Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
Validation of ocean colour satellite products in coastal waters (HIGHROC project)
David Doxaran, LOV
& the HIGHROC consortium: RBINS (PI), LOV, NIVA, BC, VITO, CEFAS, U.Hull
amt4sentinelfrm Workshop, 20-21 June 2017, Plymouth, UK
Outline
1. Objectives
2. Satellite data and products
3. Methods (validation)
4. Test sites / In situ data
5. Results
6. Conclusions and Perspectives
Objectives
The HIGHROC (HIGH spatial and temporal Resolution Ocean Colour) project develops the next generation coastal water products and services from ocean colour space-borne data
Uncertainties associated to satellite products are quantifield at regional scales and provided to scientists and users
2014-2017
Satellite data and products
High Spatial resolution: S2plus
Landsat—OLI (30 m), Sentinel2-MSI (20 m)
Medium resolution: S3plus
MODIS, VIIRS, OLCI (250 - 1000 m)
High Temporal resolution: GEO
MSG-SEVIRI (15 mn, 4000 m)
L2R products: nLw, Rrs, rhow L2W products: SPM, TUR, Chla, Kd, IOPs
Methods (validation)
Regional algorithms AC (MUMM, NIR-SWIR, SWIR, SR) + L2W inversions
Routine processing at regional scales S2plus (VITO), S3plus (BC), GEO (RBINS)
Match-ups . Quality control of satellite products and field data . . Protocols adapted to test sites (1 to 3×3 pixels, +/-1h to +/-3h)
Quantification of uncertainties . As the differences in % between satellite products and field measurements (slope, NRMSE, MAPE)
Test sites / In situ data
Multi test sites (from clear to highly turbid waters) Field data from ferries, oceanographic cruises, autonomous fixed platforms
Results: atmospheric corrections (S2plus)
Match-ups between OLI-derived (SWIR AC) & Aeronet-OC MOW1 nLw (SNS)
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7
Sate
llite
Lw
n (m
W/c
m^2
sr u
m)
In-situ Lwn (mW/cm^2 sr um)
Lwn
Lwn_443
Lwn_483
Lwn_561
Lwn_655
Lwn_865
n = 57 R2 = 0.95RMSE = 0.5MAPE = 16.86%
0
1
2
3
4
5
6
7
0 1 2 3 4 5 6 7
Sate
llite
Lw
n (m
W/c
m^2
sr u
m)
In-situ Lwn (mW/cm^2 sr um)
Lwn
Lwn_443
Lwn_483
Lwn_561
Lwn_655
Lwn_865
n = 57 R2 = 0.9RMSE = 0.65MAPE = 41%
1×1 pixel +/-0.5h 3×3 pixels +/-0.5h
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.00 0.01 0.02 0.03 0.04 0.05 0.06
Sate
llite
Rrs
(sr-
1)
In-situ Rrs (sr-1)
Rrs
Rrs_443
Rrs_483
Rrs_561
Rrs_655
Rrs_865
n = 45R2 = 0.96RMSE = 0.0032MAPE = 14.35%
Results: atmospheric corrections (S2plus)
Match-ups between OLI-derived (SWIR AC) & in situ Rrs (Gironde estuary)
1×1 pixel +/-0.5h 3×3 pixels +/-0.5h
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.00 0.01 0.02 0.03 0.04 0.05 0.06
Sate
llite
Rrs
(sr-
1)
In-situ Rrs (sr-1)
Rrs
Rrs_443
Rrs_483
Rrs_561
Rrs_655
Rrs_865
n = 40R2 = 0.96RMSE = 0.0031MAPE = 13.13%
Results: atmospheric corrections (S3plus)
NRMSE on multi-spectral nLw and Rrs varying from 10 to 25%
SNS CPower
SNS MOW1
Results: atmospheric corrections (S3plus)
NRMSE on multi-spectral nLw and Rrs varying from 10 to 25%
Rhône Mesurho
Results: TUR and SPM retrievals (S2plus)
0
100
200
300
400
500
600
0 100 200 300 400 500 600
Sate
llite
SPM
(g.m
-3)
In-situ SPM (g.m-3)
SPM (Arles-Mesuhro relationship)
SPM_561
SPM_655
SPM_865
n(561) = 16R2 (561) = 0.19RMSE(561) = 133.17MAPE(561) = 50.02%
n(655) = 15R2 (655) = 0.66RMSE(655) = 111.72MAPE(655) = 40.51%
n(865) = 13R2 (865) = 0.98RMSE(865) = 18.38MAPE(865) = 43.73%
0
20
40
60
80
100
120
0 20 40 60 80 100 120
Sate
llite
SPM
(g.m
-3)
In-situ SPM (g.m-3)
SPM (Arles-Mesuhro relationship)
n(561) = 15R2 (561) = 0.44RMSE(561) = 21.93MAPE(561) = 47.13%
n(655) = 14R2 (655) = 0.7RMSE(655) = 14.69MAPE(655) = 37.96%
n(865) = 12R2 (865) = 0.65RMSE(865) = 19.11MAPE(865) = 47.32%
Match-ups between OLI-derived and in situ SPM (Rhône River mouth) using an adaptative (green > red > NIR) algorithm
1×1 pixel +/-0.5h 3×3 pixels +/-0.5h
Results: TUR and SPM retrievals (S2plus) Match-ups between S2plus- and Smartbuoy-derived TUR (SNS)
0
5
10
15
20
25
30
0 5 10 15 20 25 30
Sate
llite
T 6
55 (F
TU)
In-situ T (FTU)
T (FTU) 3x3
DOWSING
WEST GABBARD
WARP
LIVBAY
n = 18R2 = 0.92RMSE = 2.27MAPE = 61.98%
0
5
10
15
20
25
30
0 5 10 15 20 25 30
Sate
llite
T 8
65 (F
TU)
In-situ T (FTU)
T (FTU) 3x3
DOWSING
WEST GABBARD
WARP
LIVBAY
n = 17R2 = 0.49RMSE = 4.79MAPE = 170.21%
0
5
10
15
20
25
30
0 5 10 15 20 25 30
Sate
llite
T 6
55 (F
TU)
In-situ T (FTU)
T (FTU)
DOWSING
WEST GABBARD
WARP
LIVBAY
n = 25R2 = 0.83RMSE = 2.42MAPE = 77.88%
0
5
10
15
20
25
30
0 5 10 15 20 25 30
Sate
llite
T 8
65 (F
TU)
In-situ T (FTU)
T (FTU)
DOWSING
WEST GABBARD
WARP
LIVBAY
n = 22R2 = 0.25RMSE = 5.33MAPE = 197.82%
Results: Chla retrieval (S2plus)
Chla mapping at 20 m spatial resolution (S2-MSI) (Gernez et al. 2017) based on Gons et al. (2005) algorithm (aphy(675))
S2-MSI (18/08/2015)
Rrs(NIR) bb(NIR) aphy(665) Chla
Ongoing validation exercise…
Results: TUR retrieval (S3plus)
In the SNS, numerous TUR match-ups (here 1×1) with Smartbuoys data …some remaining calibration issues?
