Soil moisture and water stress play a pivotal role in regulating stomatal behaviour of plants; however, in the last decade, the role of water availability has often been neglected in atmospheric chemistry modelling studies as well as in integrated risk assessments, despite the fact that plants remove a large amount of atmospheric compounds from the lower troposphere through stomata.
The main aim of this study is to evaluate, within the chemistry transport model CHIMERE, the effect of soil water limitation on stomatal conductance and assess the resulting changes in atmospheric chemistry testing various hypotheses of water uptake by plants in the rooting zone.
Results highlight how dry deposition significantly declines when soil
moisture is used to regulate the stomatal opening, mainly in the semi-arid
environments: in particular, over Europe the amount of ozone removed by dry
deposition in one year without considering any soil water limitation to
stomatal conductance is about 8.5 TgO
Our results shed light on the importance of improving the parameterizations of processes occurring at plant level (i.e. from the soil to the canopy) as they have significant implications for concentration of gases in the lower troposphere and resulting risk assessments for vegetation or human health.
Plant-level water cycling and exchange of air pollutants between atmosphere and vegetation are intimately coupled (Eamus, 2003; Domec et al., 2010); thus, any factor affecting root water absorption by plants is expected to impact the concentration of gases in the lower troposphere by changing deposition rates. In fact, atmospheric gases, including air pollutants, are primarily removed from the troposphere by dry deposition to the Earth's surface (Hardacre et al., 2015; Monks et al., 2015). A major part of dry deposition to vegetation is regulated by stomata opening, which strongly depends on the amount of water available in the soil (Büker et al., 2012). Therefore a proper quantification of soil water content as well as a proper understanding of stomatal response to soil moisture is required for correctly quantifying the concentration of gases in the atmosphere, particularly in water-limited ecosystems (dry and semidry environments), which cover 41 % of Earth's land surface (Reynolds et al., 2007).
Among common air gasses, ozone (O
At the European level, the model currently parameterized for European vegetation
and developed to estimate surface O
However, in the last decade the importance of soil water stress on vegetation
has been well demonstrated in several studies reporting a large reduction in
the amount of air gases taken up from the atmosphere during heat waves or
drought years (e.g. Ciais et al., 2005; Granier et al., 2007; Reichstein et
al., 2007) with species responding in different ways to scarce water
availability, depending on eco-hydrological properties (Granier et al., 1996;
Pataki et al., 2000; Pataki and Oren, 2003) and drought avoidance and
tolerance strategies (Martinez-Ferri et al., 2000; Bolte et al., 2007). For
instance, drought-avoiding species (e.g.
Moreover, it is important to take into account that soil drying does not occur at the same rate at different depths, and the drying rate is more pronounced in the superficial soil layers than in the deeper ones. Overall, deep-rooted forest systems take up water from deep to shallow soil horizons (Aranda et al., 2012). In contrast, shallow-rooted grass normally adsorbs available soil water from top–middle soil, while shrubs can take up soil water adaptively from top to deep soil layers, with increased use of topsoil water under non-drought stress and a tendency of using water from deeper soil under drought stress (Wu et al., 2017). Thus, plants able to develop a deeper root system are usually more tolerant to low water availability than plants with a more superficial root system (Canadell et al., 1996). Jackson et al. (2000) showed that differences in rooting depth patterns vary between world's major plant biomes, with plants of xeric environments having deeper root-depth distributions than plants in more humid environments. In contrast, Schenk and Jackson (2002) found that maximum rooting depths tend to be shallowest in arid regions and deepest in sub-humid regions.
Consequently, the role of root systems is fundamental in stomatal
conductance regulation and thus in atmospheric chemistry modelling: results
from a sensitivity analysis of ozone dry deposition model indicate that soil
moisture is one of the most crucial factors of deposition in the continental
climate region (Mészáros et al., 2009). For these reasons, recently
the DO
Chemistry transport models are widely used to estimate the concentration of
gases in atmosphere at both regional and global scale; in these models the
concentration of a given gas species is mainly regulated and parameterized by
three different processes: atmospheric transport, chemical
production/destruction and losses to surface by dry deposition (Monks et al.,
2015). Within these models, the dry deposition is generally simulated through
an electrical resistance analogy (Wesely, 1989; Monks et al., 2015), that is,
the transport of material to the surface is assumed to be controlled by three
different resistances: the aerodynamic resistance (
In this study, we improve the dry deposition scheme within the chemistry
transport model CHIMERE considering the effect of soil water limitation to
stomatal conductance. Our main aim was to perform several different
simulations testing various hypotheses of water uptake by plants at
different soil depths in the rooting zone, based on the main assumption that
roots maximize water uptake to fulfill resource requirements adsorbing water
at different depths depending on the water availability. Finally, we show and
discuss the resulting effects on O
We use a multi-model system to reproduce the meteorological conditions and the concentration of gases in the troposphere; this framework is composed by the WRF (Weather Research and Forecast Model) regional meteorological model and the CHIMERE chemistry transport model.
In this study, in order to have a large latitudinal gradient and assess the role of soil moisture across different climatic zones, we selected a domain extending over all of Europe (except Iceland). For both WRF and CHIMERE we performed a simulation for the whole year 2011, with a spin-up of 2 months to initialize all the fields.
Meteorological variables are simulated with the WRF regional model (v 3.6);
it is a limited-area, non-hydrostatic, terrain-following eta-coordinate
mesoscale model (Skamarock et al., 2008) widely used worldwide for climate
studies. In our configuration, the model domain is projected on a regular
latitude–longitude grid with a spatial resolution of 16 km and with
30 vertical levels extending from land surface to 50 hPa. The initial and
boundary meteorological conditions required to run the WRF model are provided
by the European Centre for Medium-Range Weather Forecast (ECMWF) analyses
with a horizontal resolution of 0.7
The exchange of heat, water and momentum between soil, vegetation and atmosphere is calculated using the Noah land surface model (Chen and Dudhia, 2001); in our configuration the soil has a vertical profile with a total depth of 2 m below the surface and it is partitioned into four layers with thicknesses of 10, 30, 60, and 100 cm (giving a total of 2 m). The root zone is fixed at 100 cm (i.e. including the top three soil layers). Thus, the lower 100 cm of soil layer acts as a reservoir with gravity drainage at the bottom (Al-Shrafany et al., 2014).
For each soil layer Noah calculates the volumetric soil water content
(
For the definition of vegetation and land cover WRF uses the United States Geological Survey (USGS) land cover dataset, which has a resolution of 1 km with 24 categories (Loveland et al., 2000; Hibbard et al., 2010; Sertel et al., 2010); this land cover dataset is derived from the 1 km satellite Advanced Very High Resolution Radiometer (AVHRR) data. In addition to land cover, WRF defines 12 soil types and 4 non-soil types, including organic material, water, bedrock, and ice. Soil types are classified based on the percentage of sand, silt, and clay in the soil (Dy and Fung, 2016); for each soil type, WRF has a default soil parameter table that generalizes the hydraulic and thermal properties of the soil. Soil texture data are derived from the Food and Agriculture Organization (FAO) 5 min soil type categories.
One useful capability of WRF is its flexibility in choosing different dynamical and physical schemes; Table 1 lists the main options used in this study for physical schemes.
WRF 3.6 physical configurations used in the model simulations.
The chemistry transport model used in this study is CHIMERE (v2014b), an Eulerian model developed to simulate gas-phase chemistry, aerosol formation, transport and deposition at a regional scale (Menut et al., 2014).
The gas-phase chemical mechanism used by CHIMERE is MELCHIOR2 (Lattuati,
1997), which consists of a simplified version (40 chemical species, 120
reactions) of the full chemical mechanism MELCHIOR; this latter mechanism describes
more than 300 reactions of 80 species. Photolysis rates are explicitly
calculated using the FastJ radiation module (Wild et al., 2000), as
described by Mailler et al. (2016, 2017). External meteorological forcing
required by CHIMERE to calculate the atmospheric concentrations of gas-phase
and aerosol species are directly provided by the WRF simulation. In
addition, to accurately reproduce the gas-phase chemistry, emissions must be
provided every hour for the specific species of the chemical mechanism. For
studies over Europe, the EMEP inventory (Vestreng et al., 2009) is usually
used for anthropogenic emissions of NO
Boundary conditions are provided as a monthly climatology of the LMDz-INCA global chemistry transport model (Hauglustaine et al., 2004; Folberth et al., 2006) for gaseous species and the GOCART model (Ginoux et al., 2001) for aerosol species. More details regarding the parameterizations of the above-mentioned processes are described in Menut et al. (2014).
Following Wesely (1989), within CHIMERE the dry deposition velocity is
described through an electrical resistance analogy, that is, the transport of
material from the lowest model layer to the surface is assumed to be
dependent on three different resistances: the aerodynamic resistance
(
The leaf-level stomatal conductance (i.e. the inverse of stomatal resistance)
is estimated by CHIMERE using the DO
Here, we improve the DO
Meteorological fields required by the DO
As the original version of CHIMERE does not account for any limitation of soil moisture to stomatal conductance, in the following analysis we use the simulation NO_SWC as reference; thus we show and discuss models' changes with respect to this original configuration (Menut et al., 2014).
In order to assess how the new parameterization of dry deposition changes the
ability of CHIMERE to reproduce the spatial distribution of surface O
Comparison of hourly precipitation simulated by WRF with
observations collected at four measurement sites along with changes in the
vertical distribution of soil moisture (m
For the validation of O
Considering the soil moisture, we retrieve precipitation data over four
forested eddy covariance sites belonging to the European flux network
(
Figure 1 shows the seasonal variation of simulated soil water content at four different locations; in order to assess the reliability of vertical soil moisture profiles we also evaluate models skills in capturing precipitation events by comparing the hourly simulated precipitation with data collected over the four measurement stations.
The first site, Leinefelde in Germany, is characterized by a
temperate/continental climate with mean annual precipitation ranging between
700 and 750 mm, covered by a beech forest (
The second temperate site, covered by a spruce forest (
In Collelongo, a
Percentage change in the amount of O
The fourth station is San Rossore, a Mediterranean
Overall, these results suggest that soil water availability was higher from April to September for the two central European sites, where soil water content remained above 50 % of total available water capacity. In the Mediterranean sites, water availability declined from spring onwards, but remained above 40 % total available water capacity until late August, while effective drought conditions occurred in October.
Comparison of seasonal amount of O
The inclusion of soil water limitation in the stomatal conductance
parameterization firstly affects the surface resistance, which in turn
affects the dry deposition velocity and thus the amount of air pollutants
removed from the surface layer by dry deposition (Seinfeld and Pandis, 2016;
Hardacre et al., 2015; Monks et al., 2015). Figure 2 shows the mean
percentage change in O
In addition, in order to point out the seasonal changes between different
climatic zones, in Fig. 3 we show the dry deposition integrated over
different domains along with its daily variability. As already discussed
above, for all the seasons and climatic regions, the NO_SWC simulation shows
the largest amount of O
Furthermore, it is noteworthy how the inclusion of soil water limitation
changes not only the amount of pollutant removed by deposition but also its
variability; specifically, in all the domains and seasons (except the
Mediterranean area during summer) we found a relevant reduction in the
standard deviation of daily O
Overall, during the whole year the amount of O
Percentage change in surface O
As plants uptake atmospheric gases into the leaves when stomata are open
(Cieslik et al., 2009), changes in stomatal behaviour (and thus in dry
deposition velocity) affect, in turn, the concentration of compounds
remaining in the lower atmosphere; Fig. 3 shows the mean percentage change
in O
Similarly, the vertical mixing in surface layers, largely driven by wind and
its interaction with frictional drag at the surface (Monks et al., 2015),
propagates the changes in O
Vertical anomaly in O
As discussed above, the inclusion of soil water limitation to stomatal
conductance leads to increased O
Figure 6 (upper panels) shows how the inclusion of the new parameterization leads to an increase of model–data misfit during the temporal period April–September, being the percentage change in RMSE positive in all the stations. Overall, the mean RMSE (average over all the stations) computed comparing hourly data is 17.8 ppb for the NO_SWC simulation, 19.5 ppb in the SWC_10cm and SWC_40cm, and 19 ppb in the SWC_1m and SWC_DYN simulations.
Conversely, the new parameterization improves the model skills in reproducing the observed hourly cycle (Fig. 6, lower panels), being the percentage change in correlation coefficient positive in all the stations. Overall, the mean correlation computed from hourly data is 0.6 for the NO_SWC simulation, 0.62 in the SWC_10cm and 0.64 in the SWC_40cm, SWC_1m and SWC_DYN simulations.
This result is in agreement with a previous study which showed how, within CHIMERE, the deposition not only acts as a shifting term on the modelled concentration but also influences the variability and timing of ozone (Solazzo et al., 2017).
