United States
Environmental Protection
Agency
Atmospheric Research and
Exposure Assessment Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S3-89/068 Sept. 1989
&EPA Project Summary
Sensitivity of a Regional Oxidant
Model to Variations in Climate
Parameters
Ralph E. Morris, Michael W. Gery, Mei-Kao Liu, Gary E. Moore, Christopher
Daley, and Stanley M. Greenfield
The continued release of CO2 and
other trace gases has in recent years
led to the concern that these gases
will result in the global warming of
the atmosphere by blocking the
escape of outgoing infrared radiation.
The resultant increase in global tem-
perature due to this so-called
"greenhouse effect" may have far-
reaching consequences from raising
the sea level of the oceans to altering
land use patterns across the globe
and possibly increased photoc-
hemical smog formation in the lower
troposphere.
In order to investigate the
sensitivity of ozone concentrations to
future climate variations, a regional
oxidant model was applied for future
climate scenarios to two regions: one
covering central California (San
Joaquin Valley, Sierra Nevada moun-
tains and the San Francisco Bay
Area) and the other covering the
midwestern and southeastern United
States. Based on model calculations,
the effects of increased temperature
on ambient ozone concentrations
results in an increase of the area of
exceedances of the ozone air quality
standard, a movement of the peak
ozone concentration closer to the
urban areas, and the resultant
increase in the exposure of people to
harmful levels of ozone concentra-
tions. The calculations for California
indicate that the maximum daily
ozone concentrations may increase
from 2 to 20 percent and the number
of people exposed to hourly ozone
concentrations in excess of the air
quality standard may triple as a result
of a temperature increase. Similar,
although less dramatic, results were
seen for the midwestern and south-
eastern applications.
Past regional oxidant model simu-
lations were analyzed to relate input
meteorological variables to ozone
concentrations in order to infer the
possible effects of future climate
perturbations on ozone concen-
trations. Days with elevated ozone
concentrations were highly cor-
related with rainfall (negative cor-
relation) and solar intensity (positive
correlation). A weaker positive
correlation between temperature and
ozone concentrations also was
exhibited.
This Project Summary was devel-
oped by EPA's Atmospheric Research
and Exposure Assessment Laboratory,
Research Triangle Park, NC, to an-
nounce key findings of the research
project that is fully documented in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Global atmospheric changes are
expected to occur within the next several
decades because of increases in levels of
pollutants in the atmosphere. These
changes are associated with (1) warming
of the atmosphere due to the greenhouse
effect of trace gases, (2) depletion of the
stratospheric ozone layer, and (3) modi-
fication of tropospheric chemistry. The
emitted materials responsible for those
changes are carbon dioxide, carbon
monoxide, methane, chlorofluoro-hydro-
-------
carbons, nitrous oxide, and other trace
gases.
As atmospheric concentrations of the
so-called "greenhouse gases" continue
to increase, the potential climate change
and consequent environmental impacts
have become issues of great concern
worldwide. Such climate changes could
significantly affect the chemistry and
dynamics of the troposphere and
ultimately endanger human health and
sensitive ecological systems. Of partic-
ular concern to many parties, including
the Environmental Protection Agency
(EPA) and National Park service (NPS), is
the possibility that increased temperature
and depletion of stratospheric ozone may
result in increases in tropospheric
concentrations of ozone. The resulting
increased reactivity of the troposphere
would also result in increased concen-
trations of other photochemically active
species, such as hydrogen peroxide and
peroxyacetyl nitrate. Hydrogen peroxide
is believed to be one of the principal
chemicals in the formation of sulfate and
consequently acid deposition, by acting
as an oxidizer of sulfur dioxide, while
increased concentrations of ozone and
peroxyacetyl nitrate may result in signif-
icant damage to forest ecosystems.
To provide an initial estimate of the
possible effects of future climate changes
on tropospheric oxidant concentrations,
the EPA, through an interagency agree-
ment with the NPS, funded a preliminary
study that examined the effects of future
climate changes on urban air quality at
several U.S. cities. Specifically, this study
used a computer simulation model,
OZIPM3, a simple photochemical trajec-
tory box model, to study the effects of
increased temperature and decreased
stratospheric ozone on ozone formation.
