ACPAtmospheric Chemistry and PhysicsACPAtmos. Chem. Phys.1680-7324Copernicus PublicationsGöttingen, Germany10.5194/acp-17-14333-2017Impact of uncertainties in inorganic chemical rate constants on tropospheric composition and ozone radiative forcingNewsomeBenhttps://orcid.org/0000-0001-7645-7793EvansMatmat.evans@york.ac.ukhttps://orcid.org/0000-0003-4775-032XWolfson Atmospheric Chemistry Laboratories, Department of Chemistry, University of York, York, YO10 5DD, UKNational Centre for Atmospheric Science, Department of Chemistry, University of York, York, YO10 5DD, UKMat Evans (mat.evans@york.ac.uk)4December2017172314333143525January20179March20171October20176October2017This work is licensed under the Creative Commons Attribution 3.0 Unported License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/3.0/This article is available from https://acp.copernicus.org/articles/17/14333/2017/acp-17-14333-2017.htmlThe full text article is available as a PDF file from https://acp.copernicus.org/articles/17/14333/2017/acp-17-14333-2017.pdf
Chemical rate constants determine the composition of the atmosphere and how
this composition has changed over time. They are central to our understanding
of climate change and air quality degradation. Atmospheric chemistry models,
whether online or offline, box, regional or global, use these rate constants.
Expert panels evaluate laboratory measurements, making recommendations for
the rate constants that should be used. This results in very similar or
identical rate constants being used by all models. The inherent uncertainties
in these recommendations are, in general, therefore ignored. We explore the
impact of these uncertainties on the composition of the troposphere using the
GEOS-Chem chemistry transport model. Based on the Jet Propulsion Laboratory
(JPL) and International Union of Pure and Applied Chemistry (IUPAC)
evaluations we assess the influence of 50 mainly inorganic rate constants and
10 photolysis rates on tropospheric composition through the use of the
GEOS-Chem chemistry transport model.
We assess the impact on four standard metrics: annual mean tropospheric ozone
burden, surface ozone and tropospheric OH concentrations, and tropospheric
methane lifetime. Uncertainty in the rate constants for NO2+ OH ⟶M HNO3 and
O3+ NO → NO2+ O2 are the two largest sources of uncertainty in these metrics. The
absolute magnitude of the change in the metrics is similar if rate constants
are increased or decreased by their σ values. We investigate two
methods of assessing these uncertainties, addition in quadrature and a Monte
Carlo approach, and conclude they give similar outcomes. Combining the
uncertainties across the 60 reactions gives overall uncertainties on the
annual mean tropospheric ozone burden, surface ozone and tropospheric OH
concentrations, and tropospheric methane lifetime of 10, 11, 16 and 16 %,
respectively. These are larger than the spread between models in recent model
intercomparisons. Remote regions such as the tropics, poles and upper
troposphere are most uncertain. This chemical uncertainty is sufficiently
large to suggest that rate constant uncertainty should be considered
alongside other processes when model results disagree with measurement.
Calculations for the pre-industrial simulation allow a tropospheric ozone radiative
forcing to be calculated of 0.412 ± 0.062 W m-2. This uncertainty (13 %) is comparable to the inter-model spread in ozone radiative forcing found
in previous model–model intercomparison studies where the rate constants
used in the models are all identical or very similar. Thus, the uncertainty of
tropospheric ozone radiative forcing should expanded to include this
additional source of uncertainty. These rate constant uncertainties are
significant and suggest that refinement of supposedly well-known chemical
rate constants should be considered alongside other improvements to enhance
our understanding of atmospheric processes.
Introduction
The concentration of gases and aerosols in the atmosphere have changed over
the last century due to human activity. This has resulted in a change in
climate and a degradation in air quality
with tropospheric ozone (O3) and
methane (CH4) playing a central role. The response of these compounds to
the changing emissions is complex and non-linear
. The hydroxyl radical (OH) plays a central role
in this chemistry, as it initiates the destruction of many pollutants (notably
CH4) and thus determines their lifetime in the atmosphere. The dominant
source of OH is the photolysis of O3 in the presence of water vapour.
The oxidation of compounds such as CH4, carbon monoxide (CO) and other
hydrocarbons can lead to the production of O3 if sufficient oxides of
nitrogen (NOx) are present. Changes in the emissions of O3
precursors between the pre-industrial (∼ 1850) and the present-day periods have
increased O3 concentrations and this has produced a radiative forcing
estimated to be 410±65 mW m-2.
The rate constants of the reactions occurring in the atmosphere have been
determined by a number of laboratory studies which are synthesised by groups
such as the International Union of Pure and Applied Chemistry (IUPAC) and Jet Propulsion Laboratory (JPL) panels. These provide
recommendations for both rate constants and their associated uncertainties.
These reactions are typically expressed in an Arrhenius form to represent the
temperature dependence. More complicated representations are needed for
three-body reactions. IUPAC and JPL provide similar but different
representations of the uncertainty in a rate constant. For IUPAC (Eq. ), the uncertainty in a rate constant is
described as the uncertainty in the log10 of the rate constant (Δlog10kT) at a temperature (T), with the panel giving values
for the log10 uncertainty at 298 K (Δlog10k298K) and
the rate of increase in uncertainty away from 298 K described by a ΔE/R term. For JPL (Eq. ), the relative
uncertainty in a rate constant (f(T)) is described as the relative uncertainty
at temperature of 298 K (f (298)) together with a term (g) that expresses how
quickly the uncertainty increases away from 298 K
(Eq. ), leading to temperature dependences
which increase away from room temperature
(Fig. ).
