The colour of the Mediterranean Sea: global versus regional bio-optical algorithm evaluation
and development of regional chlorophyll dataset in the framework of MERSEA
Gianluca Volpe1, Rosalia Santoleri1, Salvatore Marullo2, Simone Colella1
1 Istituto di Scienze dell’Atmosfera e del Clima – sezione di Roma2 Ente per le Nuove tecnologie l’Energia e l’Ambiente – Centro Ricerche Frascati
[email protected]; http://gos.ifa.rm.cnr.it;
The aims are:
1. to identify and develop of an optimal algorithm for the Mediterranean sea to produce high quality ocean colour datasets for the region.
2. to provide an accurate and consistent stream of ocean colour data at a resolution and format compatible with operational forecasting of the Mediterranean Marine Environment.
3. To setup the Mediterranean Forecasting System (MFS) core products: satellite and model structure
NRT MODIS and the SeaWiFS reanalysis from 1997 to 2005
A new regional algorithm for Chlorophyll retrieval
Impact of OC regional product in models
In situ chlorophyll-a data: 1144 chlorophyll profiles for satellite data validation
Mediterranean Ocean Color CAL/VAL DATA SETS
28 Mediterranean cruises:from 1997 up to now were organized by ISAC in the framework of Italian National Projects
+ 2 permanent stations
Bio-optical measurements: 155 chl/opt measurements to define the Mediterranean regional algorithm(red points)
Optical measurements: 938 SIMBADA for Rrs validation(blue points)
CAL/VAL cruise in the Eastern Mediterranean
CNR-JRC joint cruise in Sep 2006
)( 44
33
221010 RbRbRbRbbC
MEDOC4 : R is log10 of either the 443/560 or the 490/555 or the 510/555 band reflectance ratios. The switch from one band ratio to another one is based on the chlorophyll concentration itself (the Maximum is Chosen)
THE NEW MEDITERRANEAN ALGORITHM (Volpe et al. RSE, 2006 in press)
a
b
AlgorithmsALL N = 155 OWP < 0.4 N = 105 OWP > 0.4 N= 50
r2 RPD APD r2 RPD APD r2 RPD APD
OC4 0.8 86.7 100 0.77 127, 129 0,62 -10,3 31.5
MedOC4 0.91 0.0 29.2 0.88 7.8 25.9 0.83 -12.3 34.5
In situ
Satellite
MED_OC4
Validation of MBRA SIMBADA-SeaWiFS RRS matchup dataset was build to test the accuracy of the satellite band ratios over the Mediterranean Sea.
SeaBAM N=1174Mediterranean N=156
Optical characteristics: Mediterranean vs Global
1. Low Chlorophyll: Med Band Ration < Global Band Ratio
2. Max Ratio choice similar for Med and Global
3. Is the Med Sea Greener or less Blue then the Global Ocean?
Is the Med Sea Greener or less Blue then the Global Ocean?