The results indicated that, if anticipated
climate changes do occur, most of the
cities studied will not be able to meet the
current National Ambient Air Quality
Standard (NAAQS) for ozone (0.12 ppm)
without more emission controls than are
currently envisioned.
This preliminary study, however, did
not estimate the effects of future climate
changes on ozone concentrations in the
rural atmosphere. The simulation model
used in the study, OZIPM3, does not
completely simulate the complex inter-
action between the processes of
transport, diffusion, chemistry, and depo-
sition, nor does it include any feedback
mechanisms between these processes.
In addition, because the model does not
divide the atmosphere into more than one
vertical layer, it does not properly
account for emission, meteorological, and
chemistry variations with height, which
may be important in both urban and rural
environments. Therefore, the purpose of
the study reported here was to examine
the ability of another model to assess the
effects of future climate change on
tropospheric air quality. This model, the
RTM-III, is an Eulerian three-dimensional
regional oxidant model that has been
developed over several years.
Procedure
This study involved two tasks. One task
was to examine past RTM-III calculations
of regional ozone concentrations in order
to estimate the sensitivity of the model to
changes in climate. Ozone concentrations
predicted by the model were related to
meteorological input parameters in order
to gain insight into how potential future
perturbations of these meteorological
parameters will affect ozone concen-
trations. The other task was to determine
the sensitivity of the RTM-III to changes
in climate by simulating a base case of
current climate conditions and potential
future climate sensitivity scenarios.
To estimate the sensitivity of past
RTM-III calculations of ozone to climatic
change, past model simulations were re-
analyzed and the meteorological condi-
tions used as input were classified into
sets of variables. Measures of these vari-
ables were developed through spatial and
temporal averaging that would best relate
them to ozone concentrations. The
predicted ozone concentrations associ-
ated with 'these sets of meteorological
conditions were then examined to deter-
mine their sensitivity to variations in
climatic conditions.
An updated version of the RTM-III was
then applied to two modeling domains:
one covering central California and the
other covering the midwestern and south-
eastern United States. For each modeling
domain the model was exercised for a
base case of current temperature and
ultraviolet light conditions, and for a
future climate scenario reflecting the
effects of global warming. The model's
ozone concentration predictions were
then analyzed to determine the sensitivity
of the model's predictions of tropospheric
air quality to climate changes.
This study provides a preliminary
estimate of the sensitivity of calculations
of air quality to climate changes and
should help identify those climate param-
eters to which the calculations are most
sensitive. These results will help focus
future research on the effects of possible
changes in climate on air quality.
Results and Discussion
Application of the RTM-III for
Future Climate Scenarios
The continued release of emissions of
carbon dioxide (C02) and other trace
gases has in recent years led to the
concern that these trace gases will result
in a global warming of the atmosphere by
blocking the escape of thermal infrared
radiation. This phenomenon is commonly
referred to as the greenhouse effect. To
quantify the amount of global warming
expected in the future, global circulation
models (GCMs) and other climate models
have been exercised with various
estimates of future loadings of trace
gases in the upper atmosphere. We
analyzed the predictions of climate
change from four GCMs to obtain two
representative future climate scenarios: a
4°C temperature increase and a combin-
ation of a 4°C temperature increase with
a 10 percent reduction in stratospheric
ozone.
The future climate scenarios were
based on results from the following four
GCMs: (1) National Center for Atmos-
pheric Research (NCAR) Community
Climate Model (CCM) model; (2) National
Aeronautics and Space Administratio"
(NASA) Goddard Institute of Spai.
Studies (GISS) model; (3) National Oce-
anic and Atmospheric Administration
(NOAA) General Fluid Dynamics Labora-
tory (GFDL) model; and (4) Oregon State
University (OSU) model.