Δlog10kT=Δlog10k298K+0.4343ΔER1T-1298Kf(T)=f(298K)expg1T-1298K
For the reactions studied, the uncertainty at 298 K typically ranges from 5 %
for well-understood reactions to 30 % for those which have significant
uncertainties. Other reactions can have larger uncertainties than quoted
here. The increase in uncertainty at temperatures away from 298 K can range
from 0 to over 40 %, giving some reactions a total uncertainty of over 50 %
in the cold upper troposphere.
Models of atmospheric composition (whether using prescribed meteorology or
calculating the meteorology, single box or transport, etc.) use these
recommended rate constants, together with estimates of the meteorology,
emissions, deposition, photolysis, etc. of compounds to calculate the
concentration of species in the atmosphere. These models are a central tool
for our understanding of atmospheric processes and for making policy choices
to minimise climate change and air pollution.
Example of the uncertainty on a reaction rate constant. The relative
uncertainty of the reaction O3+ NO is plotted as a function of
temperature. The lowest uncertainty is at room temperature (298 K) with
exponentially increasing uncertainties occurring as we diverge to higher and
lower temperatures.
Although these models have been developed significantly over the last
decades, they have, in general, all used the same basic chemical rate
constants as evaluated by the IUPAC or JPL panels. Little emphasis has been
placed on understanding the uncertainty in predicted atmospheric composition
caused by the uncertainty in these rate constants. The focus has been to
investigate the impacts of novel chemical reactions or understanding
emissions, etc. (e.g. ). Here though, we investigate the impact of this
uncertainty on the composition of the troposphere. We base our assessment on
the uncertainties in rate constants described by the JPL and IUPAC panels
using the GEOS-Chem model and evaluate a range of model
diagnostics for both the present-day and the pre-industrial periods.
Model simulations
GEOS-Chem (http://www.geos-chem.org) is an offline
chemistry transport model. We use version 9-2. For computational expediency,
we use a horizontal resolution of 4∘ latitude by 5∘ longitude
with 47 vertical hybrid pressure-sigma levels from the surface to 0.01 hPa.
The chemistry is solved within the troposphere with the SMVGEAR solver
. We use a mass-based scheme for aerosol
and thus can not investigate the impact of the rate constant uncertainty on
aerosol number or size distribution. Stratospheric chemistry is unchanged in
all simulations and uses a linearised approach to the chemistry . Global anthropogenic emissions were taken from the Emission
Database for Global Atmospheric Research (EDGAR) v3 for NOx, CO, VOCs and
SOx. Regional or source-specific inventories replaced EDGAR where
appropriate (EMEP, BRAVO, Streets, CAC, NEI05, RETRO and AEIC; see the GEOS-Chem
wiki for more details). Biogenic emissions (isoprene, monoterpenes, methyl
butenol) are taken from the MEGAN v2.1 emission inventory
. Biomass burning emissions were used from the GFED3
monthly emission inventory . NOx sources from
lightning and soils were also included.
As in previous studies , pre-industrial
emissions are calculated by switching off anthropogenic emissions, reducing
biomass burning emissions to 10 % of their modern-day values and
setting CH4 concentrations to a constant 700 ppbv .
For both present-day and the pre-industrial simulations, we run the model from
1 July 2005 to 1 July 2007 with GEOS-5 meteorology. We used
the first year to spin up the composition of the troposphere. Metrics are
derived from the second year of simulation.
We follow the methodology of JPL for the representation of
uncertainties in rate constants converting IUPAC representation where
necessary. For two-body reactions, the uncertainty is given by two
parameters. f (298 K) describes the relative uncertainty at 298 K, and
g describes how the uncertainty increases as temperature diverges from 298 K, as
shown in Eq. ().
Table of reactions studied. f (298) indicates the JPL or IUPAC
panel uncertainty estimate at 298 K and g gives the rate at which this
uncertainty increases away from 298 K (see previous section). Reactions with
0 for the temperature dependence indicate there is zero temperature
dependency or not enough information to provide a temperature-varying
uncertainty. The final column gives the fractional increase in the ozone
burden by increasing the rate constant to its 1σ value. Reactions with
a ∗ are the 10 reactions used in the Monte Carlo study.
We limit our study to the inorganic (Ox, HOx, NOx, CO,
CH4) reactions together with some key organic and sulfur reactions.
Uncertainties in organic molecules degradation chemistry of the atmosphere
makes a systematic assessment of these uncertainties difficult
. Table shows a
list of reactions that are perturbed and the uncertainties assumed. We use
the uncertainty recommendations from the JPL panel if provided and the IUPAC
panel otherwise. We investigate the impact of 50 inorganic chemical reactions
and 10 photolysis reactions (Table ). Uncertainties
in photolysis rate constants are harder to define than for the other
reactions. We consider the appropriate chemical uncertainty here as the
uncertainty in the absorption cross section and the quantum yield rather than
the uncertainty in the photon flux which we attribute to the radiative
transfer calculation. A full calculation of the chemical uncertainty in a
photolysis rate is complex, as it depends upon the uncertainties at different
wavelengths, the independence of the cross section and quantum yield
parameters and the transfer of this information through the spectral bins
used for the laboratory studies and the photolysis calculations. In order to
simplify this calculation, we apply a 10 % uncertainty to all photolysis
rates. Although this is not ideal, it does allow us to place an uncertainty in
the photolysis rates into the context of other uncertainties. An improved
presentation of the photolysis uncertainty should be included in future work.
