1. Global Influences on Local Pollution1
Global Influences on Local Pollution
Co-Chairs
Frank Dentener, PhD Terry Keating, PhD EC JRC U.S. EPA
13 September 2017

2. Update of global ozone health impacts- Malley et al., EHP, 2017
Difference of new and old ozone risk factors on global scale attributable deaths
Update Jerrett (2009) dose-response estimates with Turner et al 2016 revision.
T2016: larger cohort; longer follow-up; different metrics=> larger risk factors.
OLD J2009: million => T million premature deaths due to
10 %(J2009) => 25 % (T2016) total outdoor air pollution health impacts is due to ozone
=>For HTAP this means the importance of extra-regional contributions to health impacts from O3 and PM together may be increasing from ca % to % (preliminary estimate).

3. Working to Improve Capabilities
Global Model Inputs
Emissions, Meteorology, Land Use
Global to Regional Linkages
Nesting, One Way, Two Way
Global-Regional Model System Evaluation
What do we need the global model to do?
Source Apportionment Methods and scenarios
Perturbation, Tagging, DDM, Adjoint
DDM direct differentation sensitivity analys

4. Daily Observations and Model Simulations
Global to Regional Linkages
Comparison of Observed and Simulated MDA8 O3, AQS Sites
AQMEI3: MDA8; CMAQ driven by 4 different global models Boundary conditions.
Daily Bias
BC C-IFS
BC H-CMAQ
BC AM3
BC GEOSCHEM
Daily Observations and Model Simulations
Observations
BC C-IFS
BC H-CMAQ
BC AM3
BC GEOSCHEM
Monthly Mean Bias
BC C-IFS
BC H-CMAQ
BC AM3
BC GEOSCHEM
A lot of interest from Western States on the role of hemispheric transport in causing exceedences.
1200 sites
The choice of boundary conditions can have a significant impact on model bias for MDA8 O3, especially for the non-summer months

5. Global to Regional Linkages
Impact of Different Boundary Conditions on CMAQ Simulated Spring Average Surface O3 Mixing Ratios
O3 BC C-IFS (“Base”)
O3 BC H-CMAQ - Base
Consistent with the analysis of the global models and CMAQ ozone column burdens, the CMAQ BC AM3 simulations exhibit the highest spring surface ozone mixing ratios over the entire modeling domain
Differences between the BC H- CMAQ and BC GEOS-CHEM runs w.r.t. “base” are smaller
Separate analysis shows that BC impacts are very similar for hourly and MDA8 O3
The differences point to differences in the representation of stratospheric ozone and stratosphere/troposphere exchange
O3 BC AM3 - Base
O3 BC GEOS-CHEM - Base

6. Global to Regional Linkages
Impact of Different Boundary Conditions on CMAQ Simulated Summer Average Surface O3 Mixing Ratios
O3 BC C-IFS (“Base”)
O3 BC H-CMAQ - Base
Throughout most of the domain, all three sensitivity simulations tend to have higher summertime ozone than the C- IFS driven base case
For a large portion of the domain, the BC AM3 run still has the highest concentrations, but the differences are smaller than during spring
O3 BC AM3 - Base
O3 BC GEOS-CHEM - Base
Probably skip the summer plot

7. Global-Regional Model System Evaluation
Global (HTAP2) and Regional (AQMEII3) Models for Europe Compared to monthly and hourly observations in 2010.
1300 stations of which 450 rural
Similar range of variability in Europe between global and regional models
Ensemble of regional models overall less biased

8. HTAP2-AQMEII3 Models for Europe Compared to Observations
Global-Regional Model System Evaluation
HTAP2-AQMEII3 Models for Europe Compared to Observations
Mediterranean
C Europe <1km
C Europe >1km
HTAP1 Results
Airbase
Ca 2000 sites, more than 400 rural sites.
HTAP2 global models seem to be more biased than during HTAP1
More comprehensive set of observations