Current estimates of the emission and
retention of man-made C02 in the atmos-
phere indicate a distinct possibility that
atmospheric concentrations of C02 will
double within the next century. Under
these conditions, several GCMs predict
an increase in the global average surface
temperature at sea level of from 1 to 5°C.
The above four GCMs models predict
that the doubling of C02 concentrations
would increase the global average
temperature in the range of 2 to 5°C. The
presence of other trace gases, in addition
to CO2, would increase the global
warming further. Although the four GCMs
generally agree on the level of increase
of global average temperature, they do
not agree in their predictions of temper-
ature increases in specific regions, such
as California. Studies indicate that current
GCMs cannot, as yet, provide meaningful
results for specific regions of interest.
Because of these limitations, translating
the output of GCMs to hourly averap-
temperature increases for a speci\
region of interest, as required by a
-------
regional air quality model such as the
TM-HI, is problematic, the temporal and
-patial scales of a GCM (decades and
thousands of kilometers) are not compat-
ible with those of a model like RTM-III
(hours and 10 to 50 km). Because of
difficulties in adapting the GCM output to
a regional model and the discrepancies in
the predictions of the GCMs for many
climate variables, we have assumed
simply that the temperature increases by
4°C throughout the modeling region.
Three modeling scenarios were defined
to estimate the effect of future climate
changes on tropospheric ozone concen-
trations: (1) Base case—current meteoro-
logical and ozone column conditions; (2)
Scenario #1 — 4°C temperature increase
and attendant increase in water vapor
concentrations; and (3) Scenario #2 —4°C
temperature and water vapor increases
and a 10 percent reduction in strato-
spheric ozone concentrations.
The RTM-III was exercised for ozone
episodes of approximately one-week dur-
ation for the central California and the
midwestern/southeastern modeling do-
mains for the base case and scenario #1.
Due to limitations on time and resources
for this work, scenario #2 was modeled
only for four days from the mid-
western/southeastern modeling episode.
There are considerable uncertainties
osociated with assumptions in the
modeling. The uncertainties must be kept
in mind when interpreting the modeling
results. These uncertainties can be
roughly divided into three categories: (1)
uncertainty in the assumptions used to
define the climate change scenarios, (2)
uncertainties in the model inputs, and (3)
model limitations. The climate change
scenarios studied represent a very sim-
plistic description of future climate
perturbations. The temperature increase
was assumed to occur spatially and
temporally constant and independent of
Other meteorological (except water vapor
concentrations) and other variables (e.g.,
winds, mixing heights, emissions, etc.)
that are known to be interdependent.
However, the inclusion of these interde-
pendencies would require assumptions
that would introduce additional uncer-
tainties and confuse the analysis.
The effect of an increase in tempera-
ture tended to increase the predicted
maximum daily ozone concentrations im-
mediately downwind of the major urban
areas for each of the six days of the
central California modeling episode. In
more remote areas, Such as the Sierra
Nevada mountains, the increase in
mperature had no effect on the
maximum daily ozone concentrations.
The peak predicted maximum daily
ozone concentration increased from 3 to
20 percent due to the temperature
increase.
For the midwestern/southeastern appli-
cation of the RTM-III, the increase in
temperature had less of an effect on the
maximum daily ozone concentrations.
The maximum increase in the peak daily
ozone concentrations due to the increase
in temperature was 8 percent. Downwind
of the major urban areas (Chicago,
Detroit, St. Louis, etc.) ozone concentra-
tions tended to increase due to the
increase in temperature, whereas in other
areas there was no change and some-
times ozone concentrations decreased.
The effects of increased ultraviolet irra-
diance or temperature on tropospheric
ozone concentrations depends on the
oxidant-forming potential of the system.
This in turn is generally a function of
meteorological conditions and the effic-
iency of the atmospheric system in
converting oxidant precursors to oxidants.
For some atmospheric systems the in-
creased energy results in increased
reactivity in the morning hours, depleting
enough oxidant precursors from the
system to limit afternoon ozone produc-
tion to levels lower than the base case.