Uncertainties in all metrics:
fractional uncertainties of (a) O3 tropospheric burden,
(b) OH tropospheric burden, (c) O3 surface concentration
and (d) CH4 lifetime. Each bar labelled with a reaction
represents a run with a 1σ increase in the rate constant. “Other”
represents the addition in quadrature of the reactions that were not the top
20 most influential. “Top 10” represents the addition in quadrature of the
10 most important reactions, and “Monte Carlo top 10” represents the
standard deviation of the Monte Carlo ensemble. “Total” represents the
addition in quadrature of all the simulations.
Single reaction perturbations
From each of these 60 reactions, we increase the reaction rate by the 1σ temperature-dependent uncertainty given in
Table . To allow the model to spin up, we run the
model for 2 years and take the second year of simulation for the calculation of
four metrics: tropospheric O3 burden, mean surface O3 mixing ratio,
tropospheric mass-weighted mean OH number density and tropospheric mean
CH4 lifetime. We subtract the values of these metrics from the base
value of the metric (unchanged rate constants) and then take the absolute
value to remove cases where the value decreases on an increase in the rate
constant. Figure shows the changes for all four metrics with
Table giving the values for the change in
tropospheric O3 burden. We express these values as a percentage of the
base case value.
Uncertainty linearity: a comparison of absolute uncertainties in
O3 and OH tropospheric burdens for both positive and negative changes to
the rate constants. These reactions show a similar magnitude of tropospheric
species concentration change if the rate is set to its lower or higher sigma
level of uncertainty.
It is evident that a relatively small number of reactions produce large
uncertainties in the values of these metrics. The one that offers the most
uncertainty is the reaction between NO2 and OH to product nitric acid,
which leads to uncertainties in the range of 6–11 % in the metrics
investigated here. This reaction is both highly uncertain (f (298 K) = 30 %) and
acts as a large global sink for NOx and HOx. The O3+ NO
reaction to produce NO2 is central to the partitioning of NOx in
the atmosphere. Thus, increasing its rate constant reduces NO concentrations
in the atmosphere (leading to lower O3 concentrations) and increasing
the concentration of NO2 (which favours NO2 removal) which again
reduces O3 concentrations. Another significant reaction is that between
CH4 and OH to produce CH3O2 radicals. The model assumes a
constant CH4 concentration so an increase in the rate constant between
CH4 and OH leads to an increased source of radicals but does not lead to
a commensurate drop in the CH4 concentration. Thus, an increase in this
rate constant in the model is effectively the same as an increase in the
emission of CH4, which results in a wide range of impacts such as
increased CO concentrations, etc. Reactions after the 10th most significant
reaction for all the metrics generates an uncertainty of less than 1 %.
The relative importance of the different reactions does not change much with
the metric being investigated (see Fig. ). The rate
constants of these top 10 reactions are not particularly uncertain (other
than for NO2+ OH) compared to other reactions, but they link important
chemical cycles and have a very large chemical flux flowing through them.
Thus, relatively small changes in their uncertainties will lead to large
changes in concentration.
It is just as easy to decrease the rate constants as it is to increase them.
Figure shows that the absolute uncertainties in tropospheric
O3 burden and OH global mean concentrations vary for the top 10
reactions for both increasing and decreasing the rate constant. Although
there are some differences between the impact of increasing or decreasing the
rate constant, there is a degree of consistency between the two, and so for
simplicity reasons we only consider further the impact of increasing the rate
constants.
Given the uncertainties for the individual reactions calculated here, the
next question is how these uncertainties can be combined together to
generate a single uncertainty from rate constants' uncertainty on the
composition of the atmosphere.
Addition of uncertainties
If these perturbations are independent (uncertainties in one rate constant
are not related to uncertainties in another) and the model approximately
linear, the total rate constant uncertainty can be found by finding the root
of the sum of the individual uncertainties squared (addition in quadrature)
as shown in Eq. ().
σtotal2=Σσreaction2
It is hard to assess the independence of the rate constants. Given the nature
of the laboratory experiments used to determine them, it is likely that there
is some overlap in assumptions. It would be extremely difficult to diagnose
this for all 60 reactions, and so we ignore this in further work.
Monte Carlo simulations to understand the models' linearity: the
x-axis values show the percentage change in the metric value of an
ensemble member compared to the simulation with no perturbations. The
y-axis values show the expected percentage change of the metric based on a
linear addition of the individual 1σ perturbation experiments weighted
by the Monte Carlo perturbation values. Metrics investigated are
(a) O3 tropospheric burden, (b) OH tropospheric burden,
(c) O3 mean surface concentration and (d) CH4
lifetime. We show the result of 50 Monte Carlo simulations. Each simulation
perturbs 10 of the most important reactions (∗ reactions in
Table ) 1σ by normally distributed random
numbers.
Atmospheric chemistry is though non-linear . A
doubling of a change to the model does not necessarily lead to a doubling of
the model response. Thus, is it not obvious how uncertainties from the
individual rate constant perturbations should be combined. To investigate
this, we perform a Monte Carlo analysis of the model. We take 10 of the most
significant reactions determined earlier (shown by the ∗ in
Table ) and generate 10 normally distributed random
numbers (μ=0, σ=1), one for each reaction. For each of the 10
rate constants, we add on the calculated 1σ uncertainty multiplied by
the random number and run the model. We repeat this 50 times to produce a
Monte Carlo ensemble from which we can calculate the four metrics described
earlier.
