Photo-acoustic and Filter Absorption Inter-Comparison at ... · variability (1 sigma) in signal...
1
• Direct, online measurement of aerosol absorption & scattering • 781, 532, 405 nm • Absorption (β abs ) • Scattering (β sca ) • Angstrom Exponent (AE) • 2013 & 2015 SGP deployment • 2015 upgrade 532 nm laser power increased by factor of 4. • PSAP Absorption measurements • 660, 530, 467 nm • ACSM chemical composition data • Nephelometer scattering data • 700, 530, 467 nm • 1 hour averages used in analysis Photo-acoustic and Filter Absorption Inter-Comparison at SGP: Quantifying Biases in Long-Term Records Luis A. Torres 1 , Allison C. Aiken 1 , Manvendra K. Dubey* 1,2 , Anne Jefferson 3 1 Los Alamos National Laboratory 2 National Science Foundation 3 CIRES/NOAA, *[email protected] Summary • Subramanium et al. AS&T 41, 630–637, 2007 • Lack et al. AS&T, 42, 1033–1041, 2008. • Cappa et. al. AS&T, 42, 1022–1032, 2008 • Cross, E.S. et al., AS&T, 44, 592-611, 2010. • Flowers, B.F. et al., ACP, 10,10387-98, 2010. PASS-3, PSAP, Neph. & ACSM at SGP References LA-UR-13-21854 0 50 100 0 20 40 60 80 100 120 NEPH λ Corrected (Mm -1 ) 2013 2015 2013 Regression 2015 Regression 1:1 0 20 40 60 80 PASS (Mm -1 ) 0 10 20 30 40 50 60 70 80 PASS vs NEPH - 2013 SAE = 2.1491 vs 2015 SAE = 1.8204 2013 2015 2013 Regression 2015 Regression 1:1 0 10 20 30 40 50 0 5 10 15 20 25 30 35 40 45 50 2013 2015 2013 Regression 2015 Regression 1:1 y = (1.03± 0.006)*X r 2 =0.95 y = (1.23± 0.01)*X r 2 =0.95 y = (1.36± 0.003)*X r 2 =0.97 y = (0.92± 0.007)*X r 2 =0.95 y = (1.52± 0.02)*X r 2 =0.83 y = (2.02± 0.28)*X r 2 =0.84 0 10 20 ACSM Total Organics ( μ g/m 3 ) -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 PSAP/PASS Absorption Ratio 2013 2015 2013 2015 Lack et. all 60 80 100 120 140 160 180 200 220 240 Laser Power (mW) 0 5 10 15 Uncertainty (σ ) 2013 910 920 930 940 950 960 970 980 2015 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 PSAP/PASS Ratio 0 50 100 150 200 250 300 Counts PSAP/PASS Ratio Histogram: Bin Size = 0.2 PSAP/PASS 2013 PSAP/PASS 2015 0 5 10 15 20 0 2 4 6 8 10 12 14 16 18 20 PSAP λ Corrected (Mm -1 ) 2013 2015 2013 Regression 2015 Regression 1:1 0 5 10 15 PASS-3 (Mm -1 ) 0 5 10 15 PASS vs PSAP - 2013 AAE = 1.072 vs 2015 AAE = 0.85978 2013 2015 2013 Regression 2015 Regression 1:1 0 5 10 -1 0 1 2 3 4 5 6 7 8 9 10 2013 2015 2013 Regression 2015 Regression 1:1 y = (1.78± 0.05)*X r 2 = 0.83 y = (2.84± 0.44)*X r 2 = 0.66 y = (2.38± 0.13)*X r 2 = 0.80 y = (1.86± 0.14)*X r 2 = 0.77 y =(2.22 ± 0.08)*X r 2 =0.68 y = (4.1± 1.3)*X r 2 = 0.18 0 60 120 180 240 B sca 405 nm PASS & NEPH 2015 Timeseries (Mm -1 ) PASS 2015 NEPH 2015 0 40 80 120 160 200 B sca 532 nm PASS 2015 NEPH 2015 Jun 27 Jul 07 Jul 17 Jul 27 Aug 06 Aug 16 Aug 26 Sep 05 Sep 15 Sep 25 -20 20 60 100 B sca 781 nm PASS 2015 NEPH 2015 • Lower uncertainty and noise in 532 nm absorption at higher laser power in 2015. • PASS-3 performance is much improved • PSAP and PASS-3 are well correlated • PSAP is biased high relative to PASS-3 • PSAP bias persists in 2015 • PSAP bias correlates with OA in 2013 • Retroactive corrections of PSAP record at SGP -2 12 26 40 B abs 405 nm PASS & PSAP 2015 Timeseries (Mm -1 ) PASS 2015 PSAP 2015 -2 12 26 36 B abs 532 nm PASS 2015 PSAP 2015 Jun 27 Jul 07 Jul 17 Jul 27 Aug 06 Aug 16 Aug 26 Sep 05 Sep 15 Sep 25 -2 6 14 22 30 B abs 781 nm PASS 2015 PSAP 2015 B abs B scatter B abs B scatter 405 nm: 1.57 ± 1.58 25.87 ± 18.54 1.61 ± 1.60 51.85 ± 31.32 532 nm: 1.02 ± 0.99 16.83 ± 12.33 1.48 ± 0.95 23.05 ± 16.68 781 nm: 0.58 ± 0.67 4.79 ± 3.87 0.80 ± 2.29 7.31 ± 6.89 Figure 2: a) Time series of PASS-3 and PSAP absorption data. b) Correlations of PASS-3 and PSAP absorption extrapolated using AE to PASS-3 wavelengths and fits for 2013 and 2015 c) Time series of PASS-3 and Nephelometer scattering data d) Correlation of PASS-3 and Nephelometer scattering extrapolated using AE to PASS-3 wavelengths and fits for 2013 and 2015 a) b) c) d) Figure 3 : PSAP/PASS Ratio vs Organic Aerosols at 532 nm. Lack et. al. and 2013 show increasing bias with organic loading in contrast to 2015. Mean 2013: -0.84 ± 99.45 2015: 2.19 ± 3.49 BIAS vs OA trend fits: 2013 : y = (0.06 ± 0.48 )*x + (1.23 ± 3.41) 2015 : y = (0.009 ± 0.004 )*x + (2.01 ± 0.04) Lack : y = (0.08 ± 0.01 )*x + (1.2 ± 0.1) The potential for biases in filter-based light absorption has been recognized. This calls into question the quality of long-term baseline absorption observations that are important to evaluate climate models. To resolve this problem we report long term direct filter free observations of optical properties with a 3-wavelength photo-acoustic spectrometer (PASS-3) of both light absorption and scattering at 781, 532 & 405 nm at SGP in 2013 and 2015. We compare our results with absorption measurements made by baseline filter-based measurements (PSAP at 660, 530 & 467 nm) and the direct scattering observations made with a nephelometer (at 700, 550 & 450 nm). All instruments were located on the same inlet and the optical observations were compared at the same wavelength using extrapolations with measured Angstrom exponents. One hour averaging was used for our comparison to gain sufficient signal/noise for the PASS. Our analysis shows that while the absorption measured by the PASS and PSAP are well correlated the PSAP is biased high by a factor of 2.2, 1.9 and 1.8 in 2013 and by 2.8, 2.4, and 4.1 in 2015 at 781, 532 and 405nm respectively. The PASS was upgraded with a high power green laser and deployed at SGP in 2015. We find that the signal/noise of absorption at 532nm improved significantly in 2015 but the biases persisted. The PASS scattering measurements are noisy but agree with those measured by the nephelometer. We find that the PSAP high absorption bias increased with aerosol organic content in 2013 consistent with earlier results but not in 2015. Our results can be used to empirically scale the historic PSAP record at SGP for more reliable climate model evaluation. PASS-3, PSAP & Neph. Comparison PSAP/PASS-3 versus OA Acknowledgement This research is funded by DOE ASR grant F265 to LANL and instrument mentoring and data archiving funded by DOE ARM program. MKD alo thanks NSF Atmospheric Chemistry for support. 532nm PASS-3 Laser Upgrade Figure 4: (Above) Uncertainty in 532 absorption as measured by variability (1 sigma) in signal versus laser power, 2013 in red, 2015 in blue. Uncertainty and scatter are both reduced by a factor of 2-5 in 2015 Figure 5: ) Histogram PSAP/PASS3 ratio measured in 2013 (green) and 2015 (red) show a bias of 2 in the 532 nm absorption. 2015 bias is less variable due to lower noise at higher power. Figure 1. : Aerosol Observing System at SGP where all measurements were made in the same inlet and conditions. 2013 2015
Transcript of Photo-acoustic and Filter Absorption Inter-Comparison at ... · variability (1 sigma) in signal...
