unt-nctcog-presentation-041714

30
An evaluation of the spatio-temporal characteristics of meteorologically-adjusted ozone trends in North Texas A preliminary analysis by Mahdi Ahmadi and Kuruvilla John Department of Mechanical and Energy Engineering University of North Texas Air Quality Technical Meeting, NCTCOG Arlington, TX. April 17, 2014

Transcript of unt-nctcog-presentation-041714

Page 1: unt-nctcog-presentation-041714

An evaluation of the spatio-temporal characteristics of meteorologically-adjusted ozone trends in North Texas

A preliminary analysis

by

Mahdi Ahmadi and Kuruvilla John

Department of Mechanical and Energy Engineering University of North Texas

Air Quality Technical Meeting, NCTCOG Arlington, TX. April 17, 2014

Page 2: unt-nctcog-presentation-041714

2

The Research Study objective: Evaluate the meteorologically adjusted trends in

the measured ozone concentrations in north Texas

Study area: North Texas region including DFW metroplex

Study period: 1997-2014 (12-17 years worth of data)

Air quality data: Ozone and meteorological parameters

Number of air quality monitoring sites: 16 CAMS operated by TCEQ

Method: Statistical analysis of raw ozone data (part I) and

meteorologically adjusted ozone trend using KZ- filtering method (part II)

Page 4: unt-nctcog-presentation-041714

4

Data

Time series of 8-hr average ozone concentrations were extracted from TCEQ’s Texas Air Monitoring Information System (TAMISWeb) for the 1997-2014 period (some of the monitoring stations were activated later than 1997) at sixteen state operated Continuous Ambient Monitoring Stations (CAMS).

In addition to the 8-hr ozone data, hourly and 8-hr averaged values of meteorological parameters for the same period were also acquired. Meteorological data include ambient temperature (T), solar radiation (SR), relative humidity (RH), and wind speed and direction (W).

Time series of the daily maxima of all parameters were compiled and used in

this study to evaluate the meteorologically adjusted trends in the ozone concentrations over the north Texas region.

Page 5: unt-nctcog-presentation-041714

5

Study Area Total number of gas wells:

• At the end of 2007: 5720 wells • At the end of 2013: 17,494 wells

Rate: 713 wells/year (between 2000-2006), and 1682 wells/year (2007-2013)

Source: RRC and TCEQ

Page 6: unt-nctcog-presentation-041714

6

Monitoring Sites

CAMS # Name County Parameters No. of wells - 10

mi C76 Parker County Parker O3, SR, T, W 158 C73 Granbury Hood O3, SR, T, W 428 C75 Eagle Mountain Tarrant O3, SR, T, W 2723 C77 Cleburne Airport Johnson O3, SR, T, W 1474 C13 Ft. Worth Northwest Tarrant O3, SR, T, RH, W 1092 C17 Keller Tarrant O3, SR, T, W 1364 C56 Denton Municipal Airport Denton O3, SR, T, RH, W 1362 C61 Arlington Muni. Airport Tarrant O3, SR, T, W 862 C70 Grapevine Fairway Grapevine O3, SR, T, RH, W 299 C402 Dallas Executive Airport Dallas O3, T, W 1 C60 Dallas Hinton St. Dallas O3, SR, T, RH, W 2 C63 Dallas North No. 2 Dallas O3, SR, T, W 2 C31 Frisco Collin O3, SR, T, W 0 C69 Rockwall Heat Rockwall O3, SR, T, W 0 C71 Kaufman Kaufman O3, SR, T, RH, W 0 C1006 Greenville Hunt O3, SR, T, W 0

“Fracking

region” (FR)

“Non-fracking

region” (NFR)

Page 7: unt-nctcog-presentation-041714

7

Methods

Part I: Statistical evaluation of trends in the raw ozone data • Ozone design value: “the 3-year average annual fourth-

highest daily maximum 8-hour average ozone concentration”

• Ozone exceedances: Number of days with maximum 8-hour ozone concentration exceeding NAAQS threshold (75 ppb)