For climate change scenario #2 (increase
in temperature and decrease in the
stratospheric ozone column) ozone
concentrations tended to be higher or
lower than seen in the other scenarios
depending on the amount of precursors
present. These results are consistent with
previous studies that showed that under
conditions of increased temperature and
UV radiation the highest ozone concen-
trations were frequently lower in cities
with less oxidant precursors. This is
because the increased energy, due to an
increase in temperature and/or UV radi-
ation, burns out the oxidant precursors
earlier in the day, resulting in less oxidant
precursors in the afternoon, the period of
maximum ozone formation potential. The
rather coarse grid spacing used in the
midwestern/southeastern modeling do-
main (approximately 50 km on a side)
reduces the peak precursor concen-
trations from the urban areas because of
dilution in the large grid cells.
Although this preliminary model sensi-
tivity analysis may be useful in antic-
ipating the kinds of air quality controls
that may be needed in responding to
potential global climate changes, the
uncertainties associated with predicting
just how the climate may be modified
preclude any definitive discussion here of
regulatory controls. These preliminary
modeling results can only indicate pos-
sible general trends in exceedances of
the ozone standard and increases in the
number of people exposed to unhealthy
levels of ozone as a result of global
climate change.
The study indicates that immediately
downwind of urban areas increased
temperature tends to (1) increase ozone
concentrations, (2) move the location of
the peak ozone concentration closer to
urban areas, and (3) expand the area in
which ozone concentrations exceed the
primary ozone standard of 12 pphm.
Thus the modeling study indicates that
global warming will not only lead to more
exceedances of the primary ozone
standard over a larger area, but also to an
•increase in the number of people
exposed to these elevated ozone concen-
trations. Model calculations indicate that
approximately three times as many
people in the central California modeling
domain and 60 percent more people in
the midwestern/southeastern modeling
domain will be exposed to hourly ozone
concentrations in excess of the NAAQS
as a result of a 4°C temperature increase.
In addition, the modeling results suggest
that, with the increase in temperature,
people in central California will be
exposed to ozone concentrations in
excess of 16 pphm whereas under
current temperature conditions the
modeled ozone concentrations do not
exceed 16 pphm.
These results must be viewed with
caution. As discussed earlier, numerous
simplifying assumptions were made in
modeling the impacts of climate change
on ozone, and these assumptions add
significantly to the quantitative un-
certainty normally inherent in air quality
modeling. Some of these assumptions,
e.g., that the increase in temperature
occurs everywhere, will tend to overstate
the effects of increased temperature,
while others, e.g., that hydrocarbon
emissions do not increase under
increased temperature conditions, tend to
understate the impacts. The climate
change scenarios presented here are
simplistic and most likely do not
completely describe the changes in
climate associated with global warming.
The model's calculations of ozone under
the simplified climate change scenarios
discussed here should thus be viewed as
possible trends rather than as conclusive
impacts. The basic results of this
modeling exercise are that increases in
temperature will likely result in increases
in maximum daily ozone concentrations,
increases in the areas impacted by high
ozone concentrations, and increases in
the number of people exposed to
-------
unhealthy levels of ozone. Under these
circumstances, currently planned emis-
sion control requirements to achieve
attainment of the ozone standard may not
be sufficient.
Analysis of Historical RTM-II
Simulations to Infer the Effects
of Potential Climate Changes on
Ozone Concentrations
The analysis of past RTM-III simu-
lations consisted of several steps: (1)
Meteorological variables that potentially
influence ozone concentrations were
selected; (2) Measures of these variables
were developed, through spatial and
temporal averaging that would best relate
them to ozone concentrations; (3) The
measures were grouped into climatic
types that are associated with ozone
concentrations; and (4) The feasibility of
using such grouping schemes to predict
ozone concentrations was assessed. The
following meteorological variables were
analyzed: temperature, wind speed and
direction, water vapor mixing ratio, mixing
height, emission rates, exposure class,
precipitation rates, cloud cover, and the
radiative flux for photochemical reactions.