If the model is linear, the metrics calculated from each member of the Monte
Carlo ensemble should be (to some level) the same as the linear addition of
the individual rate constant perturbations weighted by the Monte Carlo random
numbers. Figure shows the perturbation in the
value of the metric calculated for each ensemble member against the
calculated value of the metric using the single reaction values. The model
shows a strong linear relationship between the metrics examined (intercepts
of 0.21 ± 0.9 % and gradients of 0.80 ± 0.04); thus, if the
errors are uncorrelated, we can, at least to a first approximation, add the
individual 1σ perturbations together in quadrature using
Eq. () to calculate the overall uncertainty in the model
metrics. From these simulations, we estimate the quadrature approach leads to
an over-estimate of the 1σ uncertainty on the order of 10 %.
We thus conclude that the adding together of the individual perturbations in
quadrature gives a good approximation to the uncertainty calculated by the
Monte Carlo method for significantly less computational burden.
Impacts on the present-day atmosphere metrics
We show in Fig. the absolute percentage change in global
annual mean O3 burden, surface O3, tropospheric average OH and
CH4 tropospheric lifetime from increasing each of the reaction rate
constants in Table in turn by their 1σ
value. They are ordered by the magnitude of the perturbation, and for clarity
we only show the top 20, combining the remaining 40 in quadrature into the
“Other” category. The fractional change in tropospheric O3 burden for all
of the perturbations is given in Table . We show the
results of combining all of these reactions in quadrature (“Total (sum)”),
the result of combining the top 10 in quadrature (“Top 10”) and the standard
deviation from the 50 Monte Carlo simulations (“Monte Carlo top 10”). The
relative closeness (∼ 10 %) of the value calculated from the
“Top 10” and the “Monte Carlo top 10” shows that the addition in quadrature
approach provides a useful approximation to the Monte Carlo methodology with
significantly less computational burden.
Spatial distribution of uncertainties: fractional uncertainties
calculated for O3, OH and CO concentrations for the tropospheric
column (a), the zonal mean (b) and the surface (c)
from adding together the individual reaction uncertainties from the
60 reactions studied in quadrature.
The top 10 reactions contribute over 90 % of the uncertainty for all metrics
with the overall uncertainty for the annual mean tropospheric ozone burden,
surface ozone and tropospheric OH concentrations, and tropospheric methane
lifetime calculated to be 10, 11, 16 and 16 %, respectively. These
uncertainties can be compared to the inter-model spreads found from model
intercomparison exercises. The multi-model standard deviations in the ozone
burden, tropospheric OH concentration and troposphere methane lifetime were
found to be 7, 10 and 10 % in the Atmospheric Chemistry and Climate Model
Intercomparison Project (ACCMIP) studies . Thus, we find that the chemical rate constant uncertainty
is larger than the multi-model spread, which is usually used to give some
sense of our uncertainty in our understanding of a quantity. As the models
used in these intercomparisons typically use the same rate constants, this
rate constant uncertainty is not included in the inter-model spread, and thus
the inter-model spread should be considered a lower estimate.
Primary VOCs: total 1σ uncertainty in the concentrations of
C2H6, C3H8, PRPE (≥ C3 alkenes), ALK4 (≥ C4 alkanes)
and ISOP (isoprene) from the addition in quadrature of the individual
reaction uncertainties. “Column” covers the tropospheric column.
Other organics: total 1σ uncertainty in the concentrations of
CH2O, MP (methyl hydroperoxide), ALD2 (acetaldehyde), GLYC
(glycolaldehyde), MACR (methacrolein) and MKV (methyl vinyl ketone) from the
addition in quadrature of the individual reaction uncertainties. “Column”
covers the tropospheric column.
NOx: total 1σ uncertainty in the concentrations of NO,
NO2, NO3, N2O5, HNO2 and HNO4 from the addition in
quadrature of the individual reaction uncertainties. “Column” covers the
tropospheric column.
NOy: total 1σ uncertainty in the concentrations of
HNO3, PAN (peroxyacetyl nitrate), PPN (peroxymethacryloyl nitrate), PMN
(peroxymethacryloyl nitrate) and NIT (inorganic aerosol nitrates) from the
addition in quadrature of the individual reaction uncertainties. “Column”
covers the tropospheric column.
Sulfur and aerosols: total 1σ uncertainty in the
concentrations of SO2, SO42-, DMS (dimethyl sulfide) and NH4+
from the addition in quadrature of the individual reaction uncertainties.
“Column” covers the tropospheric column.
Inorganics: total 1σ uncertainty in the concentrations of
H2O2, O3, OH, CO and HO2 from the addition in quadrature of the
individual reaction uncertainties. “Column” covers the tropospheric column.
Spatial distribution of uncertainty
Figure shows the spatial distribution of the
total uncertainty in the annual mean O3, OH and CO concentrations for the
tropospheric column, the zonal mean and at the surface from the 60 reactions. Similar plots for a large number of other model species are shown
in Figs. –. There is a significant
degree of inhomogeneity in these uncertainties which respond to a range of
factors. The uncertainties in the rate constants are largest in the upper
troposphere where the temperatures are coldest and thus furthest from the
298 K base temperature used to calculate the uncertainties. However, these
uncertainties can only manifest if chemistry is the large source or sink for
a species in that region. O3 uncertainties are relatively low in the upper
troposphere, as it has a large stratospheric source in this region which we
have not perturbed (see Sect. 2). OH uncertainties on the other hand are
high (30 %) in the upper troposphere due to the low temperatures. Over
continental regions, the concentration of CO is not particularly uncertain as
the emissions and transport control the concentration. However, over the
ocean where emissions are small, the chemistry becomes more important and so
uncertainty increases. Uncertainties in CO are largest in the Southern
Hemisphere where direct emission is lower and chemical production from CH4
and other hydrocarbons is significant. In general, uncertainties are largest
over remote regions far from recent emissions, especially if they are
particularly cold or hot compared to room temperature. Thus, surface OH values
are more uncertain in the cold remote Southern Ocean than they are in the
tropics. Surface O3 values are uncertain in the warm tropics where intense
sunlight and high water vapour concentrations lead to a large chemical flux
through O3.