Evaluations performed for the FR and NFR regions over two

contiguous periods: 2000-2006 (B07) and 2007-2013 (A07)

Page 8: unt-nctcog-presentation-041714

8

Results – Part I Trend in the ozone design values within FR

Average percent change from B07 (2000-2006) to A07 (2007-2013): -10%

65

70

75

80

85

90

95

100

105

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Ozo

ne

des

ign

val

ue

(pp

b)

Year

C73 C17 C56 C70 C13 C75 C76 C77 C61

NAAQS :75 ppb

B07 A07

Page 9: unt-nctcog-presentation-041714

9

Results – Part I Trend in the ozone design values within NFR

Average percent change from B07 (2000-2006) to A07 (2007-2013): -10%

65

70

75

80

85

90

95

100

105

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Ozo

ne

des

ign

val

ue

(pp

b)

Year

C60 C63 C69 C71 C402 C1006 C31 C77 C61

NAAQS :75 ppb

B07 A07

Page 10: unt-nctcog-presentation-041714

10

Results – Part I Annual trend in the average number of exceedance days (> 75 ppb)

Average percent change from B07 (2000-2006) to A07 (2007-2013): • Fracking region: -61.6% • Non-fracking region: -65.7%

0

5

10

15

20

25

30

352

00

0

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

20

11

20

12

20

13

Nu

mb

er o

f ex

ceed

ance

s d

ays

Year

FR NFRB07 A07

Page 11: unt-nctcog-presentation-041714

11

Results – Part I Number of average ozone exceedance days by month (> 75 ppb)

0

1

2

3

4

5

6

7

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Nu

mb

er o

f o

zon

e ex

ceed

ance

day

s

FR (B07) NFR (B07)

FR (A07) NFR (A07)

Page 12: unt-nctcog-presentation-041714

12

Results – Part I Average percent change of ozone exceedances between 2000-2006

(B07) and 2007-2013 (A07)

Ozone exceedances

Time period (75 ppb threshold)

FR NFR

May -47% -78%

Jun -74% -76%

Jul -66% -72%

Aug -49% -54%

Sep -64% -64%

Whole year -61.6% -65.7%

Page 13: unt-nctcog-presentation-041714

13

Results – Part I Trend of the average of maximum daily ozone values by month

20

25

30

35

40

45

50

55

60

65

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Ave

rage

of

max

imu

m d

aily

ozo

ne

(pp

b)

FR (B07) NFR (B07)

FR (A07) NFR (A07)

Page 14: unt-nctcog-presentation-041714

14

Results – Part I Number of days with temperatures exceeding 100 F

0

10

20

30

40

50

60

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

20

11

20

12

20

13

Nu

mb

er o

f D

ays

Year

FR NFRB07 A07

Page 15: unt-nctcog-presentation-041714

15

Results – Part I Wind rose for CAMS 13 (Ft. Worth Northwest) (May, Jun, Jul, Aug, Sep)

0

5

10

15

20

25

30

35N

NE

E

SE

S

SW

W

NW

0

5

10

15

20

25

30

35

Fre

qu

ency

(p

erce

nt)

Wind speed (mph)

>= 10

8 - 10

6 - 8

4 - 6

2 - 4

0 - 2

0

5

10

15

20

25

30

35N

NE

E

SE

S

SW

W

NW

0

5

10

15

20

25

30

35

Fre

qu

ency

(p

erce

nt)

Wind speed (mph)

>= 10

8 - 10

6 - 8

4 - 6

2 - 4

0 - 2

B07 (2000-2006) A07 (2007-2013)

Page 16: unt-nctcog-presentation-041714

16

Results – Part I Wind rose at CAMS 60 (Dallas, Hinton St.) (May, Jun, Jul, Aug, Sep)

B07 (2000-2006) A07 (2007-2013)

0

5

10

15

20

25

30

35N

NE

E

SE

S

SW

W

NW

0

5

10

15

20

25

30

35

Fre

qu

ency

(p

erce

nt)

Wind speed (mph)

