Georgia Chapter of the Air & Waste Management Association Annual Conference: Improved Air Quality...
-
Upload
norma-faley -
Category
Documents
-
view
215 -
download
0
Transcript of Georgia Chapter of the Air & Waste Management Association Annual Conference: Improved Air Quality...
Georgia Chapter of the Air & Waste Management Association Annual Conference:
Improved Air Quality Modeling for Predicting the Impacts of Controlled Forest Fires
Fernando Garcia MenendezGeorgia Institute of Technology
10/8/2009
Evaluation of Air Quality Models Applied to Wildland Fire Impact Simulation
Fernando Garcia-MenendezAika Yano
Yongtao HuM. Talat Odman
Georgia Institute of Technology10/19/2010
9th Annual CMAS Conference:
Evaluation of Smoke Models and Sensitivity Analysis for Determining Emissions Related Uncertainties
Objective 1: Evaluate performance of a range of AQ models with data from prescribed burn and wildfire events in the Southeast.
Objective 2: Quantify the uncertainties in model outputs generated from uncertainties in emission inputs.
Wildland Fires & Air Quality
PM2.5 CO VOC0%
5%
10%
15%
20%
25%
Forest Fire Contribution to Air Pollutant Emissions
in Southeastern U.S. *
•Wildland fires may have detrimental impacts on air quality, health & visibility.•Air Quality Modeling can be used as a basis for decision-making.• Informed decisions require knowledge of model’s capabilities & uncertainties.
*Bernard & Sabo, 2003
Episode: February 28, 2007
Measurement Sites
Recorded Observations
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
CFA
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
FTM
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
FS8
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
MCD
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
GWI
0
20
40
60
80
100
120
140
160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
NEW
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
JST
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
YRK
020406080
100120140160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
SDK
0
20
40
60
80
100
120
140
160
17 18 19 20 21 23 0 1 2 3
PM2.
5 (µ
g/m
3)
Time
WAL
Air Quality Modeling
Air Quality Modeling
Pave Pic
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Jefferson St.
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fire Station 8
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fort McPherson
020406080
100120140160180
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
McDonough
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Confederate Av.
Model Performance
Pave Pic
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Jefferson St.
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fire Station 8
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fort McPherson
020406080
100120140160180
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
McDonough
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Confederate Av.
Performance Comparison
Pave Pic
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Jefferson St.
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fire Station 8
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fort McPherson
020406080
100120140160180
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
McDonough
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Confederate Av.
Statistical Evaluation
SITE:JST
CFAMCDSDKFTMFS8
AVERAGE
72.5%
77.5%
CMAQ78.0%
66.5%92.5%85.7%69.7%
66.4%111.8%68.5%60.6%63.8%
72.7%
82.8%131.3%94.9%48.6%97.0%
94.8%
28.947.239.032.223.2
32.1
Mean ErrorMean Normalized
ErrorNormalized Mean
ErrorMean Fractional
ErrorCMAQ21.9
CMAQ114.1%
CMAQ65.4%
Statistical Comparison
SITE: CMAQ AG-CMAQ CMAQ AG-CMAQ CMAQ AG-CMAQ CMAQ AG-CMAQJST 21.9 21.7 114.1% 71.4% 65.4% 65.0% 78.0% 58.3%
CFA 28.9 29.4 82.8% 57.0% 66.4% 67.5% 66.5% 58.3%MCD 47.2 27.6 131.3% 58.6% 111.8% 65.3% 92.5% 64.3%SDK 39.0 40.5 94.9% 70.1% 68.5% 71.0% 85.7% 78.6%FTM 32.2 33.3 48.6% 52.3% 60.6% 62.7% 69.7% 74.4%FS8 23.2 23.8 97.0% 83.4% 63.8% 65.4% 72.5% 65.0%
AVERAGE 32.1 29.4 94.8% 65.5% 72.7% 66.2% 77.5% 66.5%
Mean Fractional ErrorMean Normalized
ErrorMean Error
Normalized Mean Error
Large amount of information is generated but only a small fraction is analyzed!Opportunities for model insight are lost:
- Model strengths and weaknesses- Sensitivity to inputs and uncertainty in results- Better guidance for future model development
Model: DAYSMOKE-CMAQ
DAYSMOKE: - Developed by US Forest
Service - Stochastic plume model
- Inert model, PM2.5 only
- Typical scales < 10 km
DAYSMOKE-CMAQ: Uses DAYSMOKE as an emissions
injector into CMAQ
Model: AG-CMAQ
• Dynamic, solution-adaptive grid algorithm.• Variable t-step algorithm.• Conserves original CMAQ grid structure.
