Action 2: Measurement of ozone fluxes – Refinement of methods


Measurements of ozone fluxes

Measurements of ozone fluxes were carried out by the micrometeorological Eddy-covariance approach at forest canopy level (Figure 1) and are based on the analysis of turbulent motion of lower air layers in contact with vegetation. The vertical flux Fi of the species i in a turbulent medium are governed by a Fick-like diffusion law, which can be written as:

where K(z) is the turbulent diffusion coefficient, Ci the local concentration of the species i and z is the height above ground level. Contrary to the molecular diffusion coefficient, which is independent from space and time but depends on the nature of the transported gas, the turbulent diffusion coefficient depends on space and time but not on the nature of the gas.

Figure 1: Scheme of the ozone fluxes system implemented.

The micrometeorological method (Annex PR-36) is based on the calculation of the covariance of the vertical component w of the wind vector and of the concentration of the substance under consideration, both measured at very high sampling frequency (10Hz). The covariance is expressed by:

where the primed quantities represent the fluctuations around the time-averaged value of the corresponding variable and the overbar stands for the time-averaging process. The eddy covariance technique requires measurements of very rapid turbulent fluctuations and therefore the used sensors and analyzers should have a short response time combined with a high selectivity.


The Abscisic Acid dosage (ABA)

In order to estimate the stress state, leaf samples, strongly impacted by O3, were biochemically analyzed: dosages of ABA and chlorophylls (Annex IR-11). ABA plays an important role in plant responses to environmental stress and pathogens. Abscisic acid is also produced in the roots in response to decreased soil water potential and other situations in which the plant may be under stress.


Results and outputs of Action 2

The measurements and collected data of the Action 1 were used for the Action 2. Results were compared with foliar symptoms inventory, leaf-level stomatal conductance and variability of anti-stress markers in line with photochemical pollution as well as stomatal fluxes calculated with the DO3SE model. FO3REST provided an opportunity to infer an evaluation of the DO3SE model parameterisation through a comparison of measured (by Eddy-covariance) and modelled O3 deposition and to extend the number of DO3SE model evaluation studies conducted under “Mediterranean style” conditions for Mediterranean tree species. Thanks to FO3REST, we added 4 locations with Mediterranean tree species such as Pinus halepensis, Quercus ilex and Pinus pinea. The transferability for similar geographical areas was planned. The transfer function will allow applying and validating standards and critical levels for the forests protection.

Most experiments to establish biologically relevant plant responses have been performed under controlled conditions not representative of actual field conditions and the results may not provide realistic results for developing standards for protecting vegetation in natural environments. In comparison to other projects, the FO3REST project was directed toward a specific analysis of ozone symptoms and real damage levels in the field, trying to define more realistic thresholds for vegetation protection against ozone pollution.


Campaign at Castelporziano

The first study in Castelporziano allowed a comparison between calculated stomatal ozone fluxes and ozone fluxes measurement and, then, a first validation of the stomatal flux model DO3SE (Fares et al., 2013). Four months of continuous measurements highlighted the forest canopy as a relevant ozone sink, with total ozone fluxes up to 10 nmol m-2.s-1 during the central hours of the day. Ozone removal from forest ecosystems is attributed to both stomatal and non-stomatal sinks. In September 2011 we started an intensive field campaign, till December 2011, aimed at investigating ozone deposition from a mixed Mediterranean forest, mainly composed by Quercus suber, Quercus ilex and Pinus pinea. Measurements at canopy level with the eddy covariance technique were supported by a vegetation survey and the measurement of all environmental parameters which allowed calculating stomatal ozone fluxes. Leaf-level measurements were used to parameterize models to calculate stomatal conductance based on a Jarvis-type and Ball-Berry approach. We showed changes in magnitude of ozone fluxes from a warm (September) to a cold period (October-December). Stomatal component explained almost the totality of ozone fluxes during the cold days, but contributed only up to 50% to total ozone deposition during warm days, suggesting that other sinks (e.g. chemistry in the gas-phase) play a major role.


