Assessment of the total, stomatal, cuticular, and soil 2 year ozone budgets of an agricultural field with winter wheat and maize crops

Authors


Corresponding author: P. Stella, Biogeochemistry Department, Max Planck Institute for Chemistry, PO Box 3060, DE‒55020 Mainz, Germany. (patrick.stella@mpic.de)

Abstract

[1] This study evaluates ozone (O3) deposition to an agricultural field over a period of 2 years. A two-layer soil-vegetation-atmosphere-transfer (Surfatm-O3) model is used to partition the O3 flux between the soil, the cuticular, and the stomatal pathways. The comparison between measured and modeled O3 fluxes exhibited a good agreement, independently of the canopy structure and coverage and the climatic conditions, which implicitly validates the O3 flux partitioning. The total, soil, cuticular, and stomatal O3 budgets are then established from the modeling. Total ecosystem O3 deposition over the 2 year period was 87.5 kg ha−1. Clearly, nonstomatal deposition dominates the deposition budget, especially the soil component which represented up to 50% of the total deposition. Nevertheless, the physiological and phenological differences of maize and winter wheat induced large difference in the stomatal deposition budgets of these two crops. Then, the effect of simplified parameterizations for soil and cuticular resistances currently used in other models on the O3 budget is tested. Independently, these simplified parameterizations cause an underestimation of the O3 deposition ranging between 0% and 11.2%. However, the combination of all simplifications resulted in an underestimation of the total O3 deposition by about 20%. Finally, crop yield loss was estimated to be 1.5–4.2% for the winter wheat, whereas maize was not affected by O3.

1 Introduction

[2] Tropospheric ozone (O3) is a well-known greenhouse gas contributing to global warming. Its radiative forcing is estimated to account for 0.35 ± 0.15 W m-2, which represents 25% of the net radiative forcing attributed to human activities since the beginning of the industrial era. O3 is the largest contributor to radiative forcing after long-lived trace gases (CO2, CH4, N2O, and halocarbons) [Forster et al., 2007]. It is also a widespread secondary pollutant: several authors have proven its deleterious effects on materials [e.g., Lee et al., 1996; Ahmad et al., 2000; Boyce et al., 2001] and human health [e.g., Rastogi et al., 1991; Ito et al., 2005; Hazucha and Lefohn, 2007]. In addition, O3 penetrates plant leaves through the stomata and initiates a range of metabolic changes resulting in a decrease of photosynthetic capacity, alteration of plant biomass and structure, stomatal closure, and acceleration of senescence [e.g., Karnosky et al., 2003; Ashmore et al., 2004; Paoletti, 2005; Felzer et al., 2007; Ainsworth, 2008; Booker et al., 2009; Fuhrer, 2009; Wittig et al., 2009]. Moreover, O3 alters epicuticular waxes on leaf surfaces, reducing the plant protection against environmental constraints (e.g., acid rain and pathogens) [Barnes et al., 1988; Barnes and Brown, 1990; Percy et al., 1992, 2009]. In natural environments, these damages may lead to biodiversity losses [Hillstrom and Lindroth, 2008; Payne et al., 2011], while in agroecosystems, these damages induce losses in crop yield that could be critical for food security and economy [Ode, 2006; Betzelberger et al., 2010; Avnery et al., 2011a, 2011b]. Finally, O3 could be responsible for an increase in global warming through a positive feedback mechanism, by which the carbon sink of terrestrial ecosystems may be reduced [Felzer et al., 2007; Sitch et al., 2007].

[3] Dry deposition to the Earth's surface is an important process governing the tropospheric O3 budget and the only net removal pathway of O3 from the atmosphere [Stevenson et al., 2006; Wild, 2007]. Total O3 deposition to terrestrial ecosystems is divided into nonstomatal (i.e., soil and cuticular pathways, and eventually chemical depletion above or within the canopy) and stomatal components [e.g., Zhang et al., 2002; Bassin et al., 2004; Fowler et al., 2009; Simspon et al., 2012]. The latter component is necessary for assessing the deleterious effect of O3 on plant productivity [Fuhrer, 2000; Ashmore et al., 2004; Matyssek et al., 2007; Mills et al., 2010], while the overall O3 deposition rate is required to assess the role of terrestrial ecosystems for the removal of O3 from the atmosphere [Stevenson et al., 2006; Wild, 2007]. In order to estimate the trade-off between ecosystem productivity and its role as a sink for O3, and eventually propose mitigation options, it is thus necessary (i) to estimate the O3 stomatal flux and also (ii) to predict the total O3 deposition (which requires accurately estimating soil and cuticular components).

[4] These goals can be achieved by combining continuous O3 flux measurements and modeling approaches. Several authors have attempted to quantify the total O3 deposition over terrestrial ecosystems and its partitioning between stomatal and nonstomatal pathways. A strong focus was made on forest ecosystems [e.g., Emberson et al., 2000a; Zeller and Nikolov, 2000; Tuovinen et al., 2004; Wieser and Emberson, 2004; Gerosa et al., 2005; Fares et al., 2010; Zapletal et al., 2011]. However, only a few studies were performed to quantify total, stomatal, and nonstomatal budgets for crops, or existing studies only focused on one component of the total deposition (typically the stomatal flux), often only for a short period [e.g., Bassin et al., 2004; Gerosa et al., 2004; Coyle et al., 2009].

[5] In Europe, agricultural lands represent the second largest fraction of the terrestrial surface (23.1%) after forested areas (37.1%). However, in some regions such as eastern England, some regions of Germany, or northern France, agricultural lands cover up to 50% of the terrestrial surface. Crop production is dominated by cereals, mainly wheat, maize, and barley [Eurostat, 2011]. Consequently, predicting and partitioning O3 deposition to agroecosystems are major requirements to estimate O3 removal from the atmosphere and to assess the potential influence of O3 deposition on crop yield.

