High concentrations of ozone in ambient air are hazardous not only to humans but to the ecosystem in general. The impact of ozone damage on vegetation and agricultural plants in combination with advancing climate change may affect food security in the future. While the future scenarios in themselves are uncertain, there are limiting factors constraining the accuracy of surface ozone modeling also at present: the distribution and amount of ozone precursors and ozone-depleting substances, the stratosphere–troposphere exchange, as well as scavenging processes. Removal of any substance through gravitational settling or by uptake by plants and soil is referred to as dry deposition. The process of dry deposition is important for predicting surface ozone concentrations and understanding the observed amount and increase of tropospheric background ozone. The conceptual dry deposition velocities are calculated following a resistance-analogous approach, wherein aerodynamic, quasi-laminar, and canopy resistance are key components, but these are hard to measure explicitly. We present an update of the dry deposition scheme implemented in Oslo CTM3. We change from a purely empirical dry deposition parameterization to a more process-based one which takes the state of the atmosphere and vegetation into account. We examine the sensitivity of the scheme to various parameters, e.g., the stomatal conductance-based description of the canopy resistance and the choice of ozone surface resistance, and evaluate the resulting modeled ozone dry deposition with respect to observations and multi-model studies. Individual dry deposition velocities are now available for each land surface type and agree generally well with observations. We also estimate the impact on the modeled ozone concentrations at the surface. We show that the global annual total ozone dry deposition decreases with respect to the previous model version (
Ozone is an important trace gas for all life forms on Earth. Depending on the place of its occurrence, it has either a positive or negative connotation. In the stratosphere, ozone absorbs most of the ultraviolet (UV) light from the Sun within the range of 100–315
In the troposphere, and in particular in ambient air, ozone is considered a highly toxic pollutant. Since the industrial revolution, tropospheric background ozone concentrations have been increasing in the Northern Hemisphere
Elevated ozone levels at a site may originate from both the local production of ozone from its precursors, which are transported, and from advection of ozone itself. Long-range ozone transport occurs regularly and might be most important in regions that otherwise lack precursors. Tropospheric ozone is produced in complex photochemical cycles involving precursor gases such as carbon monoxide (
Since ozone is highly reactive, its global mean lifetime in the troposphere is roughly
Removal of any substance from the atmosphere which is not involving rain, e.g., through gravitational settling or by uptake by plants, soil, and water, is referred to as dry deposition. The process of dry deposition is important for predicting surface ozone concentrations and understanding the observed amount and increase of tropospheric background ozone. It is estimated that
In Sect.
Oslo CTM3 is an offline, three-dimensional global chemistry transport model (CTM). The key components of Oslo CTM3 have been described and evaluated by
While the meteorological data driving Oslo CTM3 are given at a resolution of T159N80L60, with the highest model level at 0.02
In the original dry deposition scheme, the state of the atmosphere was not taken into account. Dry deposition velocities were rather parameterized following the work of
Regarding the new dry deposition scheme, we mainly follow
As displayed in Eq. (
In general, the aerodynamic resistance describes the turbulent transport of any substance down to the surface. To derive
The quasi-laminar layer resistance
The surface resistance consists of both stomatal and non-stomatal resistance.
The stomatal conductance is a measure of the rate of
The factors herein are normalized and vary within the range 0–1. They account for leaf phenology (
The water vapor deficit (VPD) is proportional to the saturation partial pressure of water (
The penalty factor with respect to available soil water (SW)
The phenology of a plant typically describes its life cycle throughout a year; e.g., at midlatitudes and for deciduous species, it starts with the emergence of leaves in spring and ends in fall. In the mOSaic scheme, phenology is parameterized with respect to the start of the greening season (SGS) and its end (EGS). Details about our treatment of these are given in Sect.
