Air Quality Extreme Events
and Projected Trends for the
Southwestern United States
Erika Wise
University of Arizona
Climate Prediction Applications Science Workshop
March 22, 2006
Tucson, AZ
Project Context
‹
‹
Established to assess impacts of climate variability
and change on human and natural systems in the
Southwest
Air quality initiative: help air quality planners and
managers understand links between climate and
pollutants to improve decision-making capabilities
Central Phoenix Station, Phoenix, AZ
Previous study:
meteorologicallyadjusted, long-term
trends (Wise and Comrie
2005)
Model 1 ( - Temp)
Model 2 ( - Temp/MH)
Model 3 ( -Temp/MH/AWND)
Unadjusted Long-Term Ozone
51
49
47
Ozone (ppb)
‹
53
45
43
41
39
37
35
33
1991
1992
1993
1994
1995
1996
Year
1997
1998
1999
2000
2001
Research Needs
‹
Focus on ozone and
PM
– NAAQS regulations
– Known detrimental
effects (health,
environment, visibility)
Source: Liverman and Merideth 2002
‹
Cities in the Southwest are often close to
violating federal standards
– Local climate and weather conditions often
determine whether levels are exceeded
‹
Probabilities of exceedances, now and in
the future
Objectives
Meet research needs regarding extreme
events and future conditions by:
‹
‹
Characterizing ozone and PM air quality
exceedances under current conditions
using extreme value methods
Downscaling climate model scenarios to
determine probable changes in ozone/PM
meteorology and resulting changes in
return levels
Background: Predicted Changes
Climate
– Warmer: 5-6°C
projected for West
by 2100
– Less certainty with
precipitation; more
model-dependent
‹
Temperature
Air quality
– Ozone-focused
– Globally: uncertain
– Western US:
higher ozone
(Prather et al. 2002; Leung
and Gustafson 2005)
Precipitation
www.usgcrp.gov
‹
Data
Ozone and PM data
– Local, state, and
federal environmental
agencies
– Ozone: maximum
daily 8-hr average
– PM: 24-hour average
PM10
‹
GCM scenarios: Hadley Centre’s HadCM3;
SRES A2 and B2 scenarios
‹
NCEP reanalysis meteorological variables
‹
Tucson, AZ
www.ipcc.ch
‹
Methods: Extreme Values
‹
Central Limit vs. Extreme Types Theorem
– Normal vs. GEV/GP distributions
‹
Used to estimate:
– Return Period
that
threshold is exceeded
in a given time period
– Return Level
‹ Magnitude
of the
return period
‹
Applied using R-source ExtRemes
(Gilleland and Katz 2004 / NCAR)
www.isse.ucar.edu
‹ Probability
Methods: Statistical DownScaling Model
(SDSM - Wilby et al. 2002)
Combines stochastic
weather generator and
regression-based
approaches
‹ Model calibrated using
observational predictors /
predictand
‹ Modeled relationships and
GCM predictors generate future scenarios
‹ 1990-2001 / 2002-2050 / 2051-2099
www-esd.lbl.gov
‹
Results: Observed Extremes
Shape
parameter
= -0.24
(beta)
Shape
parameter
= 0.12
(Pareto)
Results: Ozone Return Levels
(Current Conditions)
Return Period
(yrs)
Return Level
(ppb)
Confidence
Interval (ppb)
1
81
80-82
10
86
85-89
100
90
87-93
Results: PM Return Levels
(Current Conditions)
Return Period
(yrs)
Return Level
(µg/m3)
Confidence
Interval (µg/m3)
1
100
93-109
10
146
128-167
100
207
173-241
Results: SDSM Calibration and Verification
‹
Calibration: 1990-1995
– Observed predictand (ozone/PM) and NCEP
climate predictors
‹
Verification: 1996-2001
– Observed and modeled output
NCEP
Predictor
Variables
Tucson Ozone
Tucson PM