[6] The Surfatm-O3 model was developed to predict O3 deposition to agroecosystems from sowing to harvest including deposition through each pathway (i.e., soil, cuticle, and stomata). While the model was previously validated for three maize crops in different regions in France [Stella et al., 2011a], this study completes the validation of the Surfatm-O3 model for winter wheat. In addition, the O3 partitioning and 2 year budgets are established for an agricultural field with maize and winter wheat. Finally, the parameterizations concerning the nonstomatal deposition used in Surfatm-O3 are compared with those currently used in chemistry transport models (CTMs) and discussed regarding the assessment of establishing the dry deposition component of the tropospheric O3 budget.

2 Materials and Methods

2.1 Site Description and Data Sets

[7] The experimental site is located at Grignon (48°51′N, 1°58′E) about 40 km west of Paris, France. This site is a part of the CarboEurope and NitroEurope measurement networks, and its description as well as the measurement setup is already well documented [e.g., Laville et al., 2009; Lamaud et al., 2009; Loubet et al., 2011, 2013; Stella et al., 2012].

[8] The Grignon site is a 19 ha field surrounded by roads on the east, south, and SW. The soil is a silt loam formed by 31% clay, 62.5% silt, and 6.5% sand. The parcel is a winter barley-mustard-maize-winter wheat-winter triticale rotation. The data used in this study were collected during the years 2008 and 2009.

[9] Phenological parameters, i.e., crop height and leaf index area (LAI), are shown in Figure 1f. During early 2008, the field was occupied by mustard crushed on 14 April 2008. Its canopy height and LAI reached 0.8 m and 2 m2 m−2, respectively. The field was sown on 28 April 2008 with maize (NK Perform) used for silage. The maize crop reached a maximal height of 2.2 m on 29 June 2008. The maximal LAI of the green leaves was 5.2 m2 m−2 and that of the yellow leaves was 3.6 m2 m−2. The maize was harvested on 9 September 2008. In late 2008, the field was sown with winter wheat (Premio) on 17 October 2008. The first winter wheat leaves emerged on 25 October 2008, and canopy height and leaf area index weakly increased until March 2009, when they rapidly increased to reach their maximum of 0.85 m and 5.4 m2 m−2, respectively. The first yellow leaves appeared around 4 May 2009. The wheat was harvested on 31 July 2009, when all the leaves were yellow. On 4 October 2009, winter triticale was sown, emerged on 16 October 2009, and progressively grew until the end of the year (Figure 1f).

Figure 1.

(left) Daily averages and (right) half hourly medians (solid line) and 10th and 90th percentiles (dashed lines) of (a) global radiation, (b) air temperature, (c) air relative humidity, (d) ozone mixing ratio, and (e) ozone flux measured at the Grignon site during 2008 and 2009.(f) Time series of canopy height (measured point wise and interpolated) and leaf area index for green and yellow leaves (modeled with the CERES-EGC model [Gabrielle et al., 2006; Lehuger et al., 2010]).

2.2 Measurements of Meteorological Variables and Fluxes

[10] During the experimental period, standard meteorological variables were measured: incident and reflected solar radiations (CM7B, Kipp & Zonen, Netherlands), net radiation (NR-Lite, Kipp & Zonen, Netherlands), wind speed (cup anemometer, Cimel, France) and direction (W200P, Campbell Sci. Inc., USA), air temperature and relative humidity (HMP-45, Vaisala, Finland), and precipitation (ARG100, Campbell Sci. Inc., USA). Soil temperature (homemade thermocouple probes, INRA, France) and soil water content (TDR CS 616, Campbell Sci. Inc., USA) profiles were also measured at 5, 10, 20, 30, 70, and 90 cm and at 5, 10, 20, 30, 50, and 90 cm, respectively.

[11] Turbulent fluxes of momentum, sensible heat, water vapor, CO2, and O3 were measured by eddy covariance at a 3.2 m height. The fetch was limited to around 200 m in the southeast direction but extended to up to 400 m in all other directions. A footprint analysis was performed showing that more than 60% of the field was in the footprint during the winter and more than 93% during the spring and summer [Loubet et al., 2011]. The measurement system included a 3-D sonic anemometer (R3, Gill Inc., UK) and an open-path infrared absorption spectrometer for water vapor and CO2 (IRGA 7500, LiCor, USA). O3 fluxes were determined according to the method by Muller et al. [2010], providing measurements of the deposition velocity (i.e., the flux divided by the mean concentration) using a fast-response O3 chemiluminescence analyzer (ATDD, NOAA, USA). The O3 flux was obtained by multiplying the deposition velocity with the absolute O3 concentration measured by a slow-response sensor (O341M, Environnement SA, France) at a 3.2 m height. The data were sampled and recorded at 50 Hz, and flux integration was performed over 30 min time intervals. Flux calculations from eddy covariance were made following the CarboEurope methodology [Aubinet et al., 2000]. Flux processing included despiking of scalar time series, one sector planar fit to correct for the sonic anemometer tilt, linear detrending of concentrations, correction of the time lag induced by the tube for the fast O3 sensor, and corrections for density fluctuations (WPL correction [Webb et al., 1980]) for CO2 and water vapor fluxes. In addition, high frequency loss due to sensor separation was corrected for CO2, water vapor, and O3 using the empirical ogive method described by Ammann et al. [2006].

2.3 Modeling and Partitioning of Ozone Fluxes

[12] The O3 fluxes were modeled using the Surfatm-O3 model [Stella et al., 2011a]. It is a one-dimensional soil-vegetation-atmosphere-transfer model including one vegetation layer and one soil compartment. It combines (i) an energy budget model accounting for the latent and sensible heat fluxes, (ii) a soil water balance model, and (iii) a pollutant exchange model simulating O3 fluxes between the surface and the atmosphere. The resistive scheme is the same for heat, water vapor, and O3 (i.e., identical aerodynamic resistances above and within the canopy, soil and leaf quasi-laminar boundary layer resistances, and stomatal resistances, although the latter two are modified by gas diffusivities). The model was extensively described in detail in previous studies [Personne et al., 2009; Stella et al., 2011a, 2013]. Hence, only the O3 deposition model and its specific resistances will be described in the following.