Herein, we use the SGS- and EGS-derived parameters: day of greening season (GDAY), the time elapsed starting at the SGS, and the total length of the greening season (GLEN), the time span between EGS and SGS. The parameters
Sketch of the five different phases in plant phenology
Light in the wavelength band 400–700
In the mOSaic scheme, non-stomatal conductance is explicitly calculated for
Extending the mOSaic scheme to the Southern Hemisphere, we use the growing season for crops defined in Table
Definition of growing season for crops used in Oslo CTM3 in the NH and SH.
In this way, vegetation affects the conductance also by being there, not only by uptake through the stomata. The in-canopy resistance
As initially mentioned, the necessary depth of snow to cover a certain type of vegetation differs. Therefore, we calculate a snow cover fraction
The vegetation height
Oslo CTM3 is configured to read land surface types from either the ISLSCP2 (
Mapping of land surface categories. Either land surface categories from ISLSCP2 product of MODIS or the Community Land Model (CLM) 2 can be chosen for mapping to the land surface types we use in the mOSaic scheme. Water bodies of MODIS are actually not mapped. For both MODIS and CLM 2 land surface categories, snow and ice cover is estimated from input meteorology, while water is defined as
As mentioned in the previous section, there are two variables needed for computing the stomatal conductance which are not directly available from the meteorological input data: the greening season, as the time of the year in the mid- and high latitudes when it is most likely for plants to grow, and the photosynthetic photon flux density, as the amount of light that plants need to photosynthesize. In the following, we present the necessary pre-processing of the variables. It is planned to implement an online computation of these variables into Oslo CTM3 later on.
In Eqs. (
In agriculture, there are different empirical rules to estimate the SGS and EGS. The simplest assumption is that greening starts after
Based on the empirical rule (5
Pre-processing of greening season from meteorological surface temperature fields. Shown is the total length of the greening season (GLEN) for the year 2005. The 5
From OpenIFS an accumulated surface PAR is available. It is integrated both spectrally (presumably 400–700 nm) and temporally. For practical use in Eq. (
The main obstacle is that PAR has been accumulated since model start, so that the first field kept from the original OpenIFS simulation (00:00 UTC) is
In this section, we present results from manifold Oslo CTM3 model integrations testing different parameters of the mOSaic scheme. We focus on changes in ozone total dry deposition
Due to significant differences between the mOSaic scheme and the previous Wesely scheme with respect to implementation, it is not possible to fully disentangle and trace back every single difference in results to a respective change. Therefore, we conducted one reference simulation denoted as
First, we take a closer look at the influence of certain parameters on the stomatal conductance. As indicated by the names,
Summary of specifications of all simulations discussed in this section. For simplicity, only the tested parameters are listed. An “x” denotes that the model was run exactly in the configuration as has been described in Sect.
Relative difference between reference simulations
In Fig.
In the evaluation of our model, we closely follow suggestions by
Dry deposition velocities are directly available only for the new model version. For
Comparison of the manifold Oslo CTM3 integrations with respect to
The annual zonal average with respect to surface ozone concentration (Fig.
The
The annual total ozone dry deposition is shown in Fig.
There seems to be a discrepancy between the Oslo CTM3 response and the multi-model mean, since the Wesely scheme is similar to the multi-model mean with respect to total annual ozone dry deposition, while the
In Fig.
To further disentangle the contributions of different regions to the global ozone budget, we will look at different projections of seasonal cycles.
In Fig.
Seasonal cycle of total annual amount of ozone removed from the atmosphere through dry deposition separated into the NH, tropics (TR), and SH. The multi-model mean from the evaluation of HTAP models by
As expected, the NH mid- and high latitudes display a strongly pronounced seasonal cycle, while it is less pronounced in the tropics (due to the lack of seasons) and in the SH (due to the small percentage of vegetated surface). The highest ozone dry deposition is found in the tropics and amounts on average to the peak level of dry deposition in the NH for the multi-model mean
As suggested by
In Fig.