Across the full set of simulated compounds
(Figs. –), there are even larger
uncertainties. For primary emitted hydrocarbons, large uncertainties occur in
remote, photochemically active locations such as the tropics where shorter-lived
hydrocarbons may be many OH lifetimes away from sources. Uncertainties
in the OH concentrations thus multiply in these regions, leading to
uncertainties of up to 60 % for ≥ C4 alkanes, for example. Secondary
products such as H2O2 and CH3OOH also show significant uncertainties of
up to 56 % in some locations.
Impact on model/measurement comparisons: modelled (red) and measured
(black) annual cycle in monthly mean O3, CO, C2H6, C3H8, ALK4
(≥ C4 alkanes) and NO2 mixing ratios at Cabo Verde
. Shaded area represents the 1σ uncertainty from
the 60 reactions added together in quadrature.
NOx concentrations close to emission sources are dominated by the
emission and transport and thus are not very sensitive to chemical
uncertainty (Fig. ). However, away from these emissions,
uncertainties can build up. Uncertainties in the NOx concentrations at the
poles are up to a factor of 40 %. Uncertainties in PAN concentrations
(Fig. ) are in general high (> 20 %) in most
locations (∼ 50 % over the remote ocean), reflecting the complexity
of the chemistry involving uncertainties in both ROx and NOx
concentrations. Uncertainties in nitric acid (the dominant NOx sink)
concentrations are smaller however (∼ 5 %), reflecting the mass
balance constraint of emissions of NOx having to balance NOy sinks.
Large variability in nitric acid concentrations in the Southern Ocean
reflects non-linearities in aerosol thermodynamics of HNO3/ NO3-
partitioning.
SO2 concentrations show the largest uncertainties in the tropical upper
troposphere where OH is also highly uncertain. However, SO42- shows
much smaller uncertainty, again reflecting mass conservation constraints.
NH4+ concentrations show little sensitivity to the rate constants
analysed. Overall, this suggests that aerosol mass is not particularly
sensitive to the gas-phase chemistry examined here.
Ozone site comparison: modelled (red) and measured (black)
concentrations of ozone at a range of sites. The pink shaded area shows the
1σ uncertainty from the chemical kinetics. The error bars represent
the 1σ uncertainty of these observations. Monthly mean observational
data were obtained from , using multiple years between
2004 and 2010 to create more complete datasets.
Ozonesondes: comparisons between the variability of annual
ozonesonde measurements and model data with uncertainties. The black line
shows the annual mean observation data and the shaded gray area shows the
range of data. The red line shows the model data and the pink shaded line
shows the chemical 1σ uncertainty. Observations are obtained from
.
Overall, we see a complex pattern of uncertainty with geographically highly
variable uncertainty.
Implications for model-measurement comparisons
Comparisons between the predictions made by models and observations underpin
the assessment of model fidelity. Deviations between the model and measurements
are often used to diagnose model failings. Attributing these differences to
uncertainties in the emissions is particularly popular (see, for
example, ).
Figure shows observed monthly mean and standard
deviations for CO, O3, C2H6, C3H8, C4H10 and NO2
from the World Meteorological Organisation's Global Atmosphere Watch Cape
Verde Atmospheric Observatory , overlaid with the base
model simulation and the chemical uncertainty (1σ) calculated from the
addition in quadrature of the 60 1σ simulations. We chose this
location as it is far from recent emissions and thus should show large
uncertainties for primary emitted species.
Consistent with Figs. –, the uncertainty
in the model calculation ranges from 5 to 30 % depending upon the species. For
some of the species (CO, O3, C2H6, C4H10), much of the
difference between the model and the measurements lies within the model
1σ uncertainty. For others, such as C3H8 or NO2, the
differences are harder to explain and other processes (emissions, transport,
unknown chemistry, etc.) would need to be explored.
Figure shows a comparison of the O3 measured
at a number of locations around the world and the model.
The shaded areas show the 1σ uncertainty due the kinetics for the
model and the 1σ standard deviation in the measurements. The
uncertainty in the model varies between 1 and 5 ppbv depending on location.
In some locations, the model uncertainty falls within the measured values. In
others, there are significant deviations.
Figures – show significant changes in
uncertainty with changes in the vertical due to increasing uncertainty with
reducing temperature. Figure shows a selection of
ozonesonde observations from the World Ozone and Ultraviolet Data
Centre () compared to equivalent modelled concentrations and
uncertainties. Observations are derived from the surface into the middle
troposphere as the temperature drops. The uncertainty thus maximises at
around 10 km. Above this, much of the ozone in the model is produced in the
stratosphere, which is unperturbed in these simulations. Above this height, the
uncertainty in the ozone due to tropospheric chemistry uncertainty reduces.
These comparisons with observations highlight the complexity of attributing
model failure to a particular cause. For some locations and some species,
the chemical uncertainty can be large. For the same species, in a different
location, the uncertainties may be much smaller. Inversion studies which
attempt to attribute model failure to a single cause (for example,
uncertainties in emissions) need to have a detailed understanding of the
magnitude and geographical distribution of the other model errors. We show
here that they vary between different species and can be large and highly
spatially varying. This should be considered when model inversion studies are
undertaken.