[13] The model includes four deposition pathways for O3: stomatal for green leaves (able to photosynthesize), stomatal for yellow leaves (not able to photosynthesize but transpiring), cuticular, and soil. For yellow leaves, it is assumed that the exchanges through the stomata are still under stomatal control [Stella et al., 2011a]. The chemical sinks or sources due to reactions above or within the canopy with other compounds such as nitric oxide, nitrogen dioxide, or volatile organic compounds are not accounted for. However, the soil NO emission at the Grignon site is quite weak and ranges between 0 and 2 nmol m−2 s−1, despite few exceptional NO emission peaks that could reach 5 nmol m−2 s−1 [Laville et al., 2009, 2011; Loubet et al., 2011]. In comparison, the O3 flux ranged between 0 and 14 nmol m−2 s−1 (Figure 1e). If we suppose that all NO emitted from soil reacts with O3 (the worst case since it was evaluated in Stella et al. [2011b] that the transport time was faster than the chemical reaction time), that would overestimate the O3 flux by a maximum of 15%. Noting that on average over 3 years, the NO emission is 0.16 nmol m−2 s−1 [Loubet et al., 2011] while the O3 deposition is 2.9 nmol m−2 s−1 over 2 years (see section 3.1), chemical reactions may only represent around 5% of the O3 flux. Finally, it must be considered that advected NO can react with O3 in the canopy. This effect would only be important for tall canopies (with enough large transport time). Lamaud et al. [2009] observed over maize at the same site a nonnegligible effect only when NO concentrations were larger than 1 ppb. But Loubet et al. [2013] reported median NO concentrations of 0.1 ppb with episodic peaks mainly when wind is low at night. Thus, the chemical sink due to the reaction between O3 and NO is negligible in Grignon.

[14] The model has been validated for three different maize crops in France, including those addressed in this study [Stella et al., 2011a]. It was also compared with O3 flux measurements at a meadow in southern Germany [Stella et al., 2013]. The O3 soil deposition pathway was parameterized using results derived at the Grignon site [Stella et al., 2011b].

[15] The O3 leaf stomatal resistance (inline image) is modeled with a Jarvis approach as described in Stella et al. [2011a]:

display math(1)

where gmax (in mmol O3 m−2 s−1) is the maximum leaf stomatal conductance; flight, fT, fVPD, and fSWP represent the response of gmax to radiation, temperature, vapor pressure deficit, and soil water potential, respectively; fmin is the minimum stomatal conductance that occurs during the daylight period; and 41,000 is the factor to convert mmol m−2 s−1 to m s−1 [Jones, 1992]. The generic functions fPAR, fT, fVPD, fSWP, and fmin are plant species specific. The parameterizations for each function are presented in Table 1 using coefficients from Emberson et al. [2000b] and Loubet et al. [2012]. Radiation, temperature, and vapor pressure deficit were deduced from meteorological measurements, while the soil water potential was modeled with Surfatm-O3. The upscaling from leaf to canopy is performed using the effective leaf area index (LAIe) as proposed by Rochette et al. [1991], integrating a shielding factor σc (= 0.25; LAIe = σc × LAI). In addition, in the present model, we distinguish stomatal resistance for green leaves (able to photosynthesize) (inline image) and yellow leaves (not able to photosynthesize, but transpiring) (inline image):

display math(2a)
display math(2b)

where inline image is the effective green leaf area index and inline image is the effective yellow leaf area index obtained by the difference between the maximum leaf area index of the cropping season and green leaf area index. Leaf area index was modeled with the CERES-EGC model and validated against measured LAI [Gabrielle et al., 2006; Lehuger et al., 2010].

Table 1. Plant Species-Dependent Parameters for Stomatal Resistance Calculationsa
 gmax (mmol O3 m−2 s−1)fminflightfTfVPDfSWP
   αTmin (°C)Topt (°C)Tmax (°C)VPDmin (kPa)VPDmax (kPa)SWPmin (MPa)SWPmax (MPa)
  1. a

    Values of gmax are given in mmol (O3) m−2 s−1. Note that parameters are identical for winter wheat, winter triticale, and mustard.

Maize1560.190.0048025512.51−0.11−0.8
Winter wheat2960.10.009026402.80.9−0.3−1.1
Winter tritical Mustard

[16] Nonstomatal (i.e., soil and cuticular) resistances for O3 were identical whatever the canopy. Soil resistance (inline image) was parameterized as a function of relative humidity at the soil surface (RHsurf) [Stella et al., 2011b], while cuticular resistance (inline image) was a function of LAI and relative humidity at the leaf surface (RHleaf) [Lamaud et al., 2009; Stella et al., 2011a]:

display math(3)
display math(4a)
display math(4b)

where inline image (= 21.15 s m−1) is the soil resistance without water adsorbed at the soil surface (i.e., at RHsurf = 0%), ksoil (= 0.024) is an empirical coefficient of the exponential function, inline image (= 5000/LAI) is the maximal cuticular resistance [Massman, 2004], RH0 (= 60%) is a threshold value of the relative humidity, and kcut (= 0.045) is an empirical coefficient of the exponential function.

3 Results

3.1 Overview on Meteorological Conditions and Fluxes

[17] Mean daily global radiation (Rg) and air temperature (Ta) increased during spring to reach a maximum in summer (200 W m−2 and 20°C, respectively) and decreased in autumn to reach their minimum in winter (less than 50 W m−2 and 0°C, respectively) (Figures 1a and 1b). In contrast, air relative humidity (RH) was higher during winter (around 90%) than during summer when it decreased to 50–60% (Figure 1c). The diurnal cycles of Rg showed increasing values during early morning, reaching maxima at noon, and decreased during the afternoon. However, the diel course of Rg was variable: the median half hourly maximum was around 400 W m−2 but varied between 100 and 800 W m−2 (10th and 90th percentiles) (Figure 1a), reflecting the seasonal variability of Rg (lower in winter and higher in summer). Similarly, Ta was lower during nighttime and increased during the morning until its maximum in the early afternoon. Median half hourly Ta was around 10°C during nighttime (10th and 90th percentiles around 0°C and 15°C) and around 14°C during early afternoon (10th and 90th percentiles around 5°C and 25°C) (Figure 1b). RH was highest during nighttime (median around 90%, 10th and 90th percentiles around 95% and 80%) and progressively decreased to reach its minimum in the early afternoon (median around 60%, 10th and 90th percentiles around 90% and 40%) (Figure 1c).

