receptor modelling of uk atmospheric aerosol
DESCRIPTION
RECEPTOR MODELLING OF UK ATMOSPHERIC AEROSOL. Roy M. Harrison University of Birmingham and National Centre for Atmospheric Science. RECEPTOR MODELLING TECHNIQUES. Multicomponent analysis in many samples followed by factor analysis (usually PMF) - PowerPoint PPT PresentationTRANSCRIPT
RECEPTOR MODELLING OF UK ATMOSPHERIC AEROSOL
Roy M. HarrisonUniversity of Birmingham
and National Centre for Atmospheric Science
RECEPTOR MODELLING TECHNIQUES1. Multicomponent analysis in many samples followed by
factor analysis (usually PMF)- we have applied to PAH and to particle number size distributions
2. Use of chemical tracers, including organic molecular markers and Chemical Mass Balance modelling- we have applied to urban and rural PM2.5
3. Targeted studies- e.g. work on brake dust particles
4. Aerosol mass spectrometry
ROADSIDEURBAN
BACKGROUND RURAL
PM10
(BROS)PM10
(BCCS)PM10
(CPSS)
Major Component Composition of PM10
Receptor Modelling Using Organic Molecular Source Tracers
• Uses approaches developed in California and mostly US source profiles
• Considers atmospheric PM chemical composition to be a linear sum of relevant source emission profiles (Chemical Mass Balance model)
• Two sites: Urban background Rural
Chemical Mass Balance Study using Molecular Markers
• PM2.5 samples were collected and analysed for
n-alkanes from C24 – C36 9 specific hopanes 13 PAH 14 carboxylic acids levoglucosan cholesterol inorganic marker elements (Si, Al)
CMB Model Results
• Model used to apportion sources of organic carbon to:
diesel engine exhaust gasoline engines smoking gasoline engines vegetative detritus dust and soil wood smoke coal combustion natural gas combustion
Source Contributions to OC at Urban Background Site
Summer Winter Annual0.0
0.7
1.4
2.1
2.8
3.5
EROSOther OC
Dust/Soil
Coal
Smoking Engines
Gasoline Engines
Diesel Engines
Natural Gas
Woodsmoke
VegetationOC
con
trib
utio
ns (
mg
m -3
)
Source Contributions to OC at Rural Site
Summer Winter Annual0.0
0.5
1.0
1.5
2.0
2.5
3.0
CPSS
Other OC
Dust/Soil
Coal
Smoking Engines
Gasoline Engines
Diesel Engines
Natural Gas
Woodsmoke
Vegetation
OC
con
trib
utio
ns (
mg
m -3
)
EROS
y = 1.01x + 0.34R2 = 0.92
0.0
2.0
4.0
6.0
0.0 1.0 2.0 3.0 4.0 5.0Other OC
Sec
OC
Relationship of “Other OC” from CMB Model with Secondary OC from Graphical Method
(µg m-3)
(µg
m-3)
Main Conclusions from CMB Model
• Road traffic contribution to primary OC is dominant.
• Split between diesel, gasoline and gasoline smoker emissions requires further study.
• Vegetative detritus is significant at the rural site.
• Small contributions from coal and natural gas combustion, very small from meat cooking.
• “Other” OC correlates highly with secondary OC estimated by the method of Castro et al. (1999).
• Wood smoke contribution is small, but studies at other sites using a multi-wavelength aethalometer show substantial concentrations.
Sources of particles from a vehicle
Emissions dependent upon• vehicle speed (resuspension, tyre and road surface wear)• engine revs and load (exhaust)• driving mode (exhaust, brake, tyre, road surface)• materials (brakes, tyres, road surface)• fuel and lubricant (exhaust)• vehicle weight and aerodynamics (resuspension)• road surface silt loading (resuspension)
exhaust
resuspension
tyre wear
road surface wear
brake wear
Median Concentrationof PM10
Studies of Non-Exhaust Particles at Marylebone Road – Chemical Composition
as a Tracer
Ba and Cu are clear tracers of brake wear particles
Al appears most plausible tracer for resuspension, but this appears difficult
Tyre wear remains a problem
Size Distribution ofBa, Cu, Fe, and Sbat (a) Marylebone Roadand (b) Regent’s Park
a
b
Inorganic coarse particles rich in Fe and Ba, Specific fingerprint for Brake wear
154
13863
5656392723
88
72
43
60
35
32
2617
19
0 50 100 150 200 250
Da = 1.25 µm
m/z
Barium
Aluminium
Iron
Iron Oxide
Single Particle Mass Spectrometry
CONCLUSIONS• Receptor modelling techniques are a blunt tool but nonetheless
can identify components which emissions inventories are poor at quantifying.
• To a large extent receptor modelling techniques (especially CMB) will only find what you tell them to look for.
• There is much scope for extending receptor modelling methods to reveal more, especially by exploiting newer techniques (e.g. high resolution aerosol mass spectrometry) and by using techniques in combination. This will be expensive.
ACKNOWLEDGEMENTS – to collaborators who collected the data,especially Dr Jianxin Yin, Dr David Beddows and Dr Johanna Gietl.