Pr

1
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour Antoine Mangin1, Seppo Kaitala2,
Stéphane Maritorena3, Odile Fanton dAndon1, and
Philippe Garnesson1 (1) ACRI-ST , (2) FIMR, (3)
ICESS UCSB Session Observations May 22, 2008
2
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Outline of this presentation

• Background of Marcoast
• Background of GlobColour
• Validation over Baltic

Gulf of Finland
3
Background, Marcoast Project

• MarCoast Project Presentation SERV10, 21 May by
  Planetek
• GMES Service Element project (ESA Funding)
• The main goals are
• the surveillance and monitoring of the marine
  and coastal environment
• to deliver services at a European scale.
• integration and Sustainability
• was launched in 2005 and will be completed
• at the end of 2008.

4
Background, MarCoast Project

• ACRI-ST activities within Marcoast project
• Ocean Colour Delivery Service (NRT/Archive
  MERIS/MODIS data) to other service providers
• Pan-European service of eutrophication statistics
  information
• Water quality indicator for EEA see EuroGOOS
  poster P19
• Validation activities
• Regional consistency assessment with in situ
  data
• over Baltic sea in collaboration with FIMR
• More on Marcoast http//gmes-marcoast.com/

5
Background, Globcolour project 2005-2008

• GlobColour ESA DUE project i.e. driven by end
  users IOCCG, IOCCP,UK Met-Office,
• Achievements
• Provision of a long time-series (1997-2007) of
  consistently calibrated and merged global ocean
  colour products
• MERIS (ESA), SeaWiFS (NASA), MODIS-AQUA (NASA)
• Validation and characterisation performed with
  available in situ observations
• Future development Global Ocean Colour Thematic
  Assembly Centre of the future EU GMES Marine Core
  Service (MyOcean)
• More on Globcolour www.globcolour.info
• ( Data available on line )

6
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Validation with FIMR observations
7
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Towards the turbidity validation . through the
In situ monitoring route Helsinki - Travemûnde

Turbidity is routinely measured and recorded
during cruises
The Nephelometric Turbidity Unit (NTU) that is
used is directly comparable to the scattering of
particulate matter bp
Helsinki
Travemûnde
8
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Illustration of turbidity mapping

Helsinki
Issues High dissolved organic matter (CDOM)
and Turbidity High variability
Åland
2006
2007
Travemûnde
Well identified seasonal increase of turbidity
Spatial and temporal uniformity of turbidity
9
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Chl-a - Gulf of Finland In situ obs. Source
FIMR
Issues High dissolved organic matter (CDOM)
and Turbidity High variability
Spring bloom
2005
10
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Chl-a - Gulf of Finland MODIS/SeaWiFS standard
algorithms 1997-2006
Issues High dissolved organic matter (CDOM)
and Turbidity High variability
False summer bloom of Chl
Spring bloom is not captured
11
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Chl-a validation - Gulf of Finland in situ /
MERIS case 2
The Meris-Chla algorithm for Case 2 waters gives
quite good fit with in situ data for the Gulf of
Finland But Spatial and temporal coverage
issue
12
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
GlobColour Data merging

• Why do ocean color data merging ?
• Several simultaneous global ocean color missions
• Several versions of the same product
• Benefits
• Development of unified, consistent ocean color
  time-series from multiple sensors
• Improved spatial and temporal coverage
• More diverse ocean color products with lower
  uncertainties
• GlobColour
• Access to products uncertainties
• Exploitation of all spectral bands from 412 nm
  to 670 nm.

13
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
GlobColour GSM algorithm
MERIS, MODIS, SeaWiFS LwN Estimates of the
uncertainties on input LwN (from sensors obs.
characterisation)
Estimates of the model error (from NOMAD Database)
Estimates of Chla, bbp, cdm uncertainties on
outputs (Co-variance matrix)
14
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
GlobColour Error estimates - Validation
Inputs GC products Results very close to
expectancy a small bias is detected the error
estimates by GSM (with ad hoc inputs) is slightly
underestimated.
Inputs In situ observations (Nomad) Results
very close to expectancy no significant bias
15
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
GlobColour Example
16
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
GlobColour Example
August 7, 2007 bbp
bbp Error
17
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
GlobColour Example
March 26, 2007 bbp
bbp Error
18
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic elements of validation
Some empirical laws are proposed between particle
backscattering coefficient (bbp) and particle
scattering coefficient (bp), e.g., bbp0.015 bp
(in MERIS ATBD)
In other words it is there admitted that it
exists a linear relationship between the two
coefficients.
However, it should vary with size and nature of
particulate matters, and consequently could vary
with time and space.
The challenging objective here is therefore to
make the bridge between measured bp (through NTU)
and EO-derived bbp in order to bring novel
insights about the linearity of the relationship
and wherever possible its suitability with
respect to turbidity ranges.
19
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic elements of validation
Helsinki
EO side
Extraction of a 3x3 GC macro-pixel each 10 km for
available in situ truth
Travemûnde
In situ side
2006
Filtering Bplt1,5 removal of  obvious 
artefacts
NTU
Month
20
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic elements of validation
2006
21
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic Confrontation EO (bbp)
against in situ (bp)
GSM - GlobColour (EO) bbp vs In-situ bp
22
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic Confrontation EO (bbp)
against in situ (bp)
Monthly scale Correlation exists Still high
(spatial) variability
GSM - GlobColour (EO) bbp vs In-situ bp
23
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic Confrontation EO (bbp)
against in situ (bp)
to be compared to 0.015
September 2006 Bbp0.0106 bp 0.0005 R20.39 N10
12
Bbp0.0109 bp 0.01 R20.15
.. and other sources for validation (Babin,
Maritorena, personal com.)
GSM - GlobColour (EO) bbp vs In-situ bp
GSM (in situ) vs In-situ
24
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
Turbidity in Baltic Confrontation EO (bbp)
against in situ (bp)

• First conclusions
• A simple linear relationship between bp and bbp
  cannot blindly be used as is in the Baltic sea.
• On a monthly basis, a linear relationship between
  bbp and bp can be used to cross-check  observed
  bp  against  EO-derived bbp 
• Still some work to sort out  valid  EO-derived
  information (e.g. by using exclusion threshold on
  standard deviation) and  valid  observation
  (e.g. by checking spatial and temporal
  consistencies).
• Work will be expanded to previous observations
  (2003/2004/2005)

25
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
FIMR feedback on the service The Meris-Chla
algorithm for Case 2 waters and by extension GSM
method gives quite good fit with in situ data for
the Gulf of Finland ( quite problematic due to
high variation of water quality parameters).
26
Validation of turbidity products in the Baltic
derived from multi-sensors observations of ocean
colour
FIMR feedback on the service The service
gives valuable information for the water quality
evaluation of the Baltic Sea and reporting to
HELCOM about the water quality status of the
Baltic Sea. 
http//www.helcom.fi/environment2/ifs/ifs2007/secc
hi/en_GB/secchi/ http//www.helcom.fi/environment2
/ifs/ifs2007/en_GB/springbloom/