Post on 10-Mar-2020
Closure of particle backscattering coefficient in oligotrophic waters
ZhongPing Lee 1, Yannick Huot 2
1. University of Massachusetts Boston
2. Université de Sherbrooke
Acknowledgements: NASA, Jim Sullivan, Robert Brewin, Giorgio Dall’Olmo
particle backscattering coefficient: bbp
(Behenfeld et al, Nature, 2006)
bbp Ccc (Balch et al. 2005, 2010)
(Stramski et al, Science, 1999)
Bulk optical property
NOMAD
0.000 0.001 0.002 0.003 0.004 0.005 0.006
0.000
0.001
0.002
0.003
0.004
0.005
0.006
Rrs_inv
reg_no
R2 = 0.92 Y = 1.07 X + 0.0004
“Excellent” closure …
In-situ bbp(555) [m-1]
Rrs
bb
p(5
55
) [
m-1
]
1:1
(Huot et al 2008)
However:
(Brewin et al 2012)
Rrs
bbp much higher than in-situ bbp
for oligotrophic waters!
No closure for such ‘simple’ waters !!
Chl < 0.1 mg/m3 makes ~50% of the global surface waters
)555()555(
)555(
)555()555(
)555(
)555()555(
)555()555()555(
10b
b
b
b
b
brs
ba
b
ba
bGG
ba
bGR
Brief review of QAA:
Rrs(555) bbp(555)
Based on:
)555()555()555()555( ww aaaa For oligotrophic waters
For Chl = 0.1 mg/m3, Δa(555) ~ 0.002 m-1, 3% of aw(555).
Potential sources of error from Rrs inversion:
1. Rrs–IOPs relationship
2. Measured Rrs includes Raman scattering contribution
3. a(555) or aw(555) value
b
b
b
b
b
brs
ba
b
ba
bGG
ba
bGR
10
is supported by Radiative Transfer Theory (Zaneveld 1995)
bbp555
0.000 0.001 0.002 0.003 0.004 0.005
0.5
1.0
1.5
2.0
2.5
3.0
NOMAD
In-situ bbp(555) [m-1]
Rrs
bb
p(5
55
)/In
-sit
u b
bp(5
55
)
Better for larger bbp(555), but Rrs
bbp(555) is higher
1. Rrs – IOPs relationship
Same bbw(555) used for both determinations.
Impact of Rrs-model parameters Gordon (0.0949;0.0794) vs QAA (0.09;0.125)
X Data
0.000 0.001 0.002 0.003 0.004 0.005
Y D
ata
1.00
1.05
1.10
1.15
In-situ bbp(555) [m-1]
bb
p(Q
AA
)/b
bp(G
ord
on
)
Not enough to have a factor of 2 impact.
2. Measured Rrs includes Raman scattering contribution
(Westberry et al 2013)
RF
RR
Trs
rs
1
Empirical Raman correction (Lee et al 2013):
)(
12
)550()()550(
)440()()(
T
rsTrs
Trs R
R
RRF
RF: Raman Factor
:TrsR Rrs from measurements
NOMAD
In-situ bbp(555) [m-1]
Rrs
bb
p(5
55
)/In
-sit
u b
bp(5
55
)
bbp555
0.000 0.001 0.002 0.003 0.004 0.005
0.5
1.0
1.5
2.0
2.5
3.0
no Raman corr.
yes Raman corr.
Yes, remove Raman effect reduces bbp from Rrs
BIOSOPE data
bbp555
0.0000 0.0005 0.0010 0.0015
0.5
1.0
1.5
2.0
2.5
3.0
3.5
no Raman corr.
yes Raman corr.
In-situ bbp(555) [m-1]
Rrs
bb
p(5
55
)/In
-sit
u b
bp(5
55
)
Rrs
bbp(555) is still generally much higher than in-situ bbp(555), especially for waters with very sparse in particles.
)(
12
)550()()550(
)440()()(
T
rsTrs
Trs R
R
RRF
Imperfect Raman correction?