Uncertainties in O3 radiative forcing: absolute fractional
uncertainty in tropospheric O3 radiative forcing between the pre-industrial
and present-day simulations due to rate constant uncertainty. Shown on the left are the
20 most important reactions. “Other” shows the addition in quadrature of
the remaining 40 reactions. “Total (sum)” indicates the total fractional
uncertainty calculated by adding together the individual uncertainties in
quadrature. “ACCMIP” indicates the inter-model spread found from the ACCMIP
study.
Ozone radiative forcing
We repeat the 60 1σ simulations described above with pre-industrial
(notionally, the year 1850) emissions (see Sect. 2) to allow us to calculate
an uncertainty in the radiative forcing of O3. For each reaction, we
calculate the difference in the annual mean tropospheric column O3 (Dobson
units) between the present-day and pre-industrial simulations with the rate constant
increased to its 1σ value. Then, using a linear relationship between
change in O3 column and radiative forcing
of 42 mW m-2 DU-1, we
calculate a radiative forcing associated with the uncertainty associated with
each reaction. We estimate an overall uncertainty in the tropospheric O3
radiative forcing in the same way as the other metrics, by adding them
together in quadrature. In our base simulations, we calculated the
tropospheric O3 radiative forcing to be 412 mW m-2, consistent with
previous studies (410±65 mW m-2)
. Our estimate of the uncertainty in the
O3 radiative forcing from rate constant uncertainty is 56 mW m-2
(15 %), with reaction specific detail shown in
Fig. . Again, the same set of reactions
contribute the largest share to the uncertainty in the radiative forcing as
in the uncertainty in present-day O3 burden.
This uncertainty estimate of 15 % can be compared to the 17 % spread in the
O3 radiative forcing calculated between climate models in the recent
ACCMIP intercomparison (shown in
Fig. ). This spread is usually used as the
uncertainty in our understanding of O3 radiative forcing. However, as all
of these models use the same JPL or IUPAC recommended rate constants, the
inter-model spread does not include the rate constant uncertainty explored
here. Given that the rate constant uncertainty is comparable to the
inter-model spread, it should be included in future assessments of the
uncertainty in O3 radiative forcing. A naive addition in quadrature
approach would suggest that the uncertainty on tropospheric O3 radiative
forcing should be increased by roughly 30 % to account for this.
Conclusions
We have shown that the uncertainty in the inorganic rate constants leads to
significant (> 10 %) uncertainties in the concentration of policy-relevant
metrics of troposphere composition (O3 burden, surface O3, global
mean OH, tropospheric CH4 lifetime, O3 radiative forcing) with
significantly higher uncertainty in other compounds. This uncertainty may
have implications for climate policy through an underestimate of the
uncertainty on O3 radiative forcing or significant uncertainties on the
CH4 lifetime. This also has implication for how model-measurement
disagreements are interpreted. Similar conclusions have been found for
regional air quality focussed models .
The simulations performed here likely provide a lower limit to the chemical
uncertainty. We do not explore the impact in uncertainties in organic
chemistry (beyond that from the initiation of hydrocarbon oxidation) or in
organic mechanisms; we do not included tropospheric bromine, iodine or chlorine
chemistry in our analysis or the parameters that define heterogeneous
reaction rates. We have neither investigated the impact of rate constant
uncertainty on the composition of the stratosphere nor mesosphere, or how
this may propagate through to the troposphere. There are also uncertainties
in the Henry's law constants used for wet and dry parameterisations, etc. It
seems likely therefore that the true chemical uncertainty in the composition
of the atmosphere is significantly higher than that found here.
Although it may be challenging, reducing these uncertainties would provide
significant benefits. Targeting the top 10 reactions identified here
(Fig. a) would significantly reduce the overall chemical
uncertainties. Despite the fact that the rate constants for these reactions
may appear “decided”, they provide the basis for determining the composition
of the atmosphere. Given the difficulties in reducing the uncertainties in
other areas of the climate system (we will never know the pre-industrial
emissions well, etc.), a redoubled effort to reduce rate constant uncertainty
appears to be a relatively straightforward methodology to improve our
understanding of atmospheric composition.
The GEOS-Chem model is available from
http://www.geos-chem.org. The updates made to the code for this project
are available from 10.15124/4d161daa-ffc4-410b-a4b6-600615b29679.
The authors declare that they have no conflict of
interest.
Acknowledgements
Ben Newsome was supported by a NERC studentship (NE/L501761/1). This work was
supported by the NERC-funded BACCHUS project (NE/L01291X/1). The Cape Verde
Atmospheric Observatory is supported by the NERC-funded ORC3 project
(NE/K004980/1) and by the National Centre for Atmospheric Science. GEOS-Chem
(http://www.geos-chem.org) is a community effort, and we wish to thank all involved
in the development of the model. We would also thank all the JPL and IUPAC
panels for their efforts in compiling atmospheric rate constants.
Edited by: Thomas von Clarmann
Reviewed by: Rolf Sander and one anonymous referee
ReferencesAtkinson, R., Baulch, D. L., Cox, R. A., Crowley, J. N., Hampson, R. F.,
Hynes, R. G., Jenkin, M. E., Rossi, M. J., and Troe, J.: Evaluated kinetic
and photochemical data for atmospheric chemistry: Volume I – gas phase
reactions of Ox, HOx, NOx and SOx species, Atmos. Chem.