• Direct, online measurement of aerosol absorption & scattering• 781, 532, 405 nm• Absorption (βabs) • Scattering (βsca)• Angstrom Exponent (AE)• 2013 & 2015 SGP deployment• 2015 upgrade 532 nm laser power
increased by factor of 4.• PSAP Absorption measurements
• 660, 530, 467 nm• ACSM chemical composition data• Nephelometer scattering data
• 700, 530, 467 nm• 1 hour averages used in analysis
Photo-acoustic and Filter Absorption Inter-Comparison at SGP: Quantifying Biases in Long-Term Records
Luis A. Torres1, Allison C. Aiken1, Manvendra K. Dubey*1,2, Anne Jefferson31Los Alamos National Laboratory 2National Science Foundation 3CIRES/NOAA, *[email protected]
Summary
• Subramanium et al. AS&T 41, 630–637, 2007• Lack et al. AS&T, 42, 1033–1041, 2008.• Cappa et. al. AS&T, 42, 1022–1032, 2008• Cross, E.S. et al., AS&T, 44, 592-611, 2010. • Flowers, B.F. et al., ACP, 10,10387-98, 2010.
PASS-3, PSAP, Neph. & ACSM at SGP
References
LA-UR-13-21854
0 50 1000
20
40
60
80
100
120
NE
PHλ
Co
rrec
ted (
Mm
-1)
201320152013 Regression2015 Regression1:1
0 20 40 60 80PASS (Mm-1)
0
10
20
30
40
50
60
70
80PASS vs NEPH - 2013SAE = 2.1491 vs 2015SAE = 1.8204
201320152013 Regression2015 Regression1:1
0 10 20 30 40 500
5
10
15
20
25
30
35
40
45
50
201320152013 Regression2015 Regression1:1
y = (1.03±0.006)*Xr2 =0.95y = (1.23±0.01)*Xr2 =0.95
y = (1.36±0.003)*Xr2 =0.97y = (0.92±0.007)*Xr2 =0.95
y = (1.52±0.02)*Xr2 =0.83y = (2.02±0.28)*Xr2 =0.84
0 10 20ACSM Total Organics (µg/m3)
-5-4-3-2-10123456789
10
PS
AP
/PA
SS
Ab
sorp
tio
n R
atio
2013201520132015Lack et. all
60 80 100 120 140 160 180 200 220 240 Laser Power (mW)
0
5
10
15
Unc
erta
inty
(σ)
2013910 920 930 940 950 960 970 980
2015
-10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10PSAP/PASS Ratio
0
50
100
150
200
250
300
Coun
ts
PSAP/PASS Ratio Histogram: Bin Size = 0.2PSAP/PASS 2013PSAP/PASS 2015
0 5 10 15 20
0
2
4
6
8
10
12
14
16
18
20
PSA
Pλ
Cor
rect
ed (M
m-1
)
201320152013 Regression2015 Regression1:1
0 5 10 15PASS-3 (Mm-1)
0
5
10
15PASS vs PSAP - 2013AAE = 1.072 vs 2015AAE = 0.85978
201320152013 Regression2015 Regression1:1
0 5 10-1
0
1
2
3
4
5
6
7
8
9
10
201320152013 Regression2015 Regression1:1
y = (1.78±0.05)*X r2 = 0.83y = (2.84±0.44)*Xr2 = 0.66
y = (2.38±0.13)*Xr2 = 0.80
y = (1.86±0.14)*Xr2 = 0.77
y =(2.22±0.08)*Xr2 =0.68y = (4.1±1.3)*X r2 = 0.18
0
60
120
180
240
Bsc
a 405
nm
PASS & NEPH 2015 Timeseries (Mm-1)
PASS 2015NEPH 2015
04080
120160200
Bsc
a 532
nm PASS 2015
NEPH 2015
Jun 27 Jul 07 Jul 17 Jul 27 Aug 06 Aug 16 Aug 26 Sep 05 Sep 15 Sep 25-20
20
60
100
Bsc
a 781
nm PASS 2015
NEPH 2015
• Lower uncertainty and noise in 532 nm absorption at higher laser power in 2015.