-blue (30%); +Green (15%)
-blue (35%); +Green (18%)
-blue (32%); Green similar
Med and Global tend to converge
+ blue (23%); - Green (35%)
MODIS & MERIS Algorithm Validation
MODIS MERIS
chl whole range N=156 r² RPD APD r² RPD APD
Standard Algorithm 0,78 89,3 104,6 0,76 108,1 118
Regional Algorithm 0,78 7,8 30,8 0,72 8,1 33,5
0.01<chl<0,4 N=110
Standard Algorithm 0,78 133,7 134,8 0,79 153,3 154
Regional Algorithm 0,72 11,6 28,9 0,77 13,5 31
MODIS vs SeaWiFS
regional
global
MERIS vs SeaWiFS
regional
global
N° RMS BIAS R2 RPD APD
SeaWiFS 92 0.47 -0.03 0.82 -0.94 8.4
MODIS 78 0.54 -0.2 0.75 -4.3 9.9
MERIS 81 0.65 -0.42 0.83 -8.5 11.3
MBR validation against in situ measurements:
Examined period: 2003-2004
The SeaWiFS reanalysis from 1997 to 2005: reprocessing of the entire Mediterranean Sea L1A archive using MedOC4
Algorithm
Available at the MERSEA web site
SeaWiFS Re-PROCESSING ON ESA-CNR GRID infrastructure
UI CE
SE
WN:Node01.. ….Node16
WN:Node01.. ….Node16
Gridtest03.esrin.esa.int Grid0007.esrin.esa.int
Se0.artov.rm.cnr.it
1) Globus-job-submit
2) Globus-job-status3) Globus-job-get-output
Globus-url-copy Globus-url-copy
UI=User Interface
SE=Storage element
CE=Computer Element
WN=Worker node
N° SeaWiFS pass= 6127 Total processing time
~ 4 days
Previous algorithm NEW algorithm Difference OLD-NEW
OC4.V4 Med OC4 (OC4.V4) – (Med OC4)0-0.1 0.2
0.01 50.01 5
2 July 2004
The GRID products output
OC4.V4 Med OC4
(OC4.V4) – (Med OC4)
1999 yearly average
Impact on Primary Productivity
The use of MedOC4 chlorophyll in PP Models reduces the annual PP estimate by about 40% and 10% respect to corresponding estimate made using OC4 and BRIC02
Global Model + Chl(OC4)
Global Model + Chl(BRIC02)
Global Model + Chl(MedOC4)
Other MED Optical characteristic
Ctot Csat Ze Ctot
Antoine and Morel, 1996 (yellow)
Morel and Berthon, 1989 (red)
Uitz et al., 2006 (green)
Colella 2006 (blue)
Morel and Berthon, 1989 (red)
Morel and Maritorena, 2001 (green)
Colella 2006 (blue)
Impact on Primary Productivity: global vs regional PP model
(a) Global Model + Chl(OC4)
(b) Global Model + Chl(BRIC02)
(c) Global Model + Chl(MedOC4)
(d) Regional Model + Chl(BRIC02)
(e) Regional Model + Chl(MedOC4)
MAX PP in SPRING
Monthly chlorophyll climatology maps derived from SeaWiFS data, processed by GOS-ISAC-CNR, for the year 1997-2004.
Monthly chlorophyll concentration maps obtained from OGS/OPA transport model after 10y spin-up.
Mediterranean biogeochemical coupled Mediterranean biogeochemical coupled model for future Short term Forecasting:model for future Short term Forecasting:
Model sensitivity to a longitudinal variable extinction coefficient
Crise et al, 2006
•Overall overestimation of surface
•Chlorophyll
•underestimates spring bloom
•in NW
•Relative error is below 0.8
•relative error lower in late-Autumn Winter
• error bias constant throughout the basin (positive anomaly in the W.Med in Spring)
Chl Anomaly Map
[Chl(data)-Chl(model)]/Chl(data))
Basin averaged relative anomaly
Concluding remarks• The analysis demonstrate that OC overestimation
chlorophyll in the Med is due to its peculiar optical characteristics.
• the use of regional algorithm is mandatory to produce high quality OC products for ecosystem model assessment and assimilation
• The MFS requirement for the GlobColour Project is to give access to the L2/L3 products (single sensors collocated Rrs data) from which we can produced specific regional L4 products taking into account the model community needs.
• Regional OC products should be produced in the TAC
The Mediterranean L4 products using L2P Medspiration data
Data mergingISACISAC
Optimal Interpolation
Data delivery on theGOS-ISAC web-site
Data quality controll
Night-time SST using MF algorithm
Cloud detection
SST daily composite binning on model grid
(1/16x1/16)
MF AVHRR acquisition
Atlantic buffer zone + west Med
Night-time SST using Pathfinder algorithm
Cloud detection
SST daily composite binning on model grid
(1/16x1/16)
ISAC AVHRR acquisition
Entire Mediterranean
L2P MedspirationProducts
To save the Mediterranean Sea from environmental problemlook at the combination of
satellites and ocean forecast!!
Conclusions