[18] The O3 mixing ratio exhibited an annual cycle: it reached its maximum in spring and then progressively decreased to its minimum in winter. However, this annual cycle is more pronounced in 2008 than in 2009. In 2008, the mean daily O3 mixing ratio reached around 35 ppb in April-May and decreased to less than 5 ppb during the winter 2008–2009. In 2009, the mean daily O3 mixing ratio increased from January to March and ranged between 10 and 30 ppb throughout the year. The diurnal cycle of O3 featured its minimum in the early morning (median half hourly O3 mixing ratio around 12 ppb, 10th and 90th percentiles around 1 and 25 ppb), reached its maximal value in the early afternoon (median half hourly O3 mixing ratio around 28 ppb, 10th and 90th percentiles around 12 and 45 ppb), and then decreased (Figure 1d).

[19] O3 flux measurements were not available during the beginning of 2008 (from January to March) and occasionally during the rest of the experiment due to instrument breakdown. Despite these technical problems, O3 fluxes were measured during 73% of the time in 2008–2009. O3 fluxes were always directed downward, i.e., indicating deposition to the surface. The mean daily O3 flux was around −3 nmol m−2 s−1 in late March to early April 2008 when the field was occupied by mustard. It then reached −10 nmol m−2 s−1 during the bare soil period between the mustard and maize crops. When the maize was fully developed, the mean daily O3 flux varied between −4 and −7 nmol m−2 s−1. It decreased again after the maize harvest to its minimal value, around −1 nmol m−2 s−1, during winter. In 2009, the mean daily O3 flux increased in spring, was maximal around −6 nmol m−2 s−1 when the winter wheat was fully developed in May 2009, and decreased again to be around −1 nmol m−2 s−1 in December 2009. We observed a pronounced diel course of the O3 flux with increasing values during the morning, maxima in the early afternoon, and decreasing values to reach a minimum during nighttime. Nighttime O3 fluxes varied between 0 and −3 nmol m−2 s−1 (10th and 90th percentiles) with a median of around −1 nmol m−2 s−1, while early afternoon O3 fluxes ranged from −1 to −13 nmol m−2 s−1 (10th and 90th percentiles) with a median of −6 nmol m−2 s−1 (Figure 1e).

3.2 Comparison Between Measured and Modeled Fluxes

[20] The model validation for bare soil periods and the maize cropping period was already addressed in Stella et al. [2011a, 2011b], and further details can be found in these studies. For these two periods, modeled O3 fluxes agreed well with measurements (slope of the regression = 0.96 and 0.98 for maize cropping and bare soil periods, respectively), with a moderate error between measured and modeled fluxes (root mean square error (RMSE) = 1.62 and 2.02 nmol m−2 s−1 for maize cropping and bare soil periods, respectively) (Figures 2c and 2e).

Figure 2.

Comparison between measured and modeled O3 fluxes for (a) mustard (N = 582), (b) winter triticale (N = 3265), (c) maize (N = 3133), (d) winter wheat (N = 7845) cropping periods (i.e., from sowing to harvest), (e) bare soil periods (N = 3327), and (f) the whole 2008–2009 period at the Grignon site. Regression coefficient, coefficient of determination (R2), and root mean square error (RMSE) are shown on each plot. Only data for u* > 0.1 m s−1 are represented.

[21] The comparison between measured and modeled O3 fluxes for the winter wheat was performed for the whole cropping period (i.e., from sowing to harvest) for each development stage: the bare soil period from sowing to winter wheat emergence, the growth, and the senescence (Table 2). Concerning the bare soil period, the slope of the regression was 0.94, the coefficient of determination (R2) was 0.87, and the root mean square error (RMSE) was 1.06 nmol m−2 s−1. During the growth period, the modeled O3 fluxes exhibited also a good agreement with the measurements, the slope of the regression, R2, and RMSE being equal to 0.87, 0.86, and 1.19 nmol m−2 s−1, respectively. Concerning the senescence period, the modeled LE flux including transpiration from yellow leaves exhibited an overall good agreement with the measurements from sowing to early senescence of the winter wheat crop (data not shown). Due to the drying of the yellow leaves, we assume that the yellow leaves did not transpire after several weeks, and so the stomatal deposition pathway for yellow leaves was not maintained to model total O3 deposition in July 2009 for the winter wheat. During the senescence, there was an overall good agreement between measured and modeled O3 fluxes, with the slope of the regression being equal to 0.94. However, the scatter was larger than for the previous periods, indicated by smaller R2 (= 0.66) and larger RMSE (= 2.35 nmol m−2 s−1) values (Table 2). Over the whole winter wheat cropping period, measured O3 fluxes were well reproduced (R2 = 0.80; RMSE = 1.71 nmol m−2 s−1), despite a weak underestimation (slope of the regression = 0.91) (Figure 2d).

Table 2. Statistics of the Comparison Between Measured and Modeled O3 Fluxes for Winter Wheat From Sowing to Harvest at the Grignon Sitea
 Bare Soil Period (17 October 2008 to 24 October 2008)Growth (25 October 2008 to 3 May 2009)Senescence (4 May 2009 to 31 July 2009)
  1. a

    Only data for u* > 0.1 m s−1 were considered.

