Modelling Atmospheric Aerosol over the U.K.

1
Modelling Atmospheric Aerosol over the U.K.
Comparison with Aircraft Measurements

• Michael Bane and Gordon McFiggans
• Centre for Atmospheric Science
• University of Manchester

2
Requirements for Aerosol Transport Model

• Necessity to interpret aerosol process research
  in light of aircraft (etc) measurements what we
  require from a model
• Framework for testing process descriptions
• eg equilibrium properties
• Prediction of Aerosol Field Measurements
• Size distributions
• Component loading by size
• Operational Model
• Future fieldwork planning real time deployment

3
Brief Overview of CMAQ Models-3

• Models-3 CCTM
• Originally US EPA code for AQ legislative
  compliance
• predicts species concentration, deposition
  fields visibility reduction through Mie
  scattering calculations
• Modular
• Choice of chemical scheme
• Radm2 (gas phase) 4-product isoprene aerosol
  aqueous
• Inputs Required
• MET set up run MM5
• Currently evaluating UWERNs post-UM processor
• Emissions
• Gridded values for all domain over all timesteps
• Initial boundary conditions
• Photolysis rates

4
Aerosol Treatment in Models-3

• Modal (standard CMAQ)
• version 4.6 released very recently (Oct 2006)
• Sectional (MADRID)
• model of aerosol dynamics, reaction, ionisation
  and dissolution
• development release 2004 - built upon CMAQ v4.4
• various options for mass transfer equilibrium
  treatments

5
Modal or Sectional?

• sectional
• representation of
• aerosol dynamics is
• more flexible

6
Set up Emissions, IC, BC

• Standard CMAQ (modal)
• Emissions 1999 EMEP (50km res, Europe), NAEI
  (1km res, UK)
• gridded according to chemical mechanism (RADM2)
• IC, BC profiles for outer 108km domain

7
Set up Emissions, IC, BC

• Standard CMAQ (modal)
• Emissions 1999 EMEP (50km res, Europe), NAEI
  (1km res, UK)
• gridded according to chemical mechanism (RADM2)
• IC, BC profiles for outer 108km domain
• MADRID (sectional)
• Sectional emissions as per MADRID pre-processors
  (SCAQS, Aug 1987)
• Size composition disaggregated from PM2.5
  PM10
• IC, BC profiles for 108km domain import CMAQ
  values into MADRID
• Reapportion (eg) sulphate Aitken Accumulation
  mode masses into sectional representation (using
  CMAQs logNormal parameters)
• Ongoing ATMOS work
• Better size- species- resolution of PM
  emissions and IC / BC for the UK from UK
  measurements (e.g. from AMPEP flights NCAS/DIAC
  work)

8
Set-up Domains, chemistry schemes, met

• Domains
• 108km ? 36km ?12km
• c10 days spin up
• Chemical scheme, solver,met driver
• Radm2 with isoprene (4 product) chem, aerosol
  aqueous
• CMAQ radm2_ci4_ae3_aq
• MADRID radm2_ci4_aqRADM_aeMADRID1_8sec
• Rosenbrock solver (ros3)
• Met generated using MM5 and ECMWF gridded 2.5o x
  2.5o
• 2007 move to using UM output

9
Assessing most useful representation

• Remove non-aerosol discrepancies between versions
  (example)
• N2O5 CMAQ includes N2O5 heterogeneous hydrolysis
  within aerosol routines
• MADRID does not represent uptake dependence on
  aerosol nitrate (CMAQ uses gN2O5 as function of
  nitrate loading)

10
Comparison of Photochemistry
Scatter plot of MADRID .v. CMAQ Ozone
concentrations at 1200GMT 25 May 2005 (684th
timestep of 108 km domain)
11
Comparison with Aircraft Measurements

• Exploring how Models-3 (standard CMAQ MADRID)
  predicts measurements
• BAe 146 Flights 2005 / 6,
• focus on B097, AMPEP
• Anticlockwise, May 2005, (S) Westerlies
• Aerodyne AMS

1150
0748
1025
0945
12
Model/measurement comparison 0750-0830
13
Model/measurement comparison 0830-0930
14
Model/measurement comparison 0930-1030
15
Model/measurement comparison 1030-1130
16
Model/measurement comparison 1130-1150
17
Mass Loading Timeseries

• How much mass per size range?
• For each measurement sample
• Determine model cell (col, row, layer)
• Determine nearest model output timestep, ts
• Use model concentration at (ts, col, row, layer)
  for selected sizes
• Smooth by averaging any contiguous (ts, col,
  row, layer)
• Plot measurements by stacked lines
• Plot model predictions by stacked patches of
  same colour
• Difficulties in this comparison
• Plane moves quickly, model outputs less
  frequently
• future field campaigns taking this on board
• Looking at 3D visualization options
• (vis5d? AVS?)

18
Nitrate Timeseries
StandardCMAQ (modal)
MADRIDhybrid CITequi with het chem(8
sectional)
19
Sulphate Aerosol at Ground Layer
20
Dry Deposition
21
Compare mass-size distributions
CMAQ only has sulphate in Aitken and Accumulation
modes nothing in the Coarse mode. This limits
the amount of aerosol mass in largest sections
exactly those sections that will have highest
rates of deposition. No such limitation exists
for sectional MADRID.
22
Mass differences in largest section at time of
interest
Scatter plot for mass in largest section over all
timesteps
23
Conclusions Ongoing/Future Work

• Conclusions
• Models-3 is suitable framework for advancing our
  understanding of aerosol processes analysing
  measurements
• A sectional approach seems more suitable than a
  modal approach
• Ongoing/Future work
• Firmly establish suitability of MADRID
• More detailed comparisons additional aerosol
  species and flights also ground-based
  measurements
• Use of Met Office UM rather than MM5 as met
  driver
• Better emissions UK size-resolved segregation
  (and more recent emissions) from ongoing NCAS
  work
• Improving MADRID
• Improve treatment of heterogeneous chemistry
• Use kinetic gas/particle dis-equilibrium mass
  transfer
• Improve SOA treatment
• Increase sections within MADRID
• Use improved chemistry schemes (RADM2 no longer
  supported)
• Use model in operational mode