*Garcia-Menendez, et al. (2010): An Adaptive Grid Version of CMAQ for Improving the Resolution of Plumes, Atmospheric Pollution Research, 1, 239-249
CMAQ AG-CMAQ
Model Evaluation: AG-CMAQ & DAYSMOKE-CMAQ
Episode: Feb. 2007 Fires near Atlanta, GAFire Emissions: Fire emissions Production Simulator (FEPS)Background Emissions: SMOKE with projected 2002 “typical year” inventoryMeteorology: Weather Research & Forecasting model (WRF) Plume Rise & Vertical Profile: DAYSMOKE
DAYSMOKE-CMAQ AG- CMAQ
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Jefferson St.
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fire Station 8
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fort McPherson
020406080
100120140160180
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
McDonough
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Confederate Av.
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Jefferson St.
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fire Station 8
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Fort McPherson
020406080
100120140160180
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
McDonough
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
Confederate Av.
Feb. 28, 2007 22:30 Z
Mar. 1, 200702:00 Z
DAYSMOKE-CMAQ
DAYSMOKE-CMAQ
AG-CMAQ
AG-CMAQ
Mar. 1, 2007 00:30 Z
Mar. 1, 200702:15 Z
DAYSMOKE-CMAQ
DAYSMOKE-CMAQ
AG-CMAQ
AG-CMAQ
CMAQ AG-CMAQ CMAQ AG-CMAQ CMAQ AG-CMAQ CMAQ AG-CMAQJST 21.9 21.7 114.1% 71.4% 65.4% 65.0% 78.0% 58.3%
CFA 28.9 29.4 82.8% 57.0% 66.4% 67.5% 66.5% 58.3%MCD 47.2 27.6 131.3% 58.6% 111.8% 65.3% 92.5% 64.3%SDK 39.0 40.5 94.9% 70.1% 68.5% 71.0% 85.7% 78.6%FTM 32.2 33.3 48.6% 52.3% 60.6% 62.7% 69.7% 74.4%FS8 23.2 23.8 97.0% 83.4% 63.8% 65.4% 72.5% 65.0%
AVERAGE 32.1 29.4 94.8% 65.5% 72.7% 66.2% 77.5% 66.5%
Mean Normalized Error
Normalized Mean Error
Mean Fractional ErrorMean Error
CMAQ AG-CMAQ CMAQ AG-CMAQ CMAQ AG-CMAQ CMAQ AG-CMAQJST 21.9 21.7 114.1% 71.4% 65.4% 65.0% 78.0% 58.3%
CFA 28.9 29.4 82.8% 57.0% 66.4% 67.5% 66.5% 58.3%MCD 47.2 27.6 131.3% 58.6% 111.8% 65.3% 92.5% 64.3%SDK 39.0 40.5 94.9% 70.1% 68.5% 71.0% 85.7% 78.6%FTM 32.2 33.3 48.6% 52.3% 60.6% 62.7% 69.7% 74.4%FS8 23.2 23.8 97.0% 83.4% 63.8% 65.4% 72.5% 65.0%
AVERAGE 32.1 29.4 94.8% 65.5% 72.7% 66.2% 77.5% 66.5%
Mean Normalized Error
Normalized Mean Error
Mean Fractional ErrorMean Error
• 11 % average improvement in Mean Fractional Error with AG-CMAQ.• Better performance in 5 of 6 stations.
Much more insight into model performance gained beyond that achieved from statistical comparison only.
CMAQ AG-CMAQ
CMAQ AG-CMAQ
Feb. 28, 2007 19:30 Z
Feb. 28, 200722:45 Z
DAYSMOKE-CMAQ
DAYSMOKE-CMAQ
AG-CMAQ
AG-CMAQ
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
23:15 Z 00:00 Z 01:15 Z
23:15 Z 00:00 Z 01:15 Z
020406080
100120140160
17 18 19 20 21 22 23 0 1 2 3 4
PM 2
.5 (μ
g/m
3 )
Time (Z)
South DeKalb
5⁰ DegreeShift in Wind Field
Model: CALPUFF• Gaussian dispersion model.• Simulates pollution as a series of puffs interacting with terrain
& meteorological fields.• Typical scales < 100 km
Future Work: Model Evaluations• Explore other fire emissions and plume rise options available for CMAQ & SMOKE.• Further explore Advanced Plume Treatment (APT) in CMAQ. • Future episodes to evaluate:
– Prescribed burn episodes at Fort Benning, GA– Florida-Georgia wildfires May-June 2007
• Sensitivity analysis with Direct Decoupled Method of fire related parameters.
• Adaptive-Grid-Daysmoke CMAQ: – Daysmoke-CMAQ hybrid model– Currently inert
• Reactive Daysmoke-CMAQ:
Future Work: Model Development
Chemical treatment of smoke plume will require 3 considerations:1.Plume Volume2.Level of Chemical Interaction3.Reacted Mass Redistribution
Thank you!
Acknowledgements: