particle size distributions: data analysis
TRANSCRIPT
Particle Size Distributions:Data Analysis
Susanne Hering,Mark Stolzenburg,Nathan Kreisberg,
Aerosol Dynamics Inc.
Size DistributionMeasurement System
• SMPS– scanning mobility particle spectrometer– particle sizing by electrical mobility,– (assume equilibrium charge distribution)– size range: 10 nm - 350 nm
• LASAIR– optical sizing– size range: 120 nm to 2000 nm (2 µm)
• Climet– optical sizing– size range: 500 nm to 10000 nm (10 µm)
Combined Data: Number Distribution
10-2
10-1
100
101
102
103
104
105
106
Parti
cle N
umbe
r Dist
ribut
ion
(dN/
Dlog
Dp, #
/cm
3)
102 3 4 5 6
1002 3 4 5 6
10002 3 4 5 6
Particle Diameter (nm)
SMPS Mobility Sizing LASAIR Optical Sizing Climet Optical Sizing
SMPS: 2/1/2001 19:09LASAIR: 2/1/2001 19:10Climet: 2/1/2001 19:10
Fresno First Street
Same Data as Particle Volume Distribution
120
100
80
60
40
20
0
Parti
cle V
olum
e Di
stribu
tion
(dV/
Dlog
Dp, µ
m3/
cm3)
102 3 4 5 6
1002 3 4 5 6
10002 3 4 5 6
Particle Diameter (nm)
Volume Distribution SMPS LASAIR Climet
SMPS: 2/1/2001 19:09LASAIR: 2/1/2001 19:10Climet: 2/1/2001 19:10
30x103
25
20
15
10
5
0
Num
ber C
once
ntra
tion
(#/cm
3)
12/15/2000 12/17/2000 12/19/2000 12/21/2000 12/23/2000 12/25/2000 12/27/2000 12/29/2000
110-220 nm
LASair Counts SMSP Counts
4000
3000
2000
1000
0
Num
ber C
once
ntra
tion
(#/cm
3)
12/15/2000 12/17/2000 12/19/2000 12/21/2000 12/23/2000 12/25/2000 12/27/2000 12/29/2000
220-316 nm
LASair Counts SMSP Counts
Systematic Difference Among Instruments
Lowest LASair channel fewer counts at peaks
Second LASair channel always higher than SMPS
Why the Differences ?
• SMSP has assumes an equilibrium particle charge distribution, andcorrects measured counts for this assumed charge fraction
• Optical and mobility sizing differ, but performed DOS and PSLcalibrations on optical instruments. Differences persist with eithercalibration factors.
How to Handle these Differences ?• Treat each sizing method independently.
– Ultrafine=> SMPS; Accumulation=> LasAir; Coarse => Climet
• Merging into a single distribution will produce inherent “bumps” attransition.
40x103
30
20
10
Parti
cle N
umbe
r Dist
ribut
ion (d
N/Dl
ogDp
, #/c
m3)
102 3 4 5 6 7 8 9
1002 3
Particle Diameter (nm)
SMPSNumber
2/1/2001 23:59
2/2/2001 08:00
Midnight
Following Morning, 8am
SMPS Particle Number Distribution
See multiple modes
SMPS Distributions:Bimodal, Lognormal Deconvolution
,2
,,
ln( / )exp log( )
2(ln )2 ln
a b
p pm aaa p
g ag a
dN dN dN
D DNdN d Dσπ σ
= +
= −
Each size distribution is expressed as a sum oftwo lognormal distributions:
Size Distribution MovieFresno First Street
Feb 1 – 3, 2001
Look for:• How well the bimodal deconvolution generally captures
the major characteristics of the distribution.
• Differences in character from nighttime and day timedistributions.
Can we use the Bimodal, LogNormal Parametersto Quantify these Observations?
For each mode:
N = number of particles
Dgm = geometric mean diameter
σg = geometric standard deviation
With two modes => 6 parameters.
How well do these parameters to characterize thetemporal and spatial variablity of the size distributions?
Mode Diameters: Jan 31-Feb 3, 2001 Data
200
150
100
50
0
Mod
e Di
amet
er, n
anom
eter
s
12:00 AM1/31/2001
12:00 PM 12:00 AM2/1/2001
12:00 PM 12:00 AM2/2/2001
12:00 PM 12:00 AM2/3/2001
12:00 PM
Mode A Dp Mode B Dp
larger ultrafine particle size midday
Comparison with Black Carbon80x103
60
40
20
0
Num
ber C
once
ntra
tion
(#/c
m3)
12:00 AM1/31/2001
12:00 PM 12:00 AM2/1/2001
12:00 PM 12:00 AM2/2/2001
12:00 PM 12:00 AM2/3/2001
12:00 PM
20
15
10
5
0
Black Carbon (ug/m
3)
Black Carbon (ug/m3) SMPS Number Concentration
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Frac
tion
in M
ode
A (s
mal
lest
)
12:00 AM1/31/2001
12:00 PM 12:00 AM2/1/2001
12:00 PM 12:00 AM2/2/2001
12:00 PM 12:00 AM2/3/2001
12:00 PM
20
15
10
5
0
Black Carbon (ug/m
3)
Fraction in smallest mode Black Carbon (ug/m3)
•Ultrafine number tracks black carbon •Lower fraction of particle number in smallest mode
Summary: Size Distribution Analysis
• Discrepancies among instruments makes it difficult to synthesizedata into a uniform distribution.