Linear regressiony = 0.94xY = 0.87xy = 0.94x
R20.870.860.66
RMSE1.06 nmol m−2 s−11.19 nmol m−2 s−12.35 nmol m−2 s−1
Number15147042990
of data
points

[22] During early 2008 and late 2009, when the field was occupied by mustard and winter triticale, respectively, the modeled fluxes were different from measurements. During the mustard period, O3 deposition was overestimated. The slope of the regression and the RMSE were equal to 1.40 and 2.64 nmol m−2 s−1, respectively (Figure 2a). Similarly, the model slightly overestimated O3 deposition fluxes during the early stage of winter triticale (slope of the regression = 1.14), but the error of the modeled fluxes was smaller (RMSE = 0.91 nmol m−2 s−1) (Figure 2b). Nevertheless, an accurate comparison between measured and modeled fluxes is difficult for these two periods due to (i) the short period considered (senescence of the mustard and early growth of the winter triticale) and (ii) the breakdown of the O3 sensor in early 2008.

[23] Considering the comparison between measured and modeled O3 fluxes over the whole 2008–2009 period, the slope of the regression was 0.96, while the R2 and the RMSE were equal to 0.79 and 1.68 nmol m−2 s−1, respectively (Figure 2f).

3.3 Ozone Flux Partitioning and Budgets

[24] The contribution of soil, cuticular, and stomatal deposition pathways to the daily total O3 deposition exhibited large seasonal variations (Figure 3a). The contribution of the soil pathway to the daily total O3 deposition progressively decreased with crop development. Daily soil deposition decreased from 100% before leaf emergence to around 20% and 10% when maize and winter wheat were fully developed. Conversely, the weight of cuticular and stomatal deposition increased with crop development. During crop maturity, cuticular deposition represented around 40% and 30% of the daily deposition for maize and winter wheat, respectively, while stomatal deposition to green leaves was around 40% and 60% for maize and winter wheat, respectively. During the senescence, stomatal deposition to yellow leaves increased and reached around 35% of the daily deposition for both maize and winter wheat. Nevertheless, this deposition pathway prevailed until the harvest of the maize, while it occurred only during the first half of the senescence for the winter wheat. During early 2008 and late 2009, when the field was occupied by mustard and winter triticale, soil deposition was the main O3 sink and accounted for up to 60% of the daily O3 deposition. The sum of cuticular and stomatal sinks varied between less than 10% and 30% of the daily deposited O3 (Figure 3a).

Figure 3.

(a) Daily contributions of soil, cuticular, and stomatal pathways for green and yellow leaves to total O3 deposition. (b) Measured (gap filled with the Surfatm-O3 model) and modeled cumulative total O3 deposition, and modeled cumulative O3 deposition components (soil, cuticular, and stomatal for green leaves and stomatal for yellow leaves) at the Grignon site.

[25] Over the cropping period (i.e., from sowing to harvest), maize and winter wheat crops were ozone sinks of 25.6 and 28.6 kg ha−1, respectively. This budget was dominated by soil deposition, 10.6 and 8.4 kg ha−1, which represented 43.1% and 28.8% of the total deposition during these periods for maize and winter wheat, respectively (Table 3). The stomatal deposition to green leaves was 5.4 and 10.5 kg ha−1 (i.e., 28% and 28.8% of the total O3 deposition, respectively), the cuticular deposition was 6.9 and 8.4 kg ha−1 (i.e., 22% and 36% of the total O3 deposition, respectively), and the stomatal deposition to yellow leaves was 1.7 and 1.9 kg ha−1 (i.e., 6.9% and 6.4% of the total O3 deposition, respectively) for maize and winter wheat crops, respectively. The stomatal sink represented 7.1 and 12.4 kg ha−1 (i.e., 28.9% and 42.4% of the total O3 deposition, respectively), whereas the nonstomatal sink accounted for 17.5 and 16.8 kg ha−1 (i.e., 71.1% and 57.6% of the total O3 deposition, respectively) over the whole cropping season for maize and winter wheat, respectively (Table 3). In early 2008 and late 2009, when the field was occupied by mustard and winter triticale, the total O3 deposition was 11.7 and 5 kg ha−1, respectively, and was dominated by the soil sink. Bare soil periods were an O3 sink accounting for 16.6 kg ha−1 (Table 3).

Table 3. Total, Soil, Cuticular, and Stomatal (for Green and Yellow Leaves) Ozone Deposition Budgets (in kg ha−1)a
 Maize (From Sowing to Harvest)Winter Wheat (From Sowing to Harvest)Bare Soil PeriodsMustardWinter Triticale200820092008–2009
  1. a

    The ozone deposition budgets were deduced from the measurements (gap filled with the Surfatm-O3 model in case of lack of data), from the Surfatm-O3 model, and from the modeled O3 fluxes using different parameterizations of the resistances to O3 deposition (see text for details).

  2. b

    Numbers in parentheses denote the relative difference of modeled fluxes to measurements for total deposition (negative values are underestimations and positive values are overestimations).

  3. c

    Numbers in parentheses denote the contributions of each deposition pathways to the total O3 deposition at the Grignon site.

  4. d

    Numbers in parentheses denote the relative difference between the total measured deposition and the total modeled deposition in case of the use of the other parameterizations.

Measurements (Measured and Gap Filled)
Total deposition25.628.616.611.75.046.541.087.5
Surfatm-O3
Total depositionb24.6 (−3.9%)29.2 (+2.1%)16.2 (−2.4%)12.7 (+8.5%)6.2 (+24.0%)46.1 (−0.9%)42.8 (+4.4%)88.9 (+1.6%)
Soil componentc10.6 (43.1%)8.4 (28.8%)16.2 (100%)8.4 (66.1%)5.5 (88.7%)27.5 (59.7%)21.5 (50.2%)49.0 (55.1%)
Cuticular componentc6.9 (28.0%)8.4 (28.8%)0 (0%)2.6 (20.5%)0.5 (8.1%)9.7 (21.0%)8.8 (20.6%)18.5 (20.8%)
Stomatal component for green leavesc5.4 (22.0%)10.5 (36.0%)0 (0%)1.6 (12.6%)0.2 (3.2%)7.1 (15.4%)10.6 (24.8%)17.7 (19.9%)
Stomatal component for yellow leavesc1.7 (6.9%)1.9 (6.4%)0 (0%)0.1 (0.8%)0 (0%)1.8 (3.9%)1.9 (4.4%)3.7 (4.2%)
Total Deposition Modeled Using Different Resistance Parameterizations
Rsoil = 200 s m−1d22.7 (−11.3%)29.1 (+1.7%)10.5 (−36.7%)12.9 (+10.3%)6.1 (+22%)42.3 (−9.0%)39.0 (−4.9%)81.3 (−7.1%)
Rcut = 5000/LAId21.5 (−16.0%)25.3 (−11.5%)16.2 (−2.4%)11.2 (−4.3%)5.9 (+18%)41.3 (−11.2%)38.7 (−5.6%)80.1 (−8.5%)
inline image = 9999 s m−1d21.5 (−16.0%)27.7 (−3.1%)16.2 (−2.4%)12.6 (+7.7%)6.2 (+24.0%)42.9 (−7.7%)41.3 (+0.7%)84.2 (−3.8%)
All effectsd17.3 (−32.4%)23.8 (−16.8%)10.5 (−36.7%)11.3 (−1.7%)5.7 (+14%)35.2 (−24.3%)33.4 (−18.5%)68.6 (−21.6%)

[26] On an annual basis, the measured total O3 deposition was 46.5 kg ha−1 in 2008 and 41 kg ha−1 in 2009. During these 2 years, soil O3 deposition dominated the annual budgets and represented up to 50% of the annual total O3 deposition, which means that 27.5 and 21.5 kg ha−1 of O3 were removed from the atmosphere by the soil pathway in 2008 and 2009, respectively. Cuticular deposition accounted for 9.7 and 8.8 kg ha−1 in 2008 and 2009, respectively, or around 20% of the annual total deposition. Stomatal deposition to green leaves was substantially lower in 2008 (7.1 kg ha−1, or 15.4% of the annual O3 deposition) than in 2009 (10.6 kg ha−1, or 24.8% of the annual O3 deposition), whereas stomatal deposition to yellow leaves was similar in 2008 and 2009, with 1.8 and 1.9 kg ha−1, respectively, which represented around 4% of the annual O3 deposition (Table 3).

[27] For the entire 2008–2009 period, 87.5 kg ha−1 was removed from the atmosphere by dry deposition to the Grignon agricultural field. Over this 2 year period, the soil sink was the dominant removal pathway for O3 and represented 55.1% of the total O3 deposition (49 kg ha−1). The cuticular pathway and stomatal deposition to green leaves were O3 sinks of similar magnitude, both accounting for about 20% of the total deposition (18.5 and 17.7 kg ha−1 for cuticular and stomatal deposition to green leaves, respectively). Stomatal deposition to yellow leaves was the smallest O3 sink, only 3.7 kg ha−1 (less than 5% of the total O3 deposition) (Figure 3b and Table 3).

4 Discussion

4.1 Total, Stomatal, Cuticular, and Soil Ozone Deposition

[28] The aim of this study was to predict the O3 deposition only using bibliographic data without calibration (shielding factor σc, soil resistance inline image, cuticular resistance inline image, etc.). The Surfatm-O3 model was able to predict the O3 fluxes during the whole experiment, independently of the climatic conditions and the canopy development stages. Obviously, the performances of the model are more reliable for the periods during which the vegetation is dense and without senescence, but, as already discussed in Stella et al. [2011a], having good results for a time series including various vegetation coverage indicates that (i) each deposition pathway is pretty well represented and (ii) the model can be used to estimate the individual contributions of stomatal, cuticular, and soil O3 deposition pathways (for more details, see Stella et al. [2011a]).

[29] The partitioning of the total O3 deposition into soil, cuticular, and stomatal components was highly dependent on crop phenology and physiology. Although soil deposition was the only O3 sink during bare soil periods, its contribution progressively decreased with crop development, when stomatal and cuticular deposition increased (Figure 3a). This pattern can be explained by the simultaneous increase of the canopy height and the leaf area index, decreasing soil deposition (due to a larger in-canopy aerodynamic resistance) and increasing cuticular and stomatal uptakes. In addition, physiological differences between winter wheat and maize explain the difference between the stomatal sink of the two crops, which was larger for winter wheat (Figure 3a and Table 3). Since the leaf stomatal conductance for wheat is 2 times higher than that for maize (see equation (1)) [Emberson et al., 2000b], the contribution of the stomatal pathway to the ecosystem flux is larger for winter wheat. Moreover, the longer growing season of winter wheat caused less O3 deposition to soil and a larger deposition to vegetation (Table 3).

[30] Most previous studies report a large and sometimes even dominant nonstomatal pathway contribution to the total O3 deposition flux, similar to our study. For instance Fares et al. [2010] reported that annual O3 deposition in a pine forest was dominated by nonstomatal processes, although its contribution was highly variable depending on the seasons, from 10% to around 70%. In addition, estimated nonstomatal deposition accounted for 50% of the ecosystem O3 deposition in a boreal Scots pine forest [Altimir et al., 2004]. Comparably, for an evergreen Mediterranean forest in Italy, stomatal uptake represented only about 30% of the ecosystem O3 flux [Gerosa et al., 2005; Vitale et al., 2005]. Nevertheless, only few studies focused on the partitioning of the nonstomatal pathway into soil and cuticular components. A study carried out over a spring wheat crop in Italy reported that stomatal, soil, and cuticular fluxes represented 40%, 34%, and 26% of the ecosystem O3 flux from anthesis to harvest [Bassin et al., 2004]. While this generally agrees well with our study concerning the fact that the nonstomatal pathway is dominating, for winter wheat at maturity, we found that stomatal, soil, and cuticular fluxes represented around 60%, 10%, and 30%, respectively, of the total ecosystem O3 flux (Figure 3a). The difference can be explained by a lower soil water content during the study carried out in Italy, which causes (i) a lower stomatal flux due to stomatal closure because of water stress, (ii) a larger soil flux due to the smaller soil resistance of dry soil [Stella et al., 2011b], and (iii) a lower cuticular flux due to an increased cuticular resistance with lower relative humidity [Lamaud et al., 2009]. This comparison reveals the weight of the meteorological conditions on the O3 flux partitioning. The combination of the prevailing environmental conditions with the established crop rotation, including longer periods without vegetation, caused the soil to constitute the major O3 sink for the agricultural ecosystem at the Grignon site (55.1% of the total deposition during 2008 and 2009).