Devices set-up

Hourly values of air temperature were recorded with a MP100A sensor (Rototronic). An anemometer was used to instantaneously measure wind speed and directions, and a pluviometer was used to measure daily precipitation. All meteorological values were recorded using a data logger (Campbell scientifics). Flux measurements at canopy level started in September 1, 2011 and ended in December 11, 2011. The experimental facility was composed of an air conditioned cabin, where closed-path analytical equipment was housed, and a measuring tower 35-m tall, managed by the Castelporziano staff. Air was sampled continuously at one inlet at 35 m height through Teflon tubes with 4 mm internal diameter and a Teflon filter located 30 cm from the inlet. Ozone was measured with a UV absorption monitor (2B Technology). The filters were replaced every two weeks to avoid contamination or flow problems. Fast response measurements of ozone were made through a Teflon tube (35 m) and filter inlet by chemiluminescence using. The raw analog data were recorded at 10 Hz for all gases. Stomatal conductance from the eddy covariance measured evapotranspiration (Gsto) was calculated using the Monteith equation also called Evaporative/Resistance method extensively discussed in previous research. Stomatal ozone fluxes were calculated by multiplying Gsto by ozone concentration at canopy level calculated from ozone concentration at measuring height and accounting for atmospheric resistances assuming that intercellular concentration of ozone is zero.

Tower for the ozone fluxes measurements at Castelporziano.


Stomatal ozone fluxes

Ozone fluxes peaked during the central hours of the day, up to 9 nmol.m-2.s-1, during warm days and half of that during cold days. This typical bell-shape dynamics of ozone fluxes is a result of different concurring drivers which are maximized during the central hours of the day: temperature, ozone concentrations, photochemistry and emission of BVOC (Biogenic Volatile Organic Compounds). There is a mismatch (Figure 3) between the hour of day where ozone concentrations is maximum (13h-15h) and hours when ozone fluxes are maximum (10h-14h).

This can be explained by stomatal and non-stomatal sinks having different dependencies on environmental variables. The non-stomatal ozone removal processes depends on surface deposition and gas-phase chemical reaction with BVOC.

Figure2: 30-Min average of air temperature, VPD, ozone concentration, stomatal and total ozone fluxes measured at canopy level for warm (DOY 245 to DOY 280) and cold days (DOY 281 to DOY 345).

Despite this significant non-stomatal ozone sink, ozone concentration measured at two different heights did not vary significantly (data not shown) during the day hours when vertical mixing was high. This suggests that non-stomatal ozone removal may not significantly affect ozone concentration at leaf level. Stomatal ozone fluxes (Figure 2) were of similar magnitude during warm and cold days, reaching values up to 4 nmol.m-2.s-1, about 50% of the total ozone fluxes during the warm days, and almost the totality of ozone fluxes during the coldest days, supporting again the conclusion that temperature-dependent gas-phase chemical reactions can significantly contribute to ozone fluxes in warm days. During these days, stomatal fluxes peaked in the morning around 10 am, before light and VPD inhibited stomatal aperture and reduced water loss from leaves, as previously described for ecosystems of warm climates.

A second goal of the study was to parameterize the DO3SE model, used for ozone-risk assessment and test the model performance against the measured values with Eddy Covariance. For this purpose, a series of ecophysiological parameters have been collected at leaf level in the field during the measurement period using portable analyzers. For each species, we obtained the necessary parameters to model stomatal conductance as a function of vapor pressure deficit, temperature and light. Figure 3 shows hourly dynamics of observed and modeled stomatal conductance expressed per ground area, also using standard parameters for Holm Oak and Generic Evergreen forests published in the UNECE mapping manuals for Holm Oak and Generic Evergreen forests.

Figure3: 30-Min average of observed stomatal conductance (Gsto), and modeled using DO3SE (GstoJ) for the entire measuring period (DOY 245 to DOY 345) with the parameters calculated in this study, and reported in the UNECE mapping manual using standard values for evergreen and Holm Oak Mediterranean forests.

