### Sensing primary production from ocean color

```Sensing primary production from ocean
color: Puzzle pieces and their status
ZhongPing Lee
University of Massachusetts Boston
An effort started half century ago …
From over 7000 measurements
Global PP:
~15 Gt/year
Longhurst et al (1995): 45-50 Gt C year-1
Antoine et al (1996): 36.5-45.6 Gt C year-1
Behrenfeld and Falkowski (1997): 43.5 Gt C year-1
~3 times higher than
estimates in the 50-60s’!
Large spatial differences from different sensing models
(Behrenfeld et al 2005)
Puzzle pieces to sense PP in the ocean
1. Input energy
3. Phytoplankton index
2. Light at depth
(spectral attenuation)
4. Energy conversion
(nutrient)
(Platt and Sathyendranath, 2007)
1. Input energy
PAR
(Frouin et al 1989, 2003)
(Gregg and Carder 1990)
2. Light at depth
(spectral attenuation)
E d (  , z )  E d (  ,0 ) e
Q: How to get Kd(λ) of varying water bodies?
 K d ( ) z
Algorithms to get Kd
Current operational standard
Method 1:
empirical
Kd
K d ( 490
)
Rrs
Rrs(λ1)/Rrs(λ2)
Method 2:
empirical
Rrs
Method 3:
empirical
[Chl]
Kd
Kd
Rrs(λ1)/Rrs(λ2)
semi-analytical
Rrs
semi-analytical
a&bb
(QAA)
KKdd
5
5
1 :1
A ra b ia n S e a
GOM
B a ltic
( 490
K
K dd (4
9 0)) [m ]
-1
1
0 .5
0 .3
0 .1
1
0 .5
0 .3
0 .1
0 .0 5
(a: Method 1)
(b: Method 2)
0 .0 5
0 .0 3
0 .0 3
0 .0 5
0 .1
0 .3
0 .5
1
3
5
0 .0 3
0 .0 5
-1
0 .1
0 .3
0 .5
1
3
5
-1
K dd(4
9 0 )) - m e a s u re d [m ]
( 490
KKd d(4
9 0 )) - m e a s u re d [m ]
( 490
5
1 :1
A ra b ia n S e a
GOM
B a ltic
3
-1
0 .0 3
1 :1
A ra b ia n S e a
GOM
B a ltic
3
9 0 )) [m ]
( 490
KK dd (4
-1
9 0)) [m ]
KKdd ((4
490
3
1
0 .5
Oceanic &
Coastal waters
0 .3
0 .1
0 .0 5
(c: Method 3)
0 .0 3
0 .0 3
0 .0 5
0 .1
0 .3
0 .5
1
3
5
-1
9 0)) - m e a s u re d [m ]
KKdd((4
490
(Lee et al. 2005)
Kd through IOPs
measured
sensed
0.06
0.04
0.02
0.00
350
400
450
500
The NOMAD set (1243 data points)
550
X Data[nm]
Wavelength
1
1:1
IOP
-1
IOPs-Kd(490)
Y Data [m ]
-1]
Spectral
K
[m
d
Y Data
0.08
0.3
0.1
0.03
0.01
0.01
0.03
0.1
X Data
Profile K
d(490)
0.3
[m-1]
1
Different sun angles:
Through IOPs
0.25
0.25
0.15
-1
IOP-basedY K
d(490) [m ]
Data
0.20
< 30
1.09
30-60
0.99
> 60
0.89
Y Data
Ratio-derived Kd(490) [m-1]
Empirical ratio
0.10
0.05
0.00
0.00
0.05
0.10
0.15
0.20
X Data
Profile-Kd(490)
[m-1]
0.20
0.15
< 30
0.99
30-60
1.00
> 60
1.00
0.10
0.05
0.00
0.00
0.05
0.10
0.15
0.20
X Data
Profile-Kd(490)
[m-1]
Spectral Kd can be well derived based on physics!
Challenges:
How Kd in the UVA/UVB varies globally?
3. Phytoplankton index
VGPM:
Chl became the index!!
