Moorings: New Results and New Technology Overview

1
Surface Ocean CO2 Variability and Vulnerabilities
Workshop
Moorings New Results and New Technology
Overview Overview of Proxy Techniques for Data
Extrapolation and Interpolation
By Christopher Sabine (NOAA/PMEL)
With contributions from N. Bates, R. Byrne, C.
Casca, Aymeric Chazottes, M. DeGrandpre, J. Dore
, D. Sadler, R. Feely, G. Friederich, A.
Jacobson, K. Lee, N. Lefevre, K. Johnson, S.
Maenner, L. Merlivat, S. Musielewicz, F. Sayles,
U. Shuster, T. Takahashi, Maciej Telszewski, J.
Trinanes, R. Wanninkhof
Paris, France April 11-14, 2007
2
We Have Made Great Progress with underway pCO2
Measurements
Circa 2004
A number of countries are in the process of
instrumenting ships of opportunity with automated
pCO2 systems
3
We Have Made Great Progress with underway pCO2
Measurements
Circa 2006
A number of countries are in the process of
instrumenting ships of opportunity with automated
pCO2 systems
4
Yet we are still struggling to produce a global
climatology with sufficient data coverage after
35 years
From Takahashi et al. manuscript in preparation
based on a data set of approximately 2.8 million
data points collected over 35 years
Number of months

• Two Complicating Factors
• Complex Chemistry
• Variable in Time and Space

Net Flux (moles CO2 m-2 yr-1)
5
Challenge Complex Chemistry
Atmosphere
CO2
K1
K2
CO2aq (CO2 H2O)
HCO3 H
CO32 H
Productivity
CaCO3
Ca2
C106H175O42N16P
calcification
Particulate Organic C
?a Ca2CO32- Ksp ?a gt 1
Supersaturated
Decomposition to organic and inorganic C
CaCO3 CO2 H2O
dissolution
Ca2 2HCO3-
6
Challenge Carbon System is Variable in Time and
Space
N. Gruber and H. Brix (2004)
The subtropical gyres are thought to be some of
the most stable surface ocean regions, but the
time-series sites have identified significant
variability in carbon parameters on all time
scales.
7
To help address these challenges we need to
supplement the standard shipboard discrete
measurements with more autonomous instrumentation
Today we will focus on moorings but the same
principle applies to other autonomous technologies
8
The most common mooring parameter is pCO2
There are two basic system types 1) colorimetric
Carioca System ala Liliane Merlivat
SAMICO2 System ala Mike DeGrandpre
9
CARIOCA
New CO2 time-series at the PIRATA mooring at 6S,
10W Nathalie Lefèvre
Liliane Merlivat has made many deployments of
CARIOCA drifter, but it has also been tied to
some moorings (e.g. DYFAMED, BTM)
Argos antenna
Electronics
CO2 sensor (Carioca) Aanderaa Optode
June 2006-present
10
The most common mooring parameter is pCO2
There are two basic system types 1) colorimetric
Carioca System ala Liliane Merlivat
SAMICO2 System ala Mike DeGrandpre
11
Red stars show historical deployments of SAMICO2
over the last 10 years
12
The most common mooring parameter is pCO2
There are two basic system types 2) NDIR
Air intake and filter holder
Calibrated gas standard
Electronics Package
Equilibrator
Battery Compartment 138 D-cells for 400 days