We found good agreement between the observed and modeled stomatal conductance. For the all measuring period, a linear correlation between 14h and 17h showed a R2= 0.40 while for generic evergreen we obtained values of R2= 0.40 and for Holm Oak values of R2= 0.36. We conclude that DO3SE model, if not properly parameterized, can produce a large bias up to 14-17%. Differences between warm and cold days are highlighted when we compare the observed and modeled stomatal conductance. For the entire period, the slope of the linear correlation between the observed and modeled stomatal conductance in the day hours (10h-17h) was excellent (0.99), with a R2 of 0.32. During warm days, the slope of the linear correlation is excellent (0.99), although the R2 is very poor (0.10) due to the overestimation of DO3SE during the early afternoon. The modeled stomatal conductance during these afternoon hours was predicted to be higher than observed, showing a recovery after a mid-day depression which did not occur in reality in our measuring site.

Comparisons of measured and modelled stomatal conductance showed that modelled values underestimated the stomatal conductance of about 10 %.



Campaign at San Rossore

Measurements were carried out during 4 seasons in order to capture eco-physiological responses to seasonal changes in meteorology and stomatal fluxes were calculated using the Monteith evaporative/resistive methods. As an example, Figures 4 and 5 shows the data obtained in spring and summer 2013. In spring, the fluxes are similar, which implies an artifact.

Figure4: Daily trends of total (in black) and stomatal (in red) fluxes in San Rossore.

Figure 5 shows the stomatal conductance obtained from two different parameterizations used in the stomatal flux-based model, according to the ICP Manual UNECE (2010): Aleppo pine and Mediterranean evergreen species. The parameterization developed for Aleppo pine (in red) works well than the parameterization for Mediterranean evergreen species (in green) for Pinus pinea in San Rossore. Measured stomatal ozone fluxes are in black in the first graph.

Figure5: Stomatal conductance in spring at San Rossore.

A new parameterization was carried out for two representative tree species in Mediterranean climate, i.e. Quercus ilex and Pinus pinaster. The model estimations of Gs by new parameterization were good agreement with observations in both two species gathering all data measured during the 4 seasons. Soil moisture deficit is recognized as a major limiting factor of stomatal conductance in Mediterranean climate. However, contrary to our expectations, the model performance of Holm oak did not improve if we would consider the fSW in the model (R2 = 0.43 without fSW, and R2 = 0.40 with fSW). On the other hand, the model performance of Stone pine dramatically improved when considering fSW (R2 = 0.35 without fSW, and R2 = 0.47 with fSW). In general, Holm oak has a deep root system while Stone pine is a relatively shallow root species. Holm oak may have drought tolerant mechanisms, leading to maintenance of relatively high stomatal conductance during water-stressed conditions. On the other hand, Stone pine need to reduce drastically water use during drought by stomatal closure due to shallow-root systems.

Figure6: A comparison between model estimations and observations of Gs for
Holm oak and Stone pine gathering all campaigns.

The new development of the parameters of DO3SE model can also reflect such a difference of ecological and physiological characteristics of stomatal response between both species. The results allow us establishing the “biologically relevant” risk assessment of O3 impacts for Mediterranean forests.



Campaigns at Fontblanche and Puéchabon

At Fontblanche and Puéchabon, models were run using parameterization values obtained from the field measurements and also using parameterization reported in the Mapping Manual.

At Fontblanche, measurements were carried out from 2 to 22 of July 2013. Ozone fluxes peaked during the central hours of the day, reaching values up to 21 nmol.m-2s-1 on 20 July at mid-day. The mean ozone concentrations were 49.3 ppb during the campaign. At Puéchabon, measurements were carried out in August-September 2013. Ozone reached 14.6 nmol.m-2s-1 at 2pm. At Puéchabon, we used the same DO3SE model parameterization than Castelporziano and we applied a new parameterization for Pinus halepensis.