[31] Previous studies rarely reported annual O3 deposition budgets derived based on field measurements. The tropospheric O3 budgets are estimated in studies using global CTMs, in which deposition fluxes to the terrestrial surface account for 530 to 1470 Tg yr−1 [Wild, 2007]. Considering that O3 is mainly deposited to terrestrial ecosystems (i) because it is hardly soluble in water [Fowler et al., 1999] and (ii) since deposition rates over oceans are very low [Gallagher et al., 2001a, 2001b], these deposition budgets would represent an annual O3 deposition ranging from 31 to 86 kg ha−1 yr−1 for continental surfaces, which agrees with our estimates.

4.2 Comparison With Other Resistance Parameterizations for O3 Deposition

[32] Since O3 is both a greenhouse gas and a secondary pollutant impacting ecosystems, the establishment of the trade-off between the beneficial effects of O3 deposition (atmospheric cleanup and lower greenhouse gas (GHG) load) and the deleterious effect to terrestrial ecosystems is necessary. Robust estimates of both ecosystem O3 budget and the amount of O3 responsible for damages to plants are needed to understand and predict the tropospheric O3 budget and the impact on ecosystems. This also includes an analysis of feedbacks on the atmospheric carbon dioxide budget. The Surfatm-O3 model was developed to achieve this goal, but several other models addressing these issues already exist, e.g., the DO3SE model (SVAT model embedded within the EMEP photochemical model) [Büker et al., 2012], the CHIMERE model (chemistry transport model including O3 dry deposition scheme) [Anav et al., 2012], PLATIN [Grünhage and Haenel, 1997, 2008], or ECHAM4 (chemistry and climate model including O3 dry deposition scheme) [Ganzeveld et al., 2002, 2006]. Although stomatal resistance is modeled using a Jarvis-based approach, these models substantially differ from Surfatm-O3 since the effect of relative humidity on soil and cuticular resistances is not included and stomatal deposition to senescent (yellow) leaves is ignored.

[33] In order to understand to what extent the exclusion of these effects in the other models may influence the total O3 deposition budget, four additional simulations were made with the Surfatm-O3 model. We modified soil, cuticular, and stomatal resistances for yellow leaves as follows: (i) inline image was set to 200 s m−1 (as recommended by Simpson et al. [2012]), (ii) the humidity effect on inline image was ignored (i.e., inline image = 5000/LAI), and (iii) the stomatal deposition to yellow leaves was turned off (i.e., inline image = 9999 s m−1). These modifications were taken into account separately and then simultaneously (Table 3). Assuming a constant inline image led to a large underestimation (36.7%) of the O3 deposition during bare soil periods. During the maize cropping season, it caused an underestimation of O3 deposition of 11.3%, but the effect during the winter wheat cropping period was negligible. During the maize and winter wheat cropping periods, the absence of the effect of relative humidity on inline image induced an underestimation of the total O3 deposition of 16% and 11.5%, respectively. The absence of stomatal deposition to yellow leaves substantially affected the total O3 budget during the maize cropping period (−16%), but not during the winter wheat cropping period (−3.1%). The effect of the modification of one of these three resistances on the annual and 2 year O3 deposition budgets varied between an overestimation of the measured deposition of 0.7% (with inline image = 9999 s m−1 in 2009) and an underestimation of 11.2% (with inline image = 5000/LAI in 2008). Nevertheless, modeling O3 deposition by excluding the effect of relative humidity on both inline image and inline image, as well as of stomatal deposition to yellow leaves, substantially underestimates the total O3 deposition, by about 20% for the maize and winter wheat during 2008 and 2009, and the 2 year budgets.

[34] In this study, the different possible choices of the stomatal resistance algorithm were not explored. These include (i) the multiplicative Jarvis approach (this study) or (ii) a semiempirical model based on photosynthesis, the so-called Ball-Berry approach. Unfortunately, the Ball-Berry algorithm cannot be used in the Surfatm-O3 model since photosynthesis modeling is not included yet. However, several studies already compared the Jarvis-based and Ball-Berry-based algorithms and reported contradictory results. Op de Beeck et al. [2010] found that the two algorithms were equivalent, while Misson et al. [2004] concluded that the Ball-Berry approach was better under drought stress conditions. Fares et al. [2013] further stated that the Jarvis approach was superior in predicting stomatal resistance. Büker et al. [2007] indicated that the Ball-Berry algorithm was less successful to predict stomatal resistance at the seasonal scale but was equivalent to the Jarvis algorithm at the hourly and daily scales. Thus, considering these results, we may expect that the differences in modeled O3 fluxes between the two approaches would be minor. Nevertheless, it must be noted that the Jarvis approach only requires standard meteorological conditions, while the Ball-Berry approach needs an estimate of the photosynthesis. Modeling this process introduces additional uncertainties in the modeled ozone fluxes, although these uncertainties may be constrained by measured CO2 fluxes [Bunce, 2000; Uddling et al., 2005]. To conclude, the Jarvis approach may be more robust and easier to use in modeling observed O3 fluxes, but the Ball-Berry approach should be preferred to account for O3 impacts on photosynthesis and study potential feedback effects between O3 and CO2.