MAPCO2 System ala Christopher Sabine (originally
based on MBARI design)
MBARI Mooring ala Friederich and Chavez
13
The most common mooring parameter is pCO2
There are two basic system types 2) NDIR
MBARI Monterey Bay Mooring
From G. Friederich
Since strong upwelling favorable winds and high
surface CO2 levels coincide, much of the ocean to
atmosphere transfer of CO2 occurs during short
time intervals.
14
The most common mooring parameter is pCO2
There are two basic system types 2) NDIR
Air intake and filter holder
Calibrated gas standard
Electronics Package
Equilibrator
Battery Compartment 138 D-cells for 400 days
MAPCO2 System ala Christopher Sabine (originally
based on MBARI design)
MBARI Surface Drifter ala Friederich and Chavez
15
Red stars show all current deployments of MAPCO2
Blue stars show current deployments of MBARI CO2
systems
Red PMEL
Blue MBARI
NOAAs strategy is to outfit OceanSITES flux
reference stations and 25 NDBC US coastal buoys
16
Challenge Complex Chemistry
pCO2 alone is not enough to understand the
underlying mechanisms controlling carbon changes
or how these changes may be affecting the
ecosystem
After Turley et al., 2005
A second parameter is needed for autonomous
instruments to look at the full range of carbon
speciation
17
Developing A Second Moored C Parameter
YSI and others have electrode based systems
DeGrandpre (Univ. Montana) is developing a
SAMI-pH system for commercial production
Testing the SAMI-pH system at Scripps pier
18
Developing A Second Moored C Parameter
CO2 System Measurements Using a Teflon AF
Waveguide
Calculated Parameter
System Configuration
acidified seawater
CTf(AT,pH)
AT std indicator
acidified seawater
acidified CT standard
seawater indicator
AT f(CT,pH)
acidified CT standard
LCW
seawater indicator
pHT
seawater
seawater indicator
seawater
pCO2 f(AT,pH)
AT std indicator
seawater
From B. Byrne, USF
19
Developing A Second Moored C Parameter
Cabled Profiler
Bob Byrne (USF) is developing Spectrophotometric
Elemental Analysis Systems (SEAS) that has
potential applications on moorings
AUV
Power 12W at 8-24 V DC Size 75 cm x 24
cm Weight 25 kg (w/ batteries) Comm RS-232
Ethernet Depth range 2800 meters Multiple
reagent capability
Autonomous Profiler
SEAS
From B. Byrne, USF
20
Developing A Second Moored C Parameter
Kimoto Electric has developed flow-through
analyzers for alkalinity and total CO2
Flow-through Alkalinity
Flow-through TCO2
Response Time 1 minute Lab. Precision 2 µmol/kg
Response Time 1 minute Lab. Precision 2 µmol/kg
A small version of the alkalinity system is being
modified to work on a mooring
Kimoto Electric Co. Ltd., Osaka, Japan
(E-mailhkimoto_at_Kimoto-electric.co.jp)
21
Developing A Second Moored C Parameter
Fred Sayles (WHOI) is developing an automated pH
an TCO2 system called RATS
Measure TCO2 Measure pH Precision
pH .003 TCO2 5 µmol/kg Endurance
6 months/ 700 Analyses ( with no
external batteries) In Situ Standardization
22
And dont forget the other carbon relevant
parameters