A number simple techniques for reduc-
ing the thousands of data points available
for each day to a dozen or so which have
the greatest predictive power for maxi-
mum daily ozone concentrations were
used: (1) Variables were combined into
composite indexes based on a knowl-
edge of chemistry, dispersion, and
emissions. For example, wind speed and
the mixing height can be multiplied into a
product which is sometimes known as the
Ventilation'; (2) Redundant data were
eliminated by combining or removing
intercorrelated variables and deempha-
sizing areas where there is no need for
representation by a classification variable;
(3) Data were systematically averaged in
time and space; and (4) Variables were
extracted from the model.
The data base for this analysis was
developed from three RTM-III simula-
tions: Eastern US Simulation (a domain
that encompasses nearly all of the
eastern United States and a large portion
of the Midwest; the modeling period was
15 August to 15 September 1978); Mid-
western/Southeastern US Simulation (a
domain that extends from the Great
Plains to just west of the mid-Atlantic
seaboard, and from the northern Great
Lakes area to central; three episodes
were available April 1980, July 1980, and
August 1980); Central California Simula-
tion (a domain is the smallest of the
three, covering an area roughly bounded
by the Pacific coast and Sierra Nevada,
and the San Francisco Bay Area and
Tehachapi Mountains; the six-day
episode occurred in August, 1981).
A correlational analysis was performed
to investigate relationships between (1)
ozone concentration and the meteorologi-
cal variables and (2) the meteorological
variables themselves. The strength of
relationships of the former type is a mea-
sure of the ability of a variable to predict
ozone concentration, while that of the
latter type is a measure of the degree of
redundancy within a set of variables. In
the northeastern region the maximum
daily ozone concentration is significantly
correlated (95% confidence level) to rain-
fall, maximum temperature, morning and
total daily solar radiation, daily average
and afternoon ventilation, and daily
average and afternoon wind speed. Most
highly correlated are morning and daily
total solar radiation (r = 0.56 and 0.53,
respectively) and rainfall (r = -0.51).
In the midwestern episode 1 (April 17-
25 1980), the maximum daily ozone is
significantly correlated (95% confidence
level) with daily average and maximum
temperature, daily average and maximum
water vapor mixing ratio, and daily total
and morning solar radiation. Most highly
correlated are daily average temperature
(r = 0.67) and daily total solar radiation (r
= 0.54). In the midwestern episode 2 (7-
21 July 1980), the relationships between
the meteorological variables and maxi-
mum daily ozone concentrations are rela-
tively weak. Those significant at the 95%
confidence level are daily total solar
radiation (r = 0.37), afternoon ventilation
(r = -0.30), and afternoon mixing height
(r = -0.29). In the midwestern episode 3
(8-18 August 1980), the maximum daily
ozone concentration is significantly corre-
lated (at the 95% confidence level) with
rainfall, daily average and maximum
water vapor mixing ratio, daily average
and maximum ventilation, and daily aver-
age mixing height. All correlation coef-
ficients were negative and relatively low;
none exceeded 0.34 (absolute value).
In the central California episode, no
meteorological variables are correlated
with maximum daily ozone at the 95%
confidence level. This may be due to the
small sample size (N = 12). The strong-
est relationship was that of daily average
ventilation and ozone, (r = -0.590). This
is typical for the region, where complex
terrain can restrict air flow and produce
an ozone episode if wind speeds and
mixing heights are low.
Overall, solar radiation appears to have
the strongest and most consistent rela-
tionship with ozone. One or both
measures of solar radiation (morning
total daily) were among the top thrt.
variables with the strongest relationship
with ozone in three out of the five
episodes analyzed. However, typically
clear skies during the ozone season in
California make solar radiation a poor
descriptor of ozone. Here the dominant
influence on air quality is the interaction
of mesoscale air flows with complex
terrain; thus, wind speed and ventilation
are highly related to ozone in California.
