U.S. EPA Office of Research & Development October 16, 2012 Prakash Bhave, Adam Reff, Alexis Zubrow, Venkatesh Rao U.S. Environmental Protection Agency CMAS Conference Chapel Hill, NC October 15 – 17, 2012 Evaluation of Urban PM 2.5 Emission Inventories across the U.S.
13
Embed
Evaluation of Urban PM 2.5 Emission Inventories across the U.S.
Evaluation of Urban PM 2.5 Emission Inventories across the U.S. Prakash Bhave, Adam Reff, Alexis Zubrow, Venkatesh Rao U.S. Environmental Protection Agency CMAS Conference Chapel Hill, NC October 15 – 17, 2012. PM 2.5 Components ( μ g m -3 ). SO 4. CMAQ v4.7. NO 3. OC. - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
U.S. EPA Office of Research & Development October 16, 2012
Prakash Bhave, Adam Reff, Alexis Zubrow, Venkatesh Rao
U.S. Environmental Protection Agency
CMAS ConferenceChapel Hill, NC
October 15 – 17, 2012
Evaluation of Urban PM2.5 Emission Inventories across the U.S.
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
2
Conclusions: CMAS 2010
• In the past decade, which modeling system refinements contributed most to PM2.5 performance improvement?→Meteorology inputs (2)→Emissions & deposition (4)→Atmospheric chemistry (2)
IMPROVE Observations (1996)
PM2.5 Components (μg m-3)
CM
AQ
v4.
1
NO3SO4
OC
IMPROVE Observations (1996)
PM2.5 Components (μg m-3)
CM
AQ
v4.
1
IMPROVE Observations (1996)
PM2.5 Components (μg m-3)
CM
AQ
v4.
1
NO3SO4
OC
PM2.5 Components (μg m-3)
IMPROVE Observations (2002 – 2006)
CM
AQ
v4.
7
NO3
SO4
OC
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
3
Background & Motivation• U.S. has most detailed national inventory for PM2.5
– Spatial resolution
– Source resolution
– Chemical resolution
• Inventory accuracy
very difficult to check– CTM is often used
– Can we find & fix gross
inventory errors without
running CMAQ? Reference: Reff et al. (ES&T, 2009)
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
4
• Cass & McRae (ES&T, 1983) demonstrated a simple approach for PM2.5 inventory evaluation• Compare emission rates
directly against ambient concentrations• Only works because,
*most trace elements are conserved*
• Results• Ti, Ni emissions too high• Zn too low• Ambient Cu data error
•We applied same method to 2001 NEI in 21 cities…
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Secondary Species
Below MDL
Reff et al. (Intl Aerosol Conf. 2006)
Al Ca Fe KSi
Prior Evaluation: 2001 NEI
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
6
Prior Evaluation: 2001 NEIEmissions Allotment of Si
Em
iss
ion
s (t
on
/yea
r)
Dallas Minneapolis St. Louis
Emissions Allotment of Si
Em
iss
ion
s (t
on
/yea
r)
Dallas Minneapolis St. Louis
Factor Dilutionc Atmospheri
ionConcentrat Ambient
• In many cities, we found positive biases in the emissions of– Agricultural soil– Unpaved road dust
Methodological Shortcomings• Limited number of sites (n = 21)• 36 km grid resolution• “old” version of NEI• Only able to identify gross
overestimates• Unable to quantify the emission
errors
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Methodology• 2005ak NEI• Mobile emissions from 2005cr, output by MOVES• Spatial allocation: 12km ConUS grid• Temporal allocation: monthly
• 85 source categories with unique PM2.5 speciation profiles
• Aggregate to 159 Core-Based Statistical Areas (CBSA)
Result> 7×104 pairs of diluted emissions & ambient concentrations
• Multiply emissions by month-
& site-specific dilution ratio
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Methodology• Apply principles of chemical mass balance (CMB)
correction factor
• Data in each city/month are fit separately
• Key result: source-specific F value for each site & month
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
MethodologyForce Fij to be positive
Account for measurement
error
Minimize this
Penalize fit for over-correcting the
emissions
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Preliminary ResultsF values for Agricultural Burning
100
1
0.01J F M A M J J A S O N D
• PM2.5 from crop burning is biased high by ~10x
• Pouliot, McCarty, et al. have diagnosed the reason for these overestimates
• Revisions will be incorporated into 2008 NEI
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Preliminary ResultsF values for Unpaved Road Dust
100
1
0.01J F M A M J J A S O N D
• PM2.5 from unpaved roads is biased high by ~30x
• Is this entirely due to emissions error?• see poster by Appel et al.
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Preliminary ResultsF values for Unpaved Road Dust
100
1
0.01J F M A M J J A S O N D
Median of Monthly F values
U.S. EPA Office of Research & Development, Atmospheric Modeling & Analysis Division
Summary•Methodology to quantify source-specific biases in
PM2.5 inventory has been developed
•Preliminary results look quite promising!
• In process of assessing our results for other source