• NO3- ISUS
• O2 Aanderaa Optode
• fluorometer
• bio-optical instruments
• etc.

Recall K. Johnsons poster Wednesday evening
23
Challenge Carbon System is Variable in Time and
Space
Feely and Wanninkhof have been making shipboard
fCO2 measurements across the Equatorial Pacific
approximately twice per year since 1995
El Niño 0.2-0.4 Pg C yr-1 Non El Niño 0.7-0.9
Pg C yr-1 La Niña 0.8-1.0 Pg C yr-1 Average 0.6
0.2 Pg C yr-1
From C. Cosca
Big changes are correlated with ENSO variability
24
How well do ships characterize temporal
variability?
Cosca et al., (2003)
25
Moorings can capture variability missed between
ship visits
TAO MAPCO2 System (140W, 0)
pCO2 at SST (µatm)
Time (date)
26
Seasonal cycle in pCO2 is relatively smallonly
about half of what one would expect from the
magnitude of the seasonal temperature signal
TAO MAPCO2 System (140W, 0)
pCO2 at SST (µatm)
Time (date)
Long-term variations (annual rate) 1.7-2.5
µatm ENSO variability 80-100 µatm Seasonal
variability 20-30 µatm Sub-seasonal
variability 50-60 µatm Diurnal variability
20-40 µatm
27
Spectral analysis indicates that sub-seasonal
variability is dominated by two frequencies
55d
17d
35d
Tropical Instability Waves with frequencies of
17-35 days
Kelvin Waves with frequencies of 53-60 days
Note that Kelvin waves show the most energy in
SST, but the TIWs show more energy in fCO2
28
Moorings can capture variability missed between
ship visits
variability is missed even at the monthly time
series sites
HALE-ALOHA Mooring (158W, 23N)
pCO2 at SST (µatm)
Time (date)
There is a seasonal cycle of low values in the
spring and slightly above atmospheric in the
fall, but there is a lot of variability around
that seasonal pattern
29
Moorings can capture variability missed between
ship visits
HALE-ALOHA Mooring (158W, 23N)
pCO2 at SST (µatm)
bloom event
Time (date)
Interannual variability 10-20(?)
µatm Seasonal variability 20-30
µatm Sub-seasonal variability 10-20
µatm Diurnal variability 2-10 µatm
30
Moorings can capture variability missed between
ship visits
Bermuda Testbed Mooring (64W, 32N)
pCO2 at SST (µatm)
Time (date)
Interannual variability 10-20(?)
µatm Seasonal variability 80-100
µatm Sub-seasonal variability 10-20
µatm Diurnal variability 2-5 µatm
31
Moorings can capture variability missed between
ship visits
Bermuda Testbed Mooring (64W, 32N)
pCO2 at SST (µatm)
Hurricane Florence (cat. 2) passes over mooring
Time (date)
32
The DYFAMED project also identified substantial
variability in the Mediterranean that were not
captured by the monthly time series visits.
High frequency variability has been identified
everywhere we have looked, so where does this
leave us?
From Copin-Montegut et al. (2004)
33
Relating Underway Data to Moored Data
Average difference between BTM and AE for
measurements within 10 km and 3 minutes is less
than 0.5 4.7 µatm (n15,462)
By comparing data over a range of distances one
can begin to assess the correlation length scales
for the region
From N. Bates
34
Relating Underway Data to Moored Data
From N. Bates
Preliminary analysis at BTM suggests that data
are coherent within about 80 km regardless of
season
35
The addition of automated sensors will help us to
fill in some of the temporal and spatial holes in
the surface CO2 network, but is it sufficient to
generate something more than an improved CO2 flux
climatology?
Moorings
If our goal is to generate annual or seasonal
air-sea flux maps we will likely have to resort
to proxy techniques
36
Researchers have been using proxies to
extrapolate surface CO2 data in time and space
for decades but these approches have generally
been limited to small regions
I am aware of at least 4 unique efforts currently
working to develop large-scale ocean CO2 flux
maps at annual or better resolutions
1.) Several groups are working on regional maps
based on empirical relationships (e.g. MLR,
neural network) derived from observations 2.)
NOAA ocean C group is working on monthly global
maps based on Kitack Lees empirical fit of the
Takahashi climatology against SST. 3.) NOAA
atmospheric C group has developed weekly global
CO2 flux (including both land and ocean)
estimates from inversions of atmospheric CO2.
Ocean fluxes are based on Takahashi et al.
(2002) 4.) Corinne Le Quéré has posted air-sea
CO2 fluxes calculated using an Ocean General
Circulation Model as part of a Global Carbon
Program (GCP) effort to evaluate the global
carbon budget.
37
Monthly pCO2 maps for 2005-2006 from UEA MLR
analyses
Maps by Ute Schuster and Andy Watson
Recall talks by Feely, Nojiri and Schuster
38
Monthly pCO2 maps for 2005-2006 from UEA neural
network analyses
Maps by Aymeric Chazottes, Maciej Telszewski and
Cyril Moulin
Cyril will discuss this in the next talk
39
NOAAs Seasonal CO2 Flux Map Program
Algorithm development pCO2 f(SST, color)
pCO2 maps
Apply algorithm to regional SST color fields to
obtain seasonal pCO2maps
Co-located satellite data
In situ sampling pCO2, SST, SSS
Flux k s ?pCO2
Regional satellite SST color data
Flux maps
Algorithm development Gas transfer, k f
(U10,SST)
Wind data
Remote sensing SST, color wind Soon SSS
40
Generating a Global Flux Map
Based on approach of Lee et al. (1998) and Park
et al. (2006)

• Use monthly pCO2sw climatology of Takahashi et
  al. 2002 as baseline
• 2. Assess interannual variability using
  climatology and local relationships of pCO2sw
  with SST, along with SST and winds from
  satellites (proxy approach)