Phys., 4, 1461–1738, 10.5194/acp-4-1461-2004, 2004.Bey, I., Jacob, D. J., Yantosca, R. M., Logan, J. A., Field, B. D., Fiore,
A. M., Li, Q., Liu, H. Y., Mickley, L. J., and Schultz, M. G.: Global
modeling of tropospheric chemistry with assimilated meteorology: Model
description and evaluation, J. Geophys. Res.-Atmos., 106, 23073–23095,
10.1029/2001JD000807, 2001.Burkholder, J. B., Sander, S. P., Abbatt, J., Barker, J. R., , Huie, R. E.,
Kolb, C. E., Kurylo, M. J., Orkin, V. L., Wilmouth, D., and Wine, P.:
Chemical Kinetics and Photochemical Data for Use in Atmospheric Studies,
Evaluation No. 18, Jet Propulsion Laboratory, available at:
http://jpldataeval.jpl.nasa.gov/ (last access: 1 October 2017), 2015.Carpenter, L. J., Fleming, Z. L., Read, K. A., Lee, J. D., Moller, S. J.,
Hopkins, J. R., Purvis, R. M., Lewis, A. C., Müller, K., Heinold, B.,
Herrmann, H., Fomba, K. W., van Pinxteren, D., Müller, C., Tegen, I.,
Wiedensohler, A., Müller, T., Niedermeier, N., Achterberg, E. P., Patey,
M. D., Kozlova, E. A., Heimann, M., Heard, D. E., Plane, J. M. C., Mahajan,
A., Oetjen, H., Ingham, T., Stone, D., Whalley, L. K., Evans, M. J., Pilling,
M. J., Leigh, R. J., Monks, P. S., Karunaharan, A., Vaughan, S., Arnold,
S. R., Tschritter, J., Pöhler, D., Frieß, U., Holla, R., Mendes,
L. M., Lopez, H., Faria, B., Manning, A. J., and Wallace, D. W. R.: Seasonal
characteristics of tropical marine boundary layer air measured at the Cape
Verde Atmospheric Observatory, J. Atmos. Chem., 67, 87–140,
10.1007/s10874-011-9206-1, 2011.Dockery, D. W., Pope, C. A., Xu, X., Spengler, J. D., Ware, J. H., Fay,
M. E.,
Ferris, B. G. J., and Speizer, F. E.: An Association between Air Pollution
and Mortality in Six U.S. Cities, N. Engl. J. Med., 329, 1753–1759,
10.1056/NEJM199312093292401, 1993.Evans, M. J. and Newsome, B.: Model Code associated with Newsome and Evans
Impact of uncertainties in inorganic chemical rate constants on tropospheric
composition and ozone radiative forcing,
10.15124/4d161daa-ffc4-410b-a4b6-600615b29679, 2017.Goldstein, A. H. and Galbally, I. E.: Known and Unexplored Organic
Constituents in the Earth's Atmosphere, Environ. Sci. Technol.,
41, 1514–1521, 10.1021/es072476p, 2007.Hartley, D. and Prinn, R.: Feasibility of determining surface emissions of
trace gases using an inverse method in a three-dimensional chemical transport
model, J. Geophys. Res.-Atmos., 98, 5183–5197, 10.1029/92JD02594,
1993.Huang, J., Golombek, A., Prinn, R., Weiss, R., Fraser, P., Simmonds, P.,
Dlugokencky, E. J., Hall, B., Elkins, J., Steele, P., Langenfelds, R.,
Krummel, P., Dutton, G., and Porter, L.: Estimation of regional emissions of
nitrous oxide from 1997 to 2005 using multinetwork measurements, a chemical
transport model, and an inverse method, J. Geophys. Res.-Atmos., 113, D17313,
10.1029/2007JD009381, 2008.Hudman, R. C., Moore, N. E., Mebust, A. K., Martin, R. V., Russell, A. R.,
Valin, L. C., and Cohen, R. C.: Steps towards a mechanistic model of global
soil nitric oxide emissions: implementation and space based-constraints,
Atmos. Chem. Phys., 12, 7779–7795, 10.5194/acp-12-7779-2012,
2012.
IPCC: Climate Change 2013: The Physical Science Basis. Contribution of
Working Group I to the Fifth Assessment Report of the Intergovernmental Panel
on Climate Change, edited by: Stocker, T. F., Qin, D., Plattner, G.-K.,
Tignor, M., Allen, S. K., Boschung, J., Nauels, A., Xia, Y., Bex, V., and
Midgley, P. M., Cambridge University Press, Cambridge, UK and New York, NY,
USA, 1535 pp., 2013.
Jacobson, M. Z. and Turco, R. P.: SMVGEAR: A sparse-matrix, vectorized Gear
code for atmospheric models, Atmos. Environ., 28, 273–284, 1994.Lin, X., Trainer, M., and Liu, S. C.: On the nonlinearity of the tropospheric
ozone production, J. Geophys. Res.-Atmos., 93, 15879–15888,
10.1029/JD093iD12p15879, 1988.McLinden, C., Olsen, S., Hannegan, B., Wild, O., Prather, M., and Sundet, J.:
Stratospheric ozone in 3-D models: A simple chemistry and the
cross-tropopause flux, J. Geophys. Res., 105, 14653–14665, 10.1029/2000JD900124, 2000.Murray, L. T., Jacob, D. J., Logan, J. A., Hudman, R. C., and Koshak, W. J.:
Optimized regional and interannual variability of lightning in a global
chemical transport model constrained by LIS/OTD satellite data, J. Geophys.
Res.-Atmos., 117, D20307, 10.1029/2012JD017934, 2012.Park, R. J., Jacob, D. J., Chin, M., and Martin, R. V.: Sources of
carbonaceous
aerosols over the United States and implications for natural visibility, J.
Geophys. Res.-Atmos., 108, 4355, 10.1029/2002JD003190, 2003.Parrella, J. P., Jacob, D. J., Liang, Q., Zhang, Y., Mickley, L. J., Miller,
B., Evans, M. J., Yang, X., Pyle, J. A., Theys, N., and Van Roozendael, M.:
Tropospheric bromine chemistry: implications for present and pre-industrial
ozone and mercury, Atmos. Chem. Phys., 12, 6723–6740,
10.5194/acp-12-6723-2012, 2012.Sherwen, T., Evans, M. J., Carpenter, L. J., Andrews, S. J., Lidster, R. T.,
Dix, B., Koenig, T. K., Sinreich, R., Ortega, I., Volkamer, R., Saiz-Lopez,
A., Prados-Roman, C., Mahajan, A. S., and Ordóñez, C.: Iodine's
impact on tropospheric oxidants: a global model study in GEOS-Chem, Atmos.
Chem. Phys., 16, 1161–1186, 10.5194/acp-16-1161-2016, 2016.Sindelarova, K., Granier, C., Bouarar, I., Guenther, A., Tilmes, S.,
Stavrakou, T., Müller, J.-F., Kuhn, U., Stefani, P., and Knorr, W.:
Global data set of biogenic VOC emissions calculated by the MEGAN model over
the last 30 years, Atmos. Chem. Phys., 14, 9317–9341,
10.5194/acp-14-9317-2014, 2014.Sofen, E. D., Alexander, B., and Kunasek, S. A.: The impact of anthropogenic
emissions on atmospheric sulfate production pathways, oxidants, and ice core
Δ17O(SO42-), Atmos. Chem. Phys., 11, 3565–3578,
10.5194/acp-11-3565-2011, 2011.Sofen, E. D., Bowdalo, D., Evans, M. J., Apadula, F., Bonasoni, P., Cupeiro,
M., Ellul, R., Galbally, I. E., Girgzdiene, R., Luppo, S., Mimouni, M.,
Nahas, A. C., Saliba, M., and Tørseth, K.: Gridded global surface ozone
metrics for atmospheric chemistry model evaluation, Earth Syst. Sci. Data, 8,
41–59, 10.5194/essd-8-41-2016, 2016.Stevenson, D. S., Young, P. J., Naik, V., Lamarque, J.-F., Shindell, D. T.,
Voulgarakis, A., Skeie, R. B., Dalsoren, S. B., Myhre, G., Berntsen, T. K.,
Folberth, G. A., Rumbold, S. T., Collins, W. J., MacKenzie, I. A., Doherty,
R. M., Zeng, G., van Noije, T. P. C., Strunk, A., Bergmann, D.,
Cameron-Smith, P., Plummer, D. A., Strode, S. A., Horowitz, L., Lee, Y. H.,
Szopa, S., Sudo, K., Nagashima, T., Josse, B., Cionni, I., Righi, M., Eyring,
V., Conley, A., Bowman, K. W., Wild, O., and Archibald, A.: Tropospheric
ozone changes, radiative forcing and attribution to emissions in the
Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP),
Atmos. Chem. Phys., 13, 3063–3085, 10.5194/acp-13-3063-2013,
2013.
van der Werf, G. R., Randerson, J. T., Giglio, L., Collatz, G. J., Mu, M.,
Kasibhatla, P. S., Morton, D. C., DeFries, R. S., Jin, Y., and van Leeuwen,
T. T.: Global fire emissions and the contribution of deforestation, savanna,
forest, agricultural, and peat fires (1997–2009), Atmos. Chem. Phys., 10,
11707–11735, 10.5194/acp-10-11707-2010, 2010.Voulgarakis, A., Naik, V., Lamarque, J.-F., Shindell, D. T., Young, P. J.,
Prather, M. J., Wild, O., Field, R. D., Bergmann, D., Cameron-Smith, P.,
Cionni, I., Collins, W. J., Dalsøren, S. B., Doherty, R. M., Eyring, V.,
Faluvegi, G., Folberth, G. A., Horowitz, L. W., Josse, B., MacKenzie, I. A.,
Nagashima, T., Plummer, D. A., Righi, M., Rumbold, S. T., Stevenson, D. S.,
Strode, S. A., Sudo, K., Szopa, S., and Zeng, G.: Analysis of present day and
future OH and methane lifetime in the ACCMIP simulations, Atmos. Chem. Phys.,
13, 2563–2587, 10.5194/acp-13-2563-2013, 2013.WOUDC: WOUDC Ozone Monitoring Community, World Meteorological Organization –
Global Atmosphere Watch Program (WMO-GAW), 10.14287/10000001, 2017.Yang, Y.-J., Wilkinson, J. G., Talat Odman, M., and Russell, A. G.: Air
Pollution Modeling and Its Application XIII, Ozone Sensitivity and
Uncertainty Analysis Using DDM-3D in a Photochemical Air Quality Model,
Springer US, Boston, MA, USA, 183–194, 10.1007/978-1-4615-4153-0_19,
2000.Young, P. J., Archibald, A. T., Bowman, K. W., Lamarque, J.-F., Naik, V.,
Stevenson, D. S., Tilmes, S., Voulgarakis, A., Wild, O., Bergmann, D.,
Cameron-Smith, P., Cionni, I., Collins, W. J., Dalsøren, S. B., Doherty,
R. M., Eyring, V., Faluvegi, G., Horowitz, L. W., Josse, B., Lee, Y. H.,
MacKenzie, I. A., Nagashima, T., Plummer, D. A., Righi, M., Rumbold, S. T.,
Skeie, R. B., Shindell, D. T., Strode, S. A., Sudo, K., Szopa, S., and Zeng,
G.: Pre-industrial to end 21st century projections of tropospheric ozone from
the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP),
Atmos. Chem. Phys., 13, 2063–2090, 10.5194/acp-13-2063-2013,
2013.