X Data
0.000 0.001 0.002 0.003 0.004 0.005
Y D
ata
0.95
0.96
0.97
0.98
0.99
1.00
RF increased by 15%
In-situ bbp(555) [m-1]
bb
p(1
5%
mo
re R
F)/b
bp
3. a(555) or aw(555) value
aw(555) bbp(555)
Reference aw(555)
Pope and Fry (1997) 0.0596
Smith and Baker (1981) ~0.0673
Tom and Patel (1979) ~0.063
Sogandares and Fry (1997) ~0.072
Buiteveld et al (1994) 0.064
The smallest value for aw(555) was used.
Potential sources of errors from in situ bbp:
1. Calibration
2. Sampling volume?
3. Measurement uncertainty?
Backscatter (active) sensor
Passive sensor Sampling volume of an active sensor
Sampling volume of a passive sensor
~ 10-6 m3 ~10-1000 m3
2. Sampling volume?
“bulk” property?
(Stramki and Kiefer 1991) >100 particles will be sampled by the 10-6 m3 sample volume
(Stramki and Kiefer 1991) Particles could be under-represented (or missed) by 10-6 m3 volume
(Dall’Olmo and Brewin)
Median vs Mean
Sample volume seems not a big issue, if averaged/handled properly.
Treat 1 min of measurements as “bulk”
3. Measurement uncertainty?
(Brewin et al 2012)
In-s
itu
bb
p(5
26
) [
m-1
] (Huot et al 2008)
bbp(555): BB3: ~0.0004 m-1
HSCAT: ~0.0007 m-1
For Chl = 0.1 mg/m3
~0.0007 m-1
bbp555
0.0000 0.0005 0.0010 0.0015
0.5
1.0
1.5
2.0
2.5
insitu bbp
insitu bbp + 0.00025
In-situ bbp(555) [m-1]
Rrs
bb
p(5
55
)/In
-sit
u b
bp(5
55
)
Measured bbp(555) + 0.00025
BIOSOPE data
Much better closure for oligotrophic waters!
If indeed insitu sensor missed (under-measured) bbp
Updated comparison
Summary:
1. For oligotrophic ocean, bbp(55x) can be retrieved very well from Rrs. Important to correct Raman effect.
2. We still have a (small) gap between inversion and insitu, though.
• Representation of “bulk” product • Extremely low signal Insitu sensor calibration and data handling
3. If ignoring the ~0.0003 m-1 bias, “excellent” closure is indeed achieved between inverted and insitu bbp(55x).
Something about spectral resolution
400 500 600 700 800
0.00
0.01
0.02
0.03
Wavelength [nm]
Rrs(λ)
[sr
-1]
(a)
(Lee, Hu, Shang, Zibordi, Applied Optics, in press)
901 spectra
400 450 500 550 600 650 700
0.90
0.92
0.94
0.96
0.98
1.00
5 nm
10 nm
15 nm
20 nm
25 nm
30 nm
Wavelength [nm]
Correlation coefficient between neighboring bands, for 6 different gaps
r Δλ(λ k
, λl)
Rrs is highly correlated between neighboring bands
400 450 500 550 600 650 700
0.00
0.03
0.06
0.09
Col 1 vs Col 3
Col 1 vs Col 4
Col 1 vs Col 5
Col 1 vs Col 6
Col 1 vs Col 7
Col 10 vs Col 11
Col 10 vs Col 12
Col 10 vs Col 13
Col 10 vs Col 14
Col 10 vs Col 15
Col 18 vs Col 19
Col 18 vs Col 20
Col 18 vs Col 21
Col 18 vs Col 22
Col 18 vs Col 23
Wavelength [nm]
Rrs
(λ)
[sr
-1]
400 450 500 550 600 650 700
0.000
0.004
0.008
0.012
0.016
Wavelength [nm]
Rrs
(λ)
[sr
-1]
Re-constructed vs measured spectral Rrs
15
1
)()(i
irsjijrcrs RKR
400 450 500 550 600 650 700
0.9998
0.9999
1.0000
1-nm
10-nm
400 450 500 550 600 650 700
0
1
2
3
1-nm
10-nm
Wavelength [nm]
Co
rrel
atio
n C
oef
fici
ent
Perc
enta
ge d
iffe
ren
ce [
%]
Wavelength [nm]
(a)
(b)
Characteristics between measured and re-constructed spectral Rrs
Hyperspectral (contiguous , 5-nm resolution) Rrs can be reconstructed from 15-band Rrs with negligible error.
Thank you!