>= 10

8 - 10

6 - 8

4 - 6

2 - 4

0 - 2

0

5

10

15

20

25

30

35N

NE

E

SE

S

SW

W

NW

0

5

10

15

20

25

30

35

Fre

qu

ency

(p

erce

nt)

Wind speed (mph)

>= 10

8 - 10

6 - 8

4 - 6

2 - 4

0 - 2

Page 17: unt-nctcog-presentation-041714

17

Meteorologically Adjusted Ozone

Part II: Statistical evaluation of trends of meteorologically adjusted (M.A.) ozone • Construction of M.A. ozone: application of Kolmogorov-

Zurbenko (KZ) filter

• M.A. ozone trends: statistical evaluation of total, long-term, and baseline M.A. ozone

Evaluations performed for the FR and NFR regions over two contiguous periods: 2000-2006 (B07) and 2007-2013 (A07)

Page 18: unt-nctcog-presentation-041714

18

Method – KZ-filter KZ-filter is a robust low-pass filter used for spectral (frequency-based)

decomposition of time-series*

Computationally KZ-filter is a simple iterative moving average technique

Log-transformed raw ozone time series

Spectral decomposition (KZ-

filter)

* Rao ST, Zurbenko IG. Detecting and tracking changes in ozone air quality. Air & waste 1994; 44: 1089-1092. Rao ST, Zalewsky E, Zurbenko IG. Determining temporal and spatial variations in ozone air quality. Journal of the Air & Waste Management Association 1995; 45: 57-61.

𝑌𝑖 =1

𝑚 𝑋𝑖+𝑗

k

𝑗=−𝑘

Long-term

Seasonal

Short-term

Page 19: unt-nctcog-presentation-041714

19

Meteorologically Adjusted Ozone • Step 1: Spectral decomposition of the ozone and meteorological time

series by applying KZ-filter

• Step 2: Baseline and short-term components of ozone time-series are calculated using a three variable linear regression model (with temperature and solar radiation)

• Step 3: Meteorologically adjusted ozone time-series is calculated as the residuals of linear regressions

• Step 4: KZ-filter is applied again to M.A. ozone time-series to calculate long-term and baseline components of M.A. ozone

𝑂 𝑡 = 𝐹 𝑇, 𝑆𝑅 + 𝑂MA(𝑡)

𝑂MA(𝑡) = 𝜖BL(𝑡) + 𝜖𝑊(𝑡)

𝑋(𝑡) = 𝑒(𝑡) + 𝑆(𝑡) +𝑊(𝑡)

&𝑒 𝑡 = 𝑋365,3(𝑡

&𝑆 𝑡 = 𝑋15,5(𝑡) − 𝑋365,3(𝑡

&𝑊 𝑡 = 𝑋 𝑡 − 𝑋15,5(𝑡

&𝐵𝐿 𝑡 = 𝑒 𝑡 + 𝑆 𝑡

Page 20: unt-nctcog-presentation-041714

20

Results – Part II

Because all expressions are given for the natural logarithm of raw ozone data, M.A. ozone values have multiplicative effects in original ozone time series

Also since OM.A. values are sufficiently small therefore:

Therefore 𝑂MA 𝑡 × 100 values represent change in mean value of the

original ozone time series

Statistical evaluation of all components of M.A. ozone (mean value, long-term, short-term, and baseline) is performed to evaluate the contribution of non-meteorological variables including anthropogenic influences affecting ozone formation and destruction.

exp 𝑂𝑀.𝐴.& ≈ 1 + 𝑂𝑀.𝐴.&

Page 21: unt-nctcog-presentation-041714

21

Results – Part II Examples of calculated and measure ozone time series and residuals (M.A. ozone)

1

1.5

2

2.5

3

3.5

4

4.5

5

2001 2003 2004 2006 2007 2008 2010 2011

Log

of

ozo

ne

(pp

b)

Measured Calculated

-2

-1.5

-1

-0.5

0

0.5

1

2001 2003 2004 2006 2007 2008 2010 2011

M.A

. ozo

ne

Year

M.A. ozone Baseline of M.A. ozone

Constructing M.A. ozone for C56 (Denton)

