Controls on Primary Productivity and its Measurement in Coastal

Report
Potential Improvements to Primary Productivity Estimates
through Subsurface Chlorophyll and Light Measurement
Michael Jacox
University of California, Santa Cruz
Raphael Kudela, Christopher Edwards (UCSC)
Mati Kahru, Daniel Rudnick (UCSD)
45th International Liége Colloquium
May 17, 2013
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Study Data: Primary Productivity and Ancillary Measurements
Shipboard:
1985-2011
CalCOFI Primary Productivity Casts
~100 cruises
>1500 stations
Satellite:
Data starting in 1997
SeaWiFS chlorophyll
SeaWiFS/MODIS PAR
AVHRR Pathfinder SST
Match-ups for 723 CalCOFI stations
Autonomous Profiling:
Data starting in 2005
Scripps Spray gliders
CTD, Fluorescence
Regular coverage of lines 80 and 90
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
The Roots of Satellite PP Models
Globally:
SC Bight:
22 cruises, ~270 stations (1974-1983)
Correlated NPP with surface environmental variables
Most of the variability explained is due to variability
in surface chlorophyll
Some explained by temperature and day length,
which may reflect seasonality
Shortcoming:
All information on vertical structure is lost
PP  1000
chl 0
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
28 Years Later, The Simplest Model is Often Among the Best
Rank
2
13
13
16
8
11
8
2
18
16
2
19
2
13
6
7
10
1
20
11
20
Friedrichs et al. 2009
PP  1000
chl 0
Saba et al. 2011
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
log (modeled productivity) (mg C m-2 d-1)
PP Model Performance for the CalCOFI dataset
VGPM
ESQRT
r2=0.55
Bias=0.11
RMSD=0.25
r2=0.64
Bias=0.20
RMSD=0.28
MARRA
VGPM-KI
r2=0.62
Bias=0.13
RMSD=0.25
r2=0.64
Bias=0.08
RMSD=0.24
log (in situ productivity) (mg C m-2 d-1)
ESQRT:
VGPM:
VGPM-KI:
MARRA:
Eppley square root model (Eppley et al. 1985)
Vertically Generalized Production Model (Behrenfeld and Falkowski 1997)
VGPM variant with two phytoplankton size classes (Kameda and Ishizaka 2005)
Vertically resolved model based on chl-specific absorption (Marra et al. 2003)
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Goal: Create an Improved PP Model for the Southern CCS
r2=0.33
r2=0.21
r2=0.00
r2=0.12
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Goal: Create an Improved PP Model for the Southern CCS
F = PP/chl
F = pp /chl
0
0
8000
0
0
100
80
6000
60
4000
40
2000
0
20
0
200
400
Distance from Shore (km)
8000
600
0
0
200
400
Distance from Shore (km)
600
100
80
6000
60
4000
40
2000
0
20
0
100
200
Year Day
300
0
0
100
200
Year Day
300
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Goal: Create an Improved PP Model for the Southern CCS
Start with VGPM:
PP  0.66125  Popt  d irr  chl 0 
B
PAR
PAR  4.1
PP
Popt ,CALC 
B
0.66125  d irr  chl 0 

Popt  f SST
B

 z eu
PAR
PAR  4.1
Popt  f chl 0 , dist

B

 z eu


Model Statistics for 2005-2010
Model
r2
Bias
RMSD
ESQRT
0.49
0.11
0.25
VGPM
0.59
0.20
0.29
VGPM-KI
0.57
0.12
0.24
MARRA
0.59
0.08
0.25
VGPM-SC
0.62
0.01
0.19
Behrenfeld and Falkowski 1997
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Revised Goal: Understand What Limits Model Performance
Fall 2002
Surface chlorophyll well correlated with chl at depth
Surface chlorophyll poorly correlated with chl at depth
Depth (m)
Depth (m)
Summer 2000
log(chlorophyll) (mg m-3)
log(chlorophyll) (mg m-3)
Model
r2
Bias
RMSD
Model
r2
Bias
RMSD
ESQRT
0.92
-0.06
0.13
ESQRT
0.41
0.16
0.23
VGPM
0.91
0.13
0.16
VGPM
0.44
0.20
0.26
VGPM-KI
0.94
0.10
0.17
VGPM-KI
0.39
0.14
0.22
MARRA
0.90
-0.05
0.15
MARRA
0.45
0.07
0.21
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
r2 (NPPMODEL, NPPIN SITU)
r2 (NPPMODEL, NPPIN SITU)
Revised Goal: Understand What Limits Model Performance
r2 (chl0, NPPIN SITU)
Model performance is strongly dependent
on chl0 being representative of NPP
r2 (PBOPT, MODEL, PBOPT, CALC)
…but not on accurate estimation of the
photosynthetic parameter
N = 14 years, 56 quarterly cruises
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Performance of a Simple Vertically Resolved Production Model
PAR
PAR  2.6
In situ surface chlorophyll
SeaWiFS chlorophyll

r2 (NPPMODEL, NPPIN SITU)
log (modeled productivity) (mg C m-2 d-1)
PP ( z )  2.9  d irr  chl ( z ) 
r2=0.59
Bias=0.02
RMSD=0.21
r2=0.64
Bias=0.03
RMSD=0.20
In situ chlorophyll
and light profiles
In situ chlorophyll profile
r2=0.74
Bias=0.02
r2 (chl0, NPPIN
RMSD=0.17
SITU)
log (in situ productivity) (mg C m-2 d-1)
r2=0.81
Bias=0.02
RMSD=0.14
Jacox et al., submitted
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Several CalCOFI Lines are Regularly Sampled by Spray Gliders
Pt. Arena
Monterey Bay
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Converting Glider Fluorescence to Chlorophyll
Los Angeles
Depth
Jul 2007
San Diego
Jan 2009
Depth
Correct profile
amplitude based on
surface chlorophyll
Lavigne et al. (2012)
3*fluorescence
Chlorophyll (mg m-3)
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Gliders Fluorescence Improves Productivity Estimates
Chlorophyll Data
Light Data
r2
RMSD
CalCOFI Surface
SeaWiFS Surface
0.52
0.20
Glider Profile
SeaWiFS Surface
0.59
0.19
CalCOFI Profile
SeaWiFS Surface
0.59
0.18
Potential for satellite/glider with fluorescence
CalCOFI Profile
CalCOFI Profile
0.74
0.14
Potential for satellite/glider with fluorescence and PAR
Potential for satellite alone
Data for CalCOFI/glider match-ups within 10km and 10 days (N=39)
100
% of Potential Improvement
% of Potential Improvement
100
r2
80
RMSD
60
40
20
0
0-10
10-20
20-30
30-40
Distance from Station (km)
40-50
r2
80
RMSD
60
40
20
0
-20
0-10
10-20
20-30
30-40
40-50
Time Difference (days)
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013
Conclusions
Satellite model performance in the SCCS is largely determined by correlations
between surface chlorophyll and NPP
Knowledge of in situ vertical chlorophyll and light profiles raises model
performance well above the variability between existing models
The combination of surface satellite data and subsurface profiler data is a
powerful one and a growing database of autonomous profiler data can now be
used to refine PP estimates
“In view of these prospects and challenges we urge our colleagues to
examine their own data on primary production and chlorophyll. There is
much yet to be done.”
-Eppley et al. 1985, J. Plankton Res.
M. Jacox | Potential Improvement to Primary Productivity Estimates through Subsurface Chlorophyll and Irradiance Measurement | May 17, 2013

similar documents