9. Performance Varies with Ozone Metric (Averaging Time)
Global-Regional Model System Evaluation
Performance Varies with Ozone Metric (Averaging Time)
Monthly Mean
M7 (daylight avg)
AOT40 (x105)
Measurements
Regional models
Global models
Regional
Global
North
South
West
Central
All
All Rural 40
Observations 20
Mean Bias
-20
North
South
West
Central
All
All Rural
-40
Regional models 60 50
AOT40m and m7 for o3
Sampling station are selected if they guarantee 90% of valid data to avoid to much discrepancy with models when integrating since models have no data gaps
AOT40= Accumulated exposure over a threshold of 40 ppb - running 3-month
M7 = Daily mean ozone concentration within the 7-hour daylight period (9am-4pm) maximum
Results are grouped in areas North, South, Central, all stations (1390), all stations only rural (453)
AOT40: Regional scales models appear in good agreement with measured values, across the continent and in the sub-regional break down. 40
North
South
West
Central
All
All Rural
Root Mean Square Error 30 20
Global models 10

10. Source Apportionment Methods and scenarios
CLRTAP 2016 Assessment Report; HTAP1 SR relationships; HTAP2 global pollution scenarios ;O3 changes [ppb] in Europe
Total surface O3
change
REF CLE Current Legislation- no climate policy
CLIM-CLE Current legislation- climate policy (energy sector)
REF MFR Maximum Feasible Reductions
In Europe:
Regional controls can still bring down ozone further, but requires ambitious and expensive air pollution policy
Ambitious air pollution policy elsewhere may be beneficial for Europe, especially in the US. For the US it would be China, etc..
Methane emission reductions crucial for reducing ozone.
Reducing methane will require strong collaboration with countries in Asia
A likely range for changes in ozone boundary conditions is therefore estimate to range between -4 and 3 ppb in 2030, and -4 and 5 ppb in 2050.
Within EU
Outside of EU
A likely range for changes in ozone boundary conditions is therefore estimate to range between -4 and 3 ppb in 2030, and -4 and 5 ppb in 2050.
Methane

11. Source Apportionment Methods and scenarios
HTAP2 scenarios HTAP2 SR relationships; Regional O3 changes [ppb]
Current Legislation
1.5 5 2
Europe
Middle East
North America 
CH4  
Total 
Extra-regional
regional  2 8 1
East Asia
South Asia
Russia
5 more macro regions
Similar annual ozone changes in Europe, North America, East Asia- role of CH4
Much larger changes in e.g. South Asia and Middle East
Preliminary results Steven Turnock, UK Metoffice

12. Maximum Feasible Reduction
Source Apportionment Methods and scenarios
HTAP2 scenarios Regional O3 changes [ppb] HTAP2 SR relationships
Maximum Feasible Reduction
CH4 
Regional 
Extra-regional 
Europe
North America
Middle East
Total  -6 -9 -5 
North America was -7 under MFR
South Asia
Russia
East Asia -9 -9 -4
Methane+extraregional > regional in most regions
Reductions between -4 and -9 ppb; but much larger compared to CLE 2050
Preliminary results Steven Turnock, UK Metoffice

13. HTAP improvement of modelling Capabilities
Global to Regional Linkages
In the US regional model performance was shown to be partly dependent on choice of boundary conditions from global model. The CAMS Boundary conditions were showing a relatively good performance. The boundary conditions were used also in Europe and for MICS-Asia.
Global-Regional Model System Evaluation
Progress on systematic comparison of global and regional models on similar performance statistics.
In Europe global model results show a similar spread on seasonal and daily timescales for ozone as regional models- regional models are less biased.
Errors in global models are more sensitive to threshold based metric like AOT40, compared to M7
Source Apportionment Methods
Harmonized experiments among global models HTAP2 and regional models from AQMEII3. New sets of models and emissions. Sparse matrix of SRs means more uncertainty for ensemble results from global models. HTAP2 global models now provide an updated evaluation of a set of air pollution mitigation scenarios- similar results as in CLRTAP assessment, but more detailed information on more regions. Results from other methods still have to be compared
Results are preliminary; More systemic analysis of errors and biases should rationalize the weighed use of regional and global model results- depending on the metric considered.