Average seasonal cycles of ozone dry deposition velocities separated by land use type. Results from
Finally, we take a look at the different global as well as hemispheric dry deposition sinks for ozone (Table
Total ozone dry deposition for the respective model experiment in
Table
Annual mean tropospheric ozone burden for all experiments and
In this section, we conclude the comparison of our results with respect to global ozone by looking at ECMWF's MACC reanalysis
Mean ozone concentrations for the year 2005.
In this section, we compare our model results to observations at a selected number of sites which provide ozone flux measurements. For all comparisons, we use the original resolution of Oslo CTM3 (
Ozone dry deposition fluxes at different observation sites.
In Fig.
We have presented an update of the dry deposition scheme in Oslo CTM3 from purely prescribed dry deposition velocities
The new dry deposition scheme named
We found the response of Oslo CTM3 to the changes in dry deposition velocities from the old and the mOSaic scheme to be consistent. A decrease in
Most of the qualitative change in ozone dry deposition in Oslo CTM3 (
Although dry deposition to ice and snow amounts to only 1 % of the total global annual ozone dry deposition in
We have studied the parameter space of the stomatal conductance parameterization and found that surface ozone in the tropics and the Northern Hemisphere is most sensitive to changes therein. In the most extreme test case, the increase in global total dry deposition amounts to 7.3 %, while the more realistic test cases, e.g., using differing years of emission, amount to changes of the order of
An important factor in the global ozone budget is emissions of precursor substances. We cover this by using the same meteorology with different years of CEDS emissions. We chose the years 2005 and 2014 for our comparison. Ozone precursor emissions in 2014 are slightly lower in the NH while enhanced in the tropics and the SH. In 2014, surface ozone burden is higher in the Southern Hemisphere and in the tropics (5 %) compared to 2005, while it is lower in the Northern Hemisphere (2 %).
We also evaluated the model with respect to observed dry deposition fluxes at six sites in the Northern Hemisphere and found that the mOSaic scheme performs better than the old one but is not able to reproduce the measurements at most sites quantitatively. This may be due to several reasons. The model resolution in both horizontal (
Future work on Oslo CTM3 should resolve the ozone high bias which may involve revising the photolysis and chemical reaction computation as well as reaction rates. For a better modeling of ozone abundances, ocean emissions of very-short-lived ozone-depleting substances (VSLSs)
Oslo CTM3 shall be publicly available on GitHub under an MIT license in the future. Until then, access can be granted upon request. Model results can be made available upon request.
The used LAI and roughness length (
CEDS historical emission inventory is provided by the Joint Global Research Institute project (
Data from the Global Fire Emissions Database, Version 4, (GFEDv4) are available online from Oak Ridge National Laboratory Distributed Active Archive Center, Oak Ridge, Tennessee, USA (
Comparison of the manifold Oslo CTM3 integrations with respect to
Partitioning of land surface types:
The supplement related to this article is available online at:
SF compiled the manuscript, finalized and revised the implementation of the stomatal conductance in the
The authors declare that they have no conflict of interest.
This work was supported by the Norwegian Research Council (NRC) through the project “The double punch: Ozone and climate stresses on vegetation” (268073).
The simulations were performed on resources provided by UNINETT Sigma2 – the National Infrastructure for High Performance Computing and Data Storage in Norway (project nn2806k).
We would like to thank Frode Stordal (Section for Meteorology and Oceanography, University of Oslo) for discussions regarding early drafts of the manuscript, Anne Fouilloux (scientific programmer at the same institute) for technical support as well as Olimpia Bruno (Karlsruhe Institute of Technology), and Franziska Hellmuth (University of Oslo) for valuable input regarding the aerodynamic resistance formulation. We would also like to thank the Center for International Climate Research (CICERO) for their support of this work.
This research has been supported by the Norwegian Research Council (grant no. 268073).
This paper was edited by Fiona O'Connor and reviewed by David Simpson and one anonymous referee.