In this study, we incorporated the soil moisture limitation into the dry
deposition parameterization of CHIMERE model and tested different hypotheses
of water uptake by roots. Model simulations with the improved
parameterization indicate that O
Percentage change in RMSE
Spatial distribution of AOT40 and SOMO35 (upper panels) along with
their percentage change (lower panels) computed using the NO_SWC
simulation as reference. The AOT40 is defined as the accumulated amount of
ozone over the threshold value of 40 ppb computed during the vegetation
growing season, i.e.
AOT40
The analysis of simulated soil moisture suggests that actual water
availability from April to September, even in the Mediterranean sites, is
higher than conventionally assumed; according to Allen et al. (1998) and
Martínez-Fernández et al. (2015), soil water content values
corresponding to 40–50 % of total available water (TAW, FC-WP) often
correspond to low stress conditions for cultivated plants. As the stress
threshold lowers with rooting depth (Allen et al., 1998), it appears likely
that the effect of water deficit on forest vegetation is limited in these
conditions. As the effect of soil water content on
stomatal aperture in the modified DO
With the modified parameterization, CHIMERE shows increased bias in the
prediction of surface hourly O
It should also be noted that the model comparison to satellite retrievals is
not obvious in this study: here, we mainly focus on O
However, in this study compared to former ones, generally the uncertainty in
the dry deposition associated to soil moisture is relatively low
(10–11 %), although it is above 30 % in a few points. Schwede et
al. (2011) compared two deposition velocity models in two long-term
monitoring networks in USA and Canada and found that the hourly median values
of ozone, and therefore the flux, can be 2 or 3 times
different depending on
the deposition velocity model used. Similarly, Flechard et al. (2011) found
differences between four dry deposition models by a factor of 2 or 3 for five
atmospheric reactive nitrogen species (NH
Moreover, our results are in agreement with Solazzo et al. (2017), which created a diagnostic methodology for model evaluation; using CHIMERE, they showed
that setting the ozone dry deposition velocity to zero causes a profound
change of the error structure of O
Finally, we point out that the uncertainty associated with different
models or dry deposition schemes (or assumptions in rooting depth, as in this
study) might have severe implications in the case of risk assessment for
vegetation or human health. For instance, Fig. 7 shows the spatial
distribution of the AOT40 (i.e. accumulated ozone over threshold of 40 ppb)
and SOMO35 (sum of ozone means over 35 ppb), which are the two metrics used for
vegetation and human health impact assessment over Europe. It should be noted
that over eastern Europe the risk for vegetation can differ by up to 90 %
between the reference case (i.e. NO_SWC) and the simulation using a shallow
rooting zone (i.e. SWC_10cm), while for the human health we report a
difference exceeding 30 % over large areas of Europe. This result clearly
shows an amplification of the percentage change with respect to both
O
Nevertheless, our results can be used to improve the representation of soil moisture stress on vegetation within chemistry transport models and to better describe the biogeochemical and biophysical feedbacks between the complex soil–plant–atmosphere system in response to a changing climate toward warmer and drier conditions. As the soil water uptake is mainly related to different rooting systems (Wu et al., 2017), chemistry models would benefit from the inclusion of species-specific parameterizations which ensure water uptake depending on species-specific eco-hydrological properties. In general, plants in water-limited regions can adapt to dry environments by accessing ground water (Craine et al., 2013) based on the depth and density of the root system (Wu et al., 2017), while deep-rooted forests can take up available water from deep soil during extreme drought events (Schwinning et al., 2005; Teuling et al., 2010). Although some of these processes are already well resolved within land surface models used by climate models, a better description of different rooting systems within the dry deposition schemes might have significant implications for stomatal regulation and thus atmospheric chemistry. We also believe that it will be challenging in the near future to use coupled land surface–chemistry models (e.g. Anav et al., 2012), which allow accounting for the different feedbacks between land surfaces and atmospheric chemistry and physics, especially in a changing climate.
The model used in this study is freely available and
provided under the GNU general public license. The source code along with the
corresponding technical documentation can be obtained from the CHIMERE web
site at
The authors declare that they have no conflict of interest.
We thank the investigators and the teams managing the eddy-flux sites. We
also acknowledge the entire EMEP and AIRBASE staff for providing ground-based
O