Fy,mky,mKopCO2sw95(dpCO2sw/
dSST)95x?SSTy,m-95-pCO2air95
Use satellite derived SST anomalies and winds to
determine seasonal fluxes
41
Relationship between SST and pCO2SW ?pCO2SW/?SST
Park et al. 2006
Seasonal (dpCO2sw/ dSST)95 relationships were
determined for each 4x5 box using a least
squares fit of Takahashis monthly climatology
everywhere except equatorial Pacific (used
Feelys functions).
42
Generating a Global Flux Map
This approach suggests a relatively small
interannual variability but it will tend to
smooth over extreme events and does not fully
capture biological effects
20-year global mean air-sea CO2 flux 1.70
0.18Pg C yr-1
For equatorial pacific 0.11Pg C yr-1
43
NOAAs Seasonal CO2 Flux Map Program
Monthly flux maps from 1995-2005
Based on NCEP II winds and SST
Ongoing work
1. Improving regional relationships by
incorporating additional parameters (e.g. mixed
layer depth, chlorophyll) 2. Improving regional
relationships using ship-based and moored pCO2sw
observations
44
CarbonTracker
Global carbon flux estimates based on atmospheric
inversions
global 6x4 resolution
North America 3x2
United States1x1
time step of three hours
vertical resolution is 25 hybrid sigma-pressure
levels
Recall poster by A. Jacobson Wednesday evening
45
CarbonTracker
Ocean fluxes based on Takahashi et al. 2002
?pCO2 values upsampled from the native 4x5
resolution to a 1x1 grid.
Gas exchange is computed every 3 hours using the
European Centre for Medium-Range Weather
Forecasts (ECMWF) forecast meteorology of the
atmospheric transport model.
Maps showing weekly estimates are available for
2000-2005
http//www.esrl.noaa.gov/gmd/ccgg/carbontracker/
46
CarbonTracker
Inversion approach can provide information on
air-sea fluxes that is independent of ocean based
approaches
For example, inversion can not reconcile the
South Pole atmospheric CO2 measurements with any
reasonable ocean fluxes. Also, flux estimates are
not dependant on choice of gas exchange
parameterization.
47
GCP Global Carbon Budget
The monthly air-sea CO2 flux is calculated for
the time period 1955-2004 using an Ocean General
Circulation Model that is coupled on-line to a
biogeochemical model.
Physical model We use the OPA physical model, a
fully prognostic OGCM based on primitive
equations (Madec et al., 1999). The model has a
horizontal resolution of 2 latitudinal
resolution is enhanced to 0.5at the equator and
at the poles. OPA is coupled to a thermodynamical
and prognostic sea-ice model (LIM). Biogeochemical
model We use the PISCES-T model, which
represents 4 Plankton Functional Types
(Nanophytoplankton, Diatoms, meso and micro
zooplankton) and co-limitation by P, Si, Fe and
light. Forcing The model was forced with NCEP
daily winds and fluxes. Initialization
Biogeochemical fields are initialized with
observations from the World Ocean Atlas. The
oceanic CO2 sink is 2.17 PgC/yr averaged over
1991-2002.
Flux maps are not available on-line, but global
monthly uptake estimates are available
at http//www.bgc-jena.mpg.de/corinne.lequere/in
terannual/
Recall talks by Lequéré, Mckinley and Doney
48
Presenting an Ocean Carbon Budget
Our community is facing increasing pressure to
produce global ocean flux maps resolved at weekly
to monthly temporal resolutions. The estimates
presented here vary by nearly 0.5 Pg C in their
estimated flux. This raises a number of questions
that we need to consider.
How well do these maps represent reality? Are
these approaches limited by ocean CO2
observations or by the availability of high
resolution global products? Can these proxy
techniques account for the changing ocean pCO2
system with time? What message are we giving the
public by presenting these maps?
49
Recap of main points for discussion
Buoys, Buoys, Buoys 1. Buoys goodShips bad!!
50
Recap of main points for discussion

• Underway systems provide an essential spatial
  component for understanding surface pCO2, but
  moorings and other autonomous in situ
  instrumentation can provide the temporal
  component which can be equally challenging.

2. A secondary carbon parameter is needed to
constrain the full carbon chemistry and the
controls on surface pCO2. Corollary other
physical and biological measurements are needed
on all systems to help evaluate pCO2 controls.
3. Even with an improved observing system we will
likely need to continue to develop proxy
approaches for making seasonal to annual large
scale flux maps. (further support for items 1 2
above)
4. There are several groups generating
multi-year, large scale flux mapsis it time to
start comparing our results and developing
strategies for advancing these efforts?