500 hPa divergence
Mean sea level pressure
850 hPa airflow strength
500 hPa geopotential height
Relative humidity at 500hPa
Relative humidity at 850hPa
Near surface relative humidity
Near surface relative humidity
Surface specific humidity
Mean temp at 2m
2
R
0.348
0.277
Results: Modeled Future Ozone
70
70
Dec
Nov
Oct
Sep
Jan
Dec
Maximum Ozone
10
A2 2051-2099
B2 2051-2099
ppb
5
www.ipcc.ch
Nov
Oct
Sep
Aug
Jul
30
Jun
30
May
40
Apr
40
Mar
50
Feb
50
Aug
60
Jul
60
Jun
ppb
80
May
Baseline
80
Jan
B2
Baseline
90
Apr
90
ppb
100
B2
Mar
100
Maximum Ozone
2051-2099
Feb
Mean Ozone
2051-2099
0
-5
-10
Winter
Spring
Summer
Autumn
Results: Modeled Future PM
Mean PM
2002-2050
36
34
32
32
30
30
µg/m3
34
28
26
28
26
24
B2
24
B2
22
Baseline
22
Baseline
-5
-10
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Dec
Nov
Oct
Sep
Dec
0
Aug
Apr
Nov
B2 2051-2099
Jul
Mar
Oct
B2 2002-2050
5
Jun
Feb
Sep
Mean PM
May
Jan
Dec
Nov
Oct
Sep
Aug
Jul
Jun
20
10
µg/m3
May
Apr
Mar
Feb
20
Jan
µg/m
3
36
Mean PM
2051-2099
Results: Modeled Ozone Extremes
Return Period
(yrs)
1990-2001 Return
Level (CI)
2002-2050 Return
Level (CI)
2051-2099 Return
Level (CI)
1
79 (78-80)
81 (81-82)
87 (86-88)
10
85 (83-87)
89 (88-91)
96 (94-98)
100
88 (86-92)
95 (92-97)
103 (100-106)
parts per billion
5
2002-2050 Return 2051-2099 Return
Return Period
Level Increase
Level Increase
(yrs)
from Baseline
from Baseline
2.53%
10.13%
10
4.71%
12.94%
100
7.95%
17.05%
4.12
4
ppb
1
Ozone Exceedances of 80 ppb
Threshold per Year
3
Modeled
Observed
2
1
0.75
0.98
1.16
0
1990-2001
2002-2050
Time Period
2051-2099
Results: Modeled PM Extremes
Return Period
(yrs)
1990-2001
Return Level
(CI)
2002-2050
Return Level
(CI)
2051-2099
Return Level
(CI)
1
85 (79-94)
81 (79-83)
83 (82-86)
10
132 (110-155)
105 (99-115)
107 (101-114)
100
213 (168-258)
133 (118-148)
131 (120-143)
3
µg/m
Conclusions
Characterization of air quality exceedances
under current climate conditions:
‹
Ozone:
– 1-yr return period for exceedances (80 ppb)
– 100-yr return level = 90 ppb
‹
PM:
– 10-yr return period for exceedances (150
ug/m3)
– 100-yr return level = 207 µg/m3
Conclusions
Downscaled GCM projections applied to air quality:
‹
SDSM models correspond well with observed
validation period air quality
– Mean, max ozone
– Mean PM
‹
Ozone
‹
PM
– Monthly means increase 4-5 ppb in summer and autumn
– Increases in summer seasonal max up to 10 ppb
– Summer monthly mean increases up to 9 µg/m3
– Winter decreases (but within confidence interval)
Conclusions
Modeled climate influence on future extreme
events:
‹
Ozone
‹
PM
– Increases in return levels at 1-yr (10%), 10-yr
(13%), and 100-yr (17%) return periods by
2099
– Quadrupling of exceedance rate/yr by 2099
– Projected decreases from 1990-2001 to 20022050
– No change 2002-2050 to 2051-2099
– 21st-century 100-yr return levels below
NAAQS
Future Research
‹
‹
Compare return levels and climate
sensitivity with other Southwest cities
Modification of PM
calibration period,
statistical methods, or
threshold for better
simulation of extremes
Incorporation of emissions scenarios /
chemical transport
seattlepi.nwsource.com
‹
Thank You!
Acknowledgments:
–
–
–
–
–
Climate Assessment for the Southwest project
National Oceanic and Atmospheric Administration
Dr. Andrew Comrie
Pima Department of Environmental Quality
National Center for Atmospheric Research
Image source: Gregg Townsend