Modelling of health-relevant dispersion of air pollution in the atmospheric Regional-scale

1
Modelling of health-relevant dispersion of air
pollution in the atmosphericRegional-scale

• Wolfgang Schöpp

2
Modelling concept for RAINS

• Use computationally efficient functional
  relationships to describe
• the response of health-relevant metrics of air
  pollution (for each 5050 km2 grid cell)
• to changes in precursor emissions in different
  countries
• Endpoints
• For PM Annual mean concentration of PM2.5
  rural concentrations and in urban background air
• For ozone SOMO35 (Sum of daily eight-hour mean
  ozone in excess of 35 ppb) rural concentrations
  and in urban background air

3
For PM2.5

• Endpoint annual mean concentrations of PM2.5
  composed of
• Primary emissions of PM2.5 from anthropogenic
  sources
• Secondary inorganic aerosols (ammonium sulfate,
  ammonium nitrate) due to precursors SO2, NOx, NH3
• Water associated with secondary inorganics
• Secondary organic aerosols (from VOC emissions)
• Natural background (mineral, sea salt, organic
  matter)
• A fraction that is chemically not identified by
  the measurements
• Thus calculations do not reproduce complete
  observed mass

• Endpoint annual mean concentrations of PM2.5
  composed of
• Primary emissions of PM2.5 from anthropogenic
  sources
• Secondary inorganic aerosols (ammonium sulfate,
  ammonium nitrate) due to precursors SO2, NOx, NH3
• Water associated with secondary inorganics
• Secondary organic aerosols (from VOC emissions)
• Natural background (mineral, sea salt, organic
  matter)
• A fraction that is chemically not identified by
  the measurements
• Thus calculations do not reproduce complete
  observed mass Focus on anthropogenic fraction!

4
Model experimentswith the EMEP Eulerian model

• Sample of EMEP model calculation covering the
  policy-relevant range of emission reductions
  (baseline/current legislation 2010 ? maximum
  technically feasible reductions)
• Explore responses of PM2.5 and ozone
  concentrations to changes in SO2, NOx, VOC, NH3,
  PPM2.5/10 emissions
• 3 emission scenarios
• CLE (current legislation 2010) CAFE baseline
  for 2010
• MFR (maximum technically feasible reductions 2010
• UFR (ultimately feasible reductions) MFR/2
• Derive computationally efficient functional
  relationships that describe response of full EMEP
  model in this range of emissions

5
Response of PM2.5 to ?PPM2.5in Germany
Change in PM2.5 concentrations µg/m3 at the
German grid cells (red) and in other countries
(blue)due to PPM2.5 emissions changing
fromCurrent LEgislation (CLE)to Ultimate
Reductions (UFR MFR/2)
Change in PM2.5 concentrations µg/m3 due to
PPM2.5 emissions changing fromCurrent
LEgislation (CLE) to Max Feasible Reductions
(MFR UFR2)
6
Response of PM2.5 to ?PPM2.5in Germany

• Linear response of PM2.5
• Over full range of primary PM2.5 emission changes
• No interactions with other pollutants

Change with all other emissions at UFR
Change with all other emissions at CLE
7
Response of secondary inorganic aerosols to ?SO2
from German emissions
Change with all other emissions at UFR
Change with all other emissions at CLE

• Rather linear response of PM2.5
• over full range of SO2 emissions
• Minor impact of other emissions

8
Response of secondary inorganic aerosols to ?NOx
from German emissions

• Strong non-linear response
• to changes in NOx emissions
• Strong influence of other emissions (NH3!)

9
Response of secondary inorganic aerosols to ?NH3
from German emissions

• Strong non-linear response
• to changes in NH3 emissions
• Strong influence of other emissions (NOx!)

10
Functional relationships for PMdeveloped for
RAINS

• PM2.5j Annual mean concentration of PM2.5 at
  receptor point j
• I Set of emission sources (countries)
• J Set of receptors (grid cells)
• pi Primary emissions of PM2.5 in country i
• si SO2 emissions in country i
• ni NOx emissions in country i
• ai NH3 emissions in country i
• aS,Wij, ?S,W,Aij, sW,Aij, pAij Linear
  transfer matrices for reduced and oxidized
  nitrogen, sulfur and primary PM2.5,
  for winter, summer and annual

11
Validation of PMCAFE baseline 2020 µg/m3
12
Response of ozone to ?NOxfrom German emissions
(summer mean of daily max O3)

• Strong non-linear response
• to changes in NOx emissions
• Some influence of other emissions (VOC)

13
Response of ozone to ?VOCfrom German emissions
(summer mean of daily max O3)

• Linear response of ozone
• to changes in VOC emissions
• Influence of other emissions (NOx!)

14
The IIASA reduced-form ozone model used for the
NEC analysis (Heyes et al., 1996)

• where enlj , the mean "effective" emissions, are
  given by
• and the following indices and abbreviations are
  used
• i - emitter country index number
• j - receptor grid index number
• l - type of AOT measure index number
• I - set of emitter countries included for grid j
  and AOT measure l
• vi - annual national emissions of non-methane
  VOCs from emitter country i
• ni - annual national emissions of NOx from
  emitter country i
• ennj - mean "effective" NOx emissions from other
  emitter countries where i ? I

15
Reduced-form ozone model Comparison with the
EMEP Lagrangian model (AOT40)
16
Conclusions

• For PM2.5
• Non-linearities for nitrogen fraction
• Decomposition into season-specific dispersion
  modes, NOx and NH3 limited regimes
• Piece-wise linear representation
• For ozone
• In areas with high NOx concentrations, non-linear
  response to NOx reductions. Functional
  relationships have been developed for RAINS.
• Non-linearity depends on concentrations and
  metric
• For SOMO35 and for emissions beyond 2010, main
  non-linearities in spring (titration of ozone
  from free troposphere?)
• Provisional approach ignoring these
  non-linearities