Other variables are occasionally highly
related to ozone concentrations; their
importance seems to be tied to the
magnitude of the variables involved. For
example, in the April 1980 midwestern
episode, there is a relatively strong
positive relationship between temperature
and ozone concentration. Temperatures
during this episode were much lower than
during the other two midwestern epi-
sodes, which occurred in summer. The
northeastern episode had the second
lowest average temperature and was the
only other episode in which temperature
was significantly correlated with ozone.
These results suggest that the lower the
temperature, the greater its influence on
ozone concentrations.
Rainfall is another example of a
variable whose magnitude influences ;
relationship with ozone. Average ar.
maximum rainfall were highest in the
northeastern episode and second highest
in the August 1980 midwestern episode;
these two episodes also exhibit the
strongest and second strongest relation-
ships between rainfall and ozone concen-
tration, respectively. From these results,
it appears that if rainfall occurs infre-
quently and in small amounts, it does
little to influence ozone concentrations;
conversely, when relatively large
amounts of rainfall occur, it has a strong
(negative) influence on ozone concen-
trations.
Conclusions and
Recommendations
This study explored the sensitivity of a
regional oxidant model, RTM-III, to
variations in atmospheric parameters in
an effort to establish the usefulness of
using a photochemical model to analyze
the impact on air quality of global climate
changes. The results of the study
indicate that the ozone concentrations
predicted by a complex model using
current atmospheric chemistry are
sensitive to the climate change scenario-
studied. Within the uncertainties prese
the modeling results suggest there could
-------
be potentially significant increases in
•otochemical pollutants due to future
..imate changes.
Given the preliminary nature of this
study, and its limitations, it is recom-
mended that in future studies the poten-
tial air quality impacts of global climate
change be examined in more detail.
While such details are impossible to
completely define because of the explor-
atory nature of such a study, it is possible
to provide an outline of possible
recommended approaches.
Include more complex climate change
scenarios. In the preliminary study
reported here only two meteorological
parameters were changed: UV intensity
at the surface and atmospheric temper-
ature. A third parameter, atmospheric
water vapor, was calculated as function of
temperature, assuming that the specific
humidity was held constant. Future
studies should examine an expanded set
of linked meteorological parameters
(wind, relative humidity, cloud type and
cover, precipitation, etc.) in addition to
those used in this preliminary exam-
ination.
Consider climatic feedback and con-
sistency in meteorological parameters. If
linked parameters are to be considered, it
is imperative that climatic feedback be
?ated in the simulations.
Broaden the range of climate pertur-
bations. To include all variations in
climate conditions contained in the GCM
scenarios, future studies should include a
broader range of meteorological param-
eters that are varied.
Analyze the impact of global change
on the effectiveness of current air quality
strategies.- The study reported here
examined future air quality conditions
assuming no change in precursor emis-
sions from today. Current regulatory
efforts to reduce ozone concentrations
involve reducing emissions of VOCs and
NOX, which could dramatically alter the
chemical mix of the atmosphere and
possibly the chemical response to poten-
tial changes in climate. Future studies
should examine the impacts of global
climate change on an environment that
more closely resembles the one that is
likely to result from the implementation of
possible control strategies.
Increase the number of regions and
meteorological conditions. This prelim-
inary study was severely restricted, by
time and resources, to an examination of
the climatic sensitivity of a model for two
specific ozone episodes in two regions of
the U.S. Future studies should increase
the number of regions and meteorological
scenarios analyzed, thereby permitting
somewhat more general conclusions
about the impact of global climate
change on air quality.
Extend the analysis of input data. The
analysis of the input data for past RTM-III
applications was both preliminary and
incomplete. Other applications of the
RTM-III and similar models as their
results become available (e.g. ROM,
RADM, DAM) should also be analyzed.
Extend the analyses to observed data.
The same statistical analysis as applied
to the RTM-III predictions could be
applied to observed ozone and meteor-
ological data.
Changes in the frequency of ozone
exceedances. The data analysis does not
address how a change in climate may
result in an increase in the number of
exceedances of the ozone air quality
standard. A methodology should be
developed that will relate changes in
meteorological parameters to changes in
the observed frequency of ozone exceed-
ances with some estimate of the uncer-
tainty included.
------- |