George Petihakis, Athanasia Iona,

1
Development and Implementation of a Data
Integration System for Eutrophication Assessment
in Coastal Watersin the framework of INSEA
project CONTRACT Nº SST4-CT-2005-012336

• George Petihakis, Athanasia Iona,
• Gerassimos Korres, Pelopidas Karagevrekis,
• Dionisis Raitsos, and
• George Triantafyllou
• IMDIS-2008, Athens

2
Agenda

• Description of the project
• Pagasitikos eutrophication problem
• Description of the Ecosystem Model
• Data Assimilation
• Data Management System

3
Overall objective

• Building an integrated tool for coastal ecosystem
  assessment combining Models, Satellite Remote
  Sensing and in situ measurements.

4
Strategy

• Build a consortium with the following
  tools/skills
• Models developed and operational,
• In situ Real-time measurement experience,
• Remote Sensing experience,
• Good data management capacities.
• Capacity for assessing the problem at a European
  scale.

5
Consortium
6
Data Flow
7
Case Studies
8
The eutrophication problem
9
Field data
10
Computational area
11
Ecosystem model
ERSEM
12
Assimilation into POM-ERSEM

• Model 1/100o Pagasitikos configuration with 25
  layers
• Downscaling from 1/30o Aegean, from 1/12o
  Mediterranean

Downscaling model solutions a problem of proper
OBCs (nesting) and initialization
Init
Init
Basin scale model (MED OPA 1/12)
Regional model (AEGEAN POM 1/30)
Coastal model (PAG POM 1/100)
OBCs
OBCs

• Interpolate coarse model solution and add
  constraints in order to dump spurious
  oscillations and propagating waves
• VIFOP (a variational
  initialization method, Auclair 1999)
• Nesting techniques developed within MFSPP, MFSTEP
  and POSEIDON projects

13
Forcing Initial conditions

• ECMWF 6 hours data
• Wind stresses
• Temperature and air humidity
• Cloud cover
• Heat fluxes
• Solar radiation
• Precipitation Jaeger (horizontal resolution
  5?x2.5o)
• Initial conditions March analyzed fields

14
Model results vs Obs
July Mean Circulation
15
Model results
16
Cost Functions
17
The Low-Rank Kalman Filter(s)

• Singular Evolutive Kalman (SEEK) filters,
  low-rank (r) square-root KFs

Analysis
Forecast

• A collection of SEEK filters

• SEEK

• SEIK Ensemble variant

• SFEK

18
Data Assimilation
19
Data Assimilation
20
(No Transcript)
21
(No Transcript)
22
Assimilation Experiments
TWIN EXPERIMENTS
FILTER ? rank OBSERVATIONS TYPE STATE VECTOR
SFEK 0.65 35 T,S PROFILES FULL
SEEK 0.65 35 T,S PROFILES FULL
SEIK 0.65 35 T,S PROFILES FULL
23
Twin Experiments
SFEK, SEEK AND SEIK PERFORMANCE(35 modes, ?0.65)
24
Assimilation into POM-ERSEM

• Observations SeaWiFS 2003 on a weekly basis
• SEEK Filter with rank 40
• Initialization EOFs computed from 2-days
  outputs of 2001-2002 model integration,
  explaining 95 of the variance
• Amplification factor 0.5

25
Experiment Setup

• Model 1/100o Pagasitikos Gulf with 25 layers

2003
31/12/03
1996
2001
Spin up
EOFs
REF OBS
2 years
1 year
5 years

• Initialization starts from the state of
  31/12/2002
• Free-run run without assimilation starting from
  mean state
• Evaluation RMS misfit relative to the misfit
  from mean state

26
(No Transcript)
27
Assimilation into ERSEM
28
Hindcast Experiments 2003
29
Hindcast Experiments 2003
30
EXP4 Mean values of Chl and BIAS (model
Forecast and Analysis)
EXP4 SEEK, 7 mod, FF0.3, Er0.1
31
Data for HNODC

• Hydrographic, Chemical and Biological data
• All the Pagasitikos gulf area, and
• Period April 1998 November 1999
• Parameters measured
• Temperature, Salinity, Density, Currents,
• Oxygen, Nitrate, Nitrite, Ammonium, Phosphate,
• Dissolved Organic Phosphate, Silicate,
• Chlorophyll-a, Bacteria, Phytoplankton, and
  Zooplankton.

32
INSEA Common Data Index Metadata

• A total of 970 xml records from 23 cruises
  transmitted by HNODC to central INSEA CDI
  database (http//www.insea.info/cdi/)
• 413 xml records describing ADCP data
• 413 xml records describing CTD data
• 144 xml records describing Bottle data
• Validation scheme CDI_V1_04.xsd
• Produced by MIKADO Java Tool

33
CONCLUSIONS

• A management tool for the ecosystem of
  Pagasitikos gulf has been developed and
  implemented successfully
• It consists of
• Hydrodynamic model
• Biogeochemical model
• Assimilation scheme
• Satellite data
• Model forecast is usually becoming different of
  the analysis which is an other indication of the
  systems strong variability requiring frequent
  observations
• Hindcast experiments show that the performance of
  the system requires
• Good quality of satellite observations
• The model exhibits biases since it produces the
  spring bloom a week before in relation to the
  satellite observations

34
Thank You