1

1.5

2

2.5

3

3.5

4

4.5

5

2003 2004 2006 2007 2008 2009 2010 2011

Lo

g o

f o

zon

e (p

pb

)

Measured Calculated

-2

-1.5

-1

-0.5

0

0.5

1

2003 2004 2006 2007 2008 2009 2010 2011

Lo

g o

f M

.A. o

zon

e (p

pb

)

Year

M.A. ozone Baseline of M.A. ozone

Constructing M.A. ozone for C31 (Frisco)

Page 22: unt-nctcog-presentation-041714

-3

-2.5

-2

-1.5

-1

-0.5

0

0.5

1

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

M.A

. ozo

ne

con

cen

trat

ion

Year

22

Results – Part II Box plot presentation of 𝑂MA(𝑡) (the maximum, 75th percentile, median,

mean, 25th percentile, and minimum) for FR and NFR

𝑂MA 𝑡 × 100 values represent change in mean value of the original ozone time series

Page 23: unt-nctcog-presentation-041714

23

Results – Part II Mean values of 𝑂MA(𝑡)

𝑂MA 𝑡 × 100 values represent change in mean value of the original ozone time series

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Ave

rage

M.A

. ozo

ne

Year

FR NFR

Page 24: unt-nctcog-presentation-041714

24

Results – Part II Mean value of of 𝑂MA 𝑡 during ozone season (May, Jun, Jul, Aug, Sep)

𝑂MA 𝑡 × 100 values represent change in mean value of the original ozone time series

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Ave

rage

of

M.A

. ozo

ne

(war

m m

on

ths)

Year

FR NFR

Page 25: unt-nctcog-presentation-041714

25

Results – Part II Mean values of 𝑂MA(𝑡) in FR and NFR during winter time (Nov, Dec, Jan, Feb)

𝑂MA 𝑡 × 100 values represent change in mean value of the original ozone time series

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Ave

rage

of

M.A

. ozo

ne

(co

ld m

on

ths)

Year

FR NFR

Page 26: unt-nctcog-presentation-041714

26

Results – Part II Mean values of long-term component of 𝑂MA(𝑡) (that is devoid of

meteorological influence)

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

2003 2004 2005 2006 2007 2008 2009 2010

Lo

ng-

term

co

mp

on

ent

of

M.A

. Ozo

ne

Year

Fracking Counties Non-fracking counties

𝑂MA 𝑡 × 100 values represent change in mean value of the original ozone time series

Page 27: unt-nctcog-presentation-041714

27

Remarks Daily vehicle miles traveled (VMT) is relatively flat during 2007-2010.

Source: TxDOT (Total VMT for Dallas, Denton, Wise, Tarrant, Hood, Parker, Hunt, Ellis, Collin, Rockwall & Kaufman counties)

Page 28: unt-nctcog-presentation-041714

28

Remarks Based on raw ozone data (B07 vs. A07):

Sites located in NFR show about 4% more reduction in the number of average ozone exceedances than in FR

Average of winter time daily maximum ozone values in FR is 3% more than in NFR

Based on the mean value of meteorologically-adjusted ozone 𝑂MA 𝑡 (post-2008): 12% increase within FR and a 4% increase in NFR was

noted During the winter time, 21% increase within FR and a

5% increase in NFR was noted

Page 29: unt-nctcog-presentation-041714

29

Future Research Needs Conduct similar adjusted trend analyses using long-term

measurements of NOx, VOC and PM2.5 in the region.

Additional rural monitoring of NOx, VOC and PM2.5 may be needed.

Perform source-receptor and source apportionment analyses using measured concentrations.

Develop a comprehensive regional emissions inventory for a more recent base year.

Conduct photochemical modeling for a recent episode as well as a year-round modeling using revised estimates of emissions and perform source apportionment analysis for various regions.

Page 30: unt-nctcog-presentation-041714

Thank you! Any questions?

Mahdi Ahmadi [email protected] Dr. Kuruvilla John [email protected]