abundant rainfall may have reduced the sensitivity of the sensing probe. We observed a lack of complete energy balance closure from the eddy covariance measurements, which indicates a biased error in evaporation measurements. Such deficiencies in monitoring protocols could lead to an error in evaporation calculations.

According to the energy balance closure results and the strict data quality control, the results calculated using the eddy covariance method was reliable, with the exception of periods of abundant rainfall (July and August). The summation of transpiration components method produced the same results as the eddy covariance method. Thus, we propose that the two methods may be used for calculating the overall transpiration in the research area.

According to our data, soil evaporation, which was small in the abundant vegetation covered area, explains only a small portion of total evapotranspiration. In the dry season (April), sap flow flux changed with intercepted rainfall evaporation, and accounted for 32% of the total evapotranspiration. The percentage of intercepted rainfall evaporation increased during the wet season as rainfall increased, while the sap flow flux decreased.

The P-M equation yielded the lowest potential evaporation, and also exhibited a different trend in variability. The peak value of potential evaporation was during April, while the peak value of actual evaporation occurred in August. The potential evaporation was larger than the actual evaporation in the dry season, and was smaller in the wet season. This was likely due to the fact that the elevated moisture content provided abundant water input for evaporation. Independent studies observed similar phenomena in a remnant forest (Wilson et al., 2001; Zeppel et al., 2008).

The total evapotranspiration in the growing season calculated using the eddy covariance technique was 598 mm, which accounts for 44% of rainfall during the same period. The total evapotranspiration estimated by summing components was 747 mm, which accounts for 55% of rainfall. Lastly, and the potential evaporation calculated using the P-M equation was 427.58 mm, which accounts for 31%. We used the mean value of the eddy covariance and summed components techniques as the total growing season evapotranspiration; and we found that the percentage of evapotranspiration, accounting for water input, was 50%.

5. Conclusions

The isotope approach, when combined with sap flow estimates of total evapotranspiration, can be a very useful alternative for calculating soil evaporation. The summed components and eddy covariance estimates are qualitatively similar during the growing season. The monitoring methods of evaporation components have unique advantages for addressing physiological responses and probing heterogeneous environments (Wilson et al., 2001). In our research, the estimated results of the summed components method were higher than those obtained using the eddy covariance method. There may be errors associated with sampling measurements, such as scaling single tress estimates, steam water sampling, measuring eddy covariance during times of abundant rainfall.

The energy balance closure was accepted, suggesting that the estimated results were reliable. We used the mean values of the eddy covariance and summed components techniques as the total evapotranspiration during the growing season; and, accounting for water input, the percentage of evapotranspiration was 50%.

The methods mentioned above were simple and easy to perform. Furthermore, our data demonstrate that these methods have proven availability at the research location.


This study was funded by the Natural Science Foundation of China (No. 40730634 and No. 40925002). The grant was awarded to Dr. Wang Genxu under the auspices of the "Hundred People" Project of the Chinese Academy of Sciences in support of some research work. We also thank Yang Cao and Li Wei for data support.