4.3 Evaluation of Ozone Damage to Crops

[35] In order to estimate damages to crops due to O3, yield reductions for maize and winter wheat were calculated following the methodology recommended by the LRTAP Convention [2010]. Hence, the accumulated phytotoxic ozone dose (i.e., the accumulated stomatal flux) of O3 above a flux threshold of 6 nmol m−2 s−1 (POD6) was calculated for winter wheat. In addition, the AOT40 (O3 concentrations accumulated over a threshold of 40 ppb) was also calculated for winter wheat. Subsequent reduction in winter wheat crop yield was estimated following Pleijel et al. [2007] (for the POD6) and Mills et al. [2007] (for the AOT40). For maize, no studies are known to us that evaluate O3 damages on crop yield using the accumulated stomatal flux approach. Consequently, the reduction in maize crop yield was estimated following Mills et al. [2007] from the exposure approach by calculating the AOT40.

[36] According to our estimates, crop yield losses due to O3 ranged between 1.5% and 4.2% for the winter wheat (considering exposure or absorbed dose, respectively), whereas the maize yield was not affected by O3 (Table 4). The difference of yield losses between maize and winter wheat can mainly be explained by the larger sensitivity of the winter wheat to O3 [e.g., Mills et al., 2007]. In addition, although yield loss using the accumulated stomatal flux approach was not estimated for maize, lower losses were expected for maize than for winter wheat since the cumulated stomatal flux was substantially lower for the maize crop (5.4 kg ha−1) than for the winter wheat (10.5 kg ha−1) (Table 3).

Table 4. Phytotoxic Ozone Dose Above a Flux Threshold of 6 nmol m−2 s−1 (POD6), Accumulated Ozone Concentration Above a Threshold of 40 ppb (AOT40), and Crop Yield Reductions Estimated From POD6 and AOT40 for the Maize and the Winter Wheat
 POD6 (mmol m−2)AOT40 (ppm h, 3 months)Yield Reduction From POD6Yield Reduction From AOT40
Maize 0.23 (from 1 June to 1 September 2008) 0%
Winter wheat1.0940.34 (from 1 April to 1 July 2009)4.2%1.5%

[37] Some studies have attempted to estimate crop yield losses at the global scale. For example, Van Dingenen et al. [2009] estimated that present day global relative yield losses range between 7% and 12% for wheat and between 3% and 5% for maize, according to the metrics used. More recently, Avnery et al. [2011a] reported yield reductions ranging from 3.9% to 15% for wheat and from 2.2% to 5.5% for maize. Although, our crop yield losses are at the lower end of these values, they still remain within the same range of estimates at the global scale.

[38] It is noticeable that the impact evaluated based on the absorbed O3 is between 2 and 3 times larger than those based on the AOT40, pointing out the need for flux-based estimation. Finally, it should be stressed that our estimation does not account for any potential feedback between O3 deposition and plant functioning (e.g., decrease in photosynthesis), which would in turn have decreased O3 absorption. Hence, the “real” impact of O3 on yield could be higher.

5 Summary and Conclusions

[39] This study presents the model results and the partitioning of O3 deposition fluxes for an agricultural field planted with mustard, maize, winter wheat, and winter triticale crops over a 2 year period using the Surfatm-O3 model. Modeled O3 fluxes exhibited a pretty good agreement with measured O3 fluxes, independently of the crop development stage and the season, which implicitly validates the partitioning of the O3 deposition in soil, cuticular, and stomatal components.

[40] The model was used to compare the total, soil, cuticular, and stomatal O3 budgets. Nonstomatal deposition dominated the total O3 budget, mainly due to the large contribution of the soil deposition pathway. Soil, cuticular, and stomatal partitioning exhibited differences due to both influence of meteorological conditions and phenological and physiological characteristics of each crop.

[41] The comparison with other parameterizations of resistances to O3 deposition used in existing models indicated that the exclusion of the new parameterizations leads to an underestimation of the dry O3 deposition by about 20%. This result is consistent with the studies of Wild [2007] and Fowler et al. [1998], showing that the simplifications applied to dry deposition schemes may be a more significant source of uncertainties than those applied to the reaction schemes for chemical O3 production. In addition, the underestimations of the dry deposition term not only affect the tropospheric O3 budget, but also that of other species such as NOx, since O3 is closely linked with NO destruction and NO2 production.

[42] We conclude that annual O3 budgets for agricultural areas under relatively dry conditions experiencing longer periods without or with low vegetation may be dominated by soil deposition. The influence of relative humidity on soil and cuticular resistances must be taken into account to establish accurate O3 budgets in the future. We showed that the Surfatm-O3 model represents a useful tool not only to improve annual O3 deposition budgets, but also to determine the trade-off between the beneficial effects of O3 deposition (atmospheric cleanup and lower GHG load) and the deleterious effect to terrestrial ecosystems. Since the Surfatm-O3 model is capable of predicting the O3 deposition via each deposition pathway, it could be used to simulate stomatal (leading to potential damages to plants) and nonstomatal deposition for different canopies (e.g., winter versus summer crops). This allows a determination of the compromise between maximal total O3 deposition and minimal stomatal absorption of O3 as a function of climatic conditions and canopy phenology.

[43] Nevertheless, although this study clearly separates the three deposition pathways, the cuticular deposition process is yet unexplained. It cannot be excluded that it is a chemical removal. It is, however, not possible that reaction with NO be the process due to low NO emissions in Grignon [Loubet et al., 2011] and occasional advection events [Loubet et al., 2013]. A reaction with a yet unknown VOC can however not be dismissed, an issue already evaluated and discussed by Wolfe et al. [2011] in a modeling study. Finally, further work is required, in particular, testing the model for other ecosystems such as forest or grassland.

Acknowledgments

[44] This work was supported by the European Commission through CarboEurope-IP and NitroEurope-IP projects, a French regional funding R2DS (Région Ile-de-France), a French-German project PHOTONA (CNRS/INSU/DFG), and a French national project Vulnoz (ANR, VMC). Michel Burban, Céline Decuq, Pascal Duprix, Brigitte Durand, Olivier Fanucci, and Nicolas Mascher are greatly acknowledged for their assistance in the maintenance of the experimental site and data acquisition and processing. The authors also gratefully acknowledge Bernard Defranssu, Dominique Tristant, and Jean-Pierre de Saint-Stéban from the experimental farm of AgroParisTech Grignon for providing access to their fields.

Ancillary