• PASS-3 performance is much improved
• PSAP and PASS-3 are well correlated• PSAP is biased high relative to PASS-3• PSAP bias persists in 2015• PSAP bias correlates with OA in 2013• Retroactive corrections of PSAP record at SGP
-2
12
26
40
Bab
s 405
nm
PASS & PSAP 2015 Timeseries (Mm-1)PASS 2015PSAP 2015
-2
12
26
36
Bab
s 532
nm PASS 2015
PSAP 2015
Jun 27 Jul 07 Jul 17 Jul 27 Aug 06 Aug 16 Aug 26 Sep 05 Sep 15 Sep 25-2
6
14
22
30
Bab
s 781
nm PASS 2015
PSAP 2015
Babs Bscatter Babs Bscatter
405 nm: 1.57 ± 1.58 25.87 ± 18.54 1.61 ± 1.60 51.85 ± 31.32532 nm: 1.02 ± 0.99 16.83 ± 12.33 1.48 ± 0.95 23.05 ± 16.68781 nm: 0.58 ± 0.67 4.79 ± 3.87 0.80 ± 2.29 7.31 ± 6.89
Figure 2: a) Time series of PASS-3 and PSAP absorption data. b) Correlations of PASS-3 and PSAP absorption extrapolated using AE to PASS-3 wavelengths and fits for 2013 and 2015 c) Time series of PASS-3 and Nephelometer scattering data d) Correlation of PASS-3 and Nephelometer scattering extrapolated using AE to PASS-3 wavelengths and fits for 2013 and 2015
a)
b)
c)
d)
Figure 3 : PSAP/PASS Ratio vs Organic Aerosols at 532 nm. Lack et. al. and 2013 show increasing bias with organic loading in contrast to 2015.
Mean
2013: -0.84 ± 99.45
2015: 2.19 ± 3.49
BIAS vs OA trend fits:2013 : y = (0.06 ± 0.48 )*x + (1.23 ± 3.41)2015 : y = (0.009 ± 0.004 )*x + (2.01 ± 0.04)Lack : y = (0.08 ± 0.01 )*x + (1.2 ± 0.1)
The potential for biases in filter-based light absorption has been recognized. This calls into question the quality of long-term baseline absorption observations that are important to evaluate climate models. To resolve this problem we report long term direct filter free observations of optical properties with a 3-wavelength photo-acoustic spectrometer (PASS-3) of both light absorption and scattering at 781, 532 & 405 nm at SGP in 2013 and 2015. We compare our results with absorption measurements made by baseline filter-based measurements (PSAP at 660, 530 & 467 nm) and the direct scattering observations made with a nephelometer (at 700, 550 & 450 nm). All instruments were located on the same inlet and the optical observations were compared at the same wavelength using extrapolations with measured Angstrom exponents. One hour averaging was used for our comparison to gain sufficient signal/noise for the PASS. Our analysis shows that while the absorption measured by the PASS and PSAP are well correlated the PSAP is biased high by a factor of 2.2, 1.9 and 1.8 in 2013 and by 2.8, 2.4, and 4.1 in 2015 at 781, 532 and 405nm respectively. The PASS was upgraded with a high power green laser and deployed at SGP in 2015. We find that the signal/noise of absorption at 532nm improved significantly in 2015 but the biases persisted. The PASS scattering measurements are noisy but agree with those measured by the nephelometer. We find that the PSAP high absorption bias increased with aerosol organic content in 2013 consistent with earlier results but not in 2015. Our results can be used to empirically scale the historic PSAP record at SGP for more reliable climate model evaluation.
PASS-3, PSAP & Neph. Comparison PSAP/PASS-3 versus OA
AcknowledgementThis research is funded by DOE ASR grant F265 to LANL and instrument mentoring and data archiving funded by DOE ARM program. MKD alo thanks NSF Atmospheric Chemistry for support.
532nm PASS-3 Laser Upgrade
Figure 4: (Above) Uncertainty in 532 absorption as measured by variability (1 sigma) in signal versus laser power, 2013 in red, 2015 in blue. Uncertainty and scatter are both reduced by a factor of 2-5 in 2015
Figure 5: ) Histogram PSAP/PASS3 ratio measured in 2013 (green) and 2015 (red) show a bias of 2 in the 532 nm absorption. 2015 bias is less variable due to lower noise at higher power.
Figure 1. : Aerosol Observing System at SGP where all measurements were made in the same inlet and conditions.
2013 2015