(Behrenfeld and Falkowski, 1997)
Essence of Rrs-ratio derived Chl product:
 R rs (  1 ) 

Chl  fun 

R
(

)
 rs 2 
R rs  G
R rs ( 440 )
R rs ( 550 )
R rs ( 440 )
R rs ( 550 )


bb
a  bb
a ( 550 ) b bw ( 440 )  b bp ( 440 )
a ( 440 ) b bw ( 550 )  b bp ( 550 )
a ( 550 )
a ( 440 )
a w ( 550 )  a dg ( 550 )  a ph ( 550 ) Chl
*

a w ( 440 )  a dg ( 440 )  a ph ( 440 ) Chl
*
Simple ratio actually involves more than one variable!
(Szeto et al 2011, JGR)
Simple ratio dismissed spatial/temporal variation!
Analytically derived a(443) [m-1]
At the center of South Pacific Gyre
Rrs-ratio derived Chl [mg/m3]
Rrs-ratio derived Chl [mg/m3]
Nature of ratio-derived “Chl”
May 2009, Global, MODIS
Analytically derived a(443) [m-1]
Ratio-derived “Chl” is re-scaled total absorption coefficient!
4. Energy conversion
(Behrenfeld and Falkowski 1997)
(Platt et al, RSE, 2008)
Variation of phytoplankton- (or chlorophyll-) specific absorption
coefficient (a*ph) contributes largely to the variation of PBopt.
“Site-specific and previously published global
models of primary production both perform
poorly and account for less than 40% of the
variance in ʃPP,” (Siegel et al 2001)
“significant improvements in estimating
oceanic primary production will not be
our ability to predict temporal and spatial
variability in PBopt”. (Behrenfeld and Falkowski 1997)
B
PP eu  Popt  PAR  Chl  z eu  DL
Chl is NOT the direction to go.
Centered on Chl
Both PBopt and Chl have a*ph associated
Increase uncertainty in PP
PP eu    PAR  a ph  z eu  DL
Centered on absorption
No engagement of a*ph
PP estimation based on phytoplankton absorption (aph):
Quantum yield for photosynthesis
Remotely sensible
PP ( z )    ( t , z )  E 0 (  , t , z )  a ph (  , z ) d  dt
Ocean color
aph
Phytoplankton index
PP
(Marra et al 2007, Deep Sea Res.)
R2 = 0.84
R2 = 0.78
The Quasi-analytical algorithm (QAA)
Rrs()
(Lee et al. Appl. Opt., 2002)
u (  )  F1  Rrs (  ) 
η (± Δη) U2
a (  0 ) (   a ) U1
b bp (  0 )  F2 u (  0 ), a (  0 ), b bw (  0 ) 
 
bbp (  )  bbp (  0 )  0 
  

a (  )  F3 u (  ), b bp (  ), b bw (  ) 
U3
U4
a ph (  2 )  F4 ( a ( 1 ), a (  2 ),  (    ),  (    ))
a dg (  2 )  F5 ( a ( 1 ), a (  2 ),  (    ),  (    ))
u( ) 
bb (  )
a (  )  bb (  )
aph(λ) (m-1)
Measured vs sensed aph
(Lee et al 2004)
Absorption-based PP compared with measured PP
10
a p h -c e n te re d
C h l-c e n te re d
1 :1
3
1
0 .3
0 .3
1
3
10
m e a s u re d p ro d u c tio n ( m o l/l/d ay )
(Lee et al. 1996)
PPeu from model [mg C m-2 d-1]
calculated production ( m ol/l/day )
700
R2 = 0.44
1:1
600
500
R2 = 0.29
400
300
Chl-PPB
R2 = 0.74
200
Chl-PPA
Aph-PP
100
100
200
300
400
500
600
PPeu from on-deck incubation [mg C m-2 d-1]
(Lee et al. 2010)
700
Challenges:
Where is the global model for φ?
Which ‘ground truth’ we remote-sensors should aim at?
Summary:
1. A frame work for sensing primary production is well
established.
2. Optical/light related parameters can now be well