• Modal analysis from one instrument (SMPS) provides means to– Investigate temporal variations– Comparison with chemical constituents
• Ultrafine distributions more stable at night– Small mode (A) appears in early morning– Larger mode (B) has larger diameter midday
• Black carbon and SMSP number track, while the fraction associatedwith the small mode (A) varies inversely with BC level.
TEMPORAL AND SPATIALDISTRIBUTION OF PARTICULATE
NITRATE AND LIGHTSCATTERING
Susanne Hering, Nathan Kreisberg,Aerosol Dynamics Inc.,
Judith Chow, John Watson,Desert Research Institute,
L. Willard Richards,Sonoma Technology Inc.
Sites and Constituentscompared using 10-min database
Multiday trends separated fromdaily pattern.
Each frequency domaincompared separately.
605040302010
0PM2.
5 N
itrat
e (u
g/m
3)
12/21/00 1/10/01 1/30/01
800
600
400
200
0
Parti
cle
Scat
terin
g at
<70
% R
H (M
m-1
)
BAKERSFIELD: 24-hr Running Average Nitrate Bsp
-40
-20
0
20
40
PM2.
5 N
itrat
e (u
g/m
3)
12/21/00 1/10/01 1/30/01
-500
0
500
Parti
cle
Scat
terin
g at
<70
% R
H (M
m-1
)
Bakersfield: Residual (24-hr running average subtracted) Nitrate Bsp
80
60
40
20
0PM2.
5 N
itrat
e (u
g/m
3)
12/21/00 1/10/01 1/30/01
120010008006004002000
Parti
cle
Scat
terin
g at
<70
% R
H (MBAKERSFIELD: Raw Data
Nitrate Bsp
=
+
=
Comparison among sites for 24-hr average
40
30
20
10
0
PM2.
5 Ni
trate
(ug/
m3)
12/11/2000 12/21/2000 12/31/2000 1/10/2001 1/20/2001 1/30/2001
1000
800
600
400
200
0
Parti
cle
Scat
terin
g at
<70
% R
H (M
m-1
)
Running 24-hr Average ANGIOLA Nitrate ANGIOLA Bsp FRESNO Nitrate FRESNO Bsp
Residual, after subtracting running 24-hr average
-40
-20
0
20
40
PM2.
5 Ni
trate
(ug/
m3)
12/11/2000 12/21/2000 12/31/2000 1/10/2001 1/20/2001 1/30/2001
-500
0
500
Parti
cle
Scat
terin
g at
<70
% R
H (M
m-1
)
Bakersfield: Residual (24-hr running average subtracted) Nitrate Bsp
Comparison among Constituents in 24-hr Pattern
-4
-2
0
2
4
PM2.
5 Ni
trate
(ug/
m3)
20151050
Hour of Day
-100
0
100
Parti
cle S
catte
ring
at <
70%
RH
(Mm
-1)
Bakersfield Nitrate Bsp
-1.0
-0.5
0.0
0.5
1.0
PM2.
5 Ni
trate
(ug/
m3)
20151050
Hour of Day
-60
-40
-20
0
20
40
60
Parti
cle S
catte
ring
at <
70%
RH
(Mm
-1)
Angiola Nitrate Bsp
-2
0
2
PM2.
5 Ni
trate
(ug/
m3)
20151050
Hour of Day
-200
-100
0
100
200
Parti
cle S
catte
ring
at <
70%
RH
(Mm
-1)
Fresno Nitrate Bsp
-4
-3
-2
-1
0
1
2
3
4
PM2.
5 Ni
trate
(ug/
m3)
20151050
Hour of Day
-60
-40
-20
0
20
40
60
Parti
cle S
catte
ring
at <
70%
RH
(Mm
-1)
Sierra Nevada Foothills Nitrate Bsp
Bsp and Nitrate Analysis
• Subtract out overall multi-day pattern to discern the daily pattern,and how this varies among constituents and sites
• Nitrate maxima occurs earlier in the day at Bakersfield and Fresnothan at Angiola or Sierra Nevada Foothills.
• Note there is also a seasonal pattern -- not shown here.