In order to parameterize the DO3SE model, used for ozone-risk assessment, and test the model performance against the measured values with Eddy Covariance, the needed parameters (e.g. temperature, soil type, soil water content) were obtained to model the stomatal ozone fluxes. At Fontblanche and Puéchabon, we found a good agreement between the observed and modeled stomatal ozone fluxes (R2 = 0.44 and R2 = 0.50 respectively) over the measuring period. A good agreement between the stomatal conductance estimated from DO3SE and canopy level measurements from Eddy-covariance was found. The model estimations of stomatal conductance and PODY by new parameterization were good agreement with observations in both two species.


The Abscisic Acid dosage

On tree species strongly impacted by ozone, biochemical analyses were carried out: dosages of ABA at 4 plots in 2012. In 2012, the sampling were carried out in September in four sites with Pinus sylvestris, Fagus sylvatica, Pinus halepensis, Quercus ilex, Pinus cembra by GIEFS and the ABA analyses are made by external assistance. We realized a comparison (Figure 7) of the ABA concentration for symptomatic (S) and healthy samples (AS) and depending of the age (one-year-old needles and two-year-old). Arolla pine and Scots pine accumulate much ABA as beech, holm oak and Aleppo pine. By comparison of the ABA concentration in needles, a higher ABA concentration is observed in healthy needles relative to symptomatic needles

Figure7: Results of ABA dosage or symptomatic (S) and healthy (AS) samples.

This conclusion is surprising. Indeed, the concentration of ABA in needles of injured trees did not increase when the trees were exposed to ozone under field conditions. However, there is a great data variability, which results in significant deviations in absolute value or in % relative to the average. Therefore, there was no significant difference between samples with or without symptoms due to ozone. Therefore, the role of ABA in plant responses to environmental stress is unclear. Following these results, some aberrations appeared (symptomatic needles with lower ABA concentrations). The accumulation of ABA in leaves of the species analyzed thus did not appear related to their reactivity with ozone. The method is not conclusive and ABA dosage did not appear as a useful and relevant method to correlate ABA concentrations with stomatal closure and visible symptoms occurrence. ABA-induced closure of the stomata would result in decreased ozone uptake. This finding has led to the abandonment of these measures for the year 2013.



Conclusions and recommendations

Ground-level ozone is an important atmospheric pollutant, a pressing sanitary problem for human and ecosystem health, and a serious climate forcer, but the quantification of its effects on real-world forests is challenging. Although the effects depend on the amount of ozone entering through the stomata rather than the amount of ozone in the air, the majority of previous epidemiological assessments used ambient ozone exposure as a metric of injury.

The forest acted as an ozone sink during both periods; during warm and dry days in particular, non-stomatal ozone sink played a major role, suggesting that surface deposition processes and chemical ozone reaction with reactive BVOC may take place. The flux dataset offered the opportunity to test model for stomatal conductance, a very important effort in light of evaluating metrics for ozone-risk assessment based on accumulated ozone fluxes over a threshold, in the context of LRTAP. A Jarvis approach conventionally used with the DO3SE model parameterized with environmental observations was able to predict well stomatal conductance although it did not account for VPD-driven depression of stomatal conductance in the afternoon hours.

A different approach based on Ball-Berry equations underestimated stomatal conductance but predicted well the dynamics of stomatal conductance during the central hours of the day. We conclude that Jarvis and Ball-Berry models, if properly parameterized, can predict well stomatal conductance, even in a complex and mixed ecosystem, and the choice of using one or another method depends on the available data.

We reported novel results from cross-comparing of soil, meteorological, ozone and plant databases for Italian and French forests, where stomatal ozone fluxes were investigated. Large-scale epidemiological investigations, where large-scale biological responses (e.g. growth, ozone visible injury, crown transparency) are compared with ambient data in the field, provide useful information for establishing the best standards and thresholds for protecting plants from ozone. Epidemiological studies help to fill the gap between field conditions and lab studies.


The new parameterizations for representative tree species in Mediterranean climate, i.e. Quercus ilex, Pinus halepensis, Pinus pinea and Pinus pinaster are defined. The DO3SE parameterization will be included in the new version of the UNECE mapping manual for ozone-risk assessment.