Results: Kd retrieval (S3plus)
In the SNS, numerous Kd match-ups (here 1×1) with Smartbuoys data …and satisfactory results
Results: TUR, SPM and Cha retrievals (S3plus)
Match-ups between S3plus satellite products (TUR, SPM, Chla) and shipborne measurements reveal typical differences (NRMSE) lower than 15%
SNS
Results: Chla retrieval (S3plus)
Results (GEO)
LR SEVIRI (SNS) HR MODIS
LR SEVIRI (SNS) LR MODIS (avg HR)
SEVIRI vs in situ Rhône River plume
Vanhellemont et al. (2013) Ody et al. (2016)
Conclusions / Perspectives
• Significant efforts made to develop multi-sensor OC algorithms to remote sense L2R (nLw, Rrs) and L2W (TUR, SPM, Chla, Kd) products at high spatial and temporal resolutions
• A large database with many quality match-ups was established to quantify the uncertainties associated to OC satellite products in coastal waters: 10-25% (NRMSE) for L2R and L2W
• Issues remain concerning the accurate remote sensing retrieval of Chla concentrations in turbid coastal waters
• Validation of high spatial (S2-MSI) remotely-sensed Chla concentrations in turbid waters
• Validation of VIIRS and OLCI (S3plus) satellite products
• High spatial and temporal remote sensing of OC products opens new perspectives for the monitoring of coastal waters
Selected publications • Novoa S., Doxaran D., Ody A., Vanhellemont Q., Lafon V., Lubac B. and P. Gernez (2017). Atmospheric
Corrections and Multi-Conditional Algorithm for Multi-Sensor Remote Sensing of Suspended Particulate Matter in Low-to-High Turbidity Levels Coastal Waters. Remote sensing, 9, 61.
• Gernez P., Doxaran D. and L. Barillé (2017). Shellfish aquaculture from space: potential of Sentinel2 to monitor tide-driven changes in turbidity, chlorophyll concentration and oyster physiological response at the scale of an oyster farm. Frontiers in Marine Science, in press.
• Baeye M., R. Quinn, S. Deleu, M. Fettweis, (2016). Detection of shipwrecks in ocean colour satellite imagery. Journal of Archaeological Science, 66, 1–6.
• Doxaran D. & Leymarie E. & Nechad B. & Dogliotti A.-I. & Ruddick K. & Gernez P. & Knaeps E. (2016). Improved correction methods for field measurements of particulate light backscattering in turbid waters. Optics Express, 24(4), 3615–3637.
• Kwiatkowska E. & Ruddick K. & Ramon D. & Vanhellemont Q. & Brockmann C. & Lebreton C. & Bonekamp H. (2016). Ocean colour opportunities from Meteosat Second and Third Generation geostationary platforms . Ocean Science, 12, 703–713.
• Ody A., Doxaran D., Vanhellemont Q., Nehad B., Novoa S., Many G., Bourrin F., Verney R., Pairaud I. et B. Gentili (2016). Potential of High Spatial and Temporal Ocean Color Satellite Data to Study the Dynamics of Suspended Particles in a Micro-Tidal River Plume. Remote Sens. 2016, 8(3), 245-279.
• Capuzzo, E., Stephens, D., Silva, T., Barry, J., & Forster, R. M. (2015). Decrease in water clarity of the southern and central North Sea during the 20th century. Global change biology, 21(6), 2206-2214.
• Vanhellemont Q. & Ruddick K. (2015). Advantages of high quality SWIR bands for ocean colour processing: Examples from Landsat-8. Remote Sensing of Environment, 161,89–106.
• Ruddick K., Brockmann C., De Keukelaere L., Doxaran D., Knaeps E., Forster R., Jaccard P., Lebreton C., Ledang A.-B., Nechad B., Norli M., Sorensen K., Stelzer K., Vanhellemont Q. & Van der Zande D. (2014). Processing and exploitation of Sentinel-2 data for coastal water applications: The HIGHROC Project . In: Proceedings of the Sentinel-2 for Science Workshop held in Frascati, Italy, 20-23 May 2014, ESA SP-726.
Extra slides
Warp
Extra slides
Warp
Extra slide: BOUSSOLE
ProVal An Argo float dedicated to the validation of ocean color data
CTD 2 sensors Ed-Lu Ed : 380, 412, 443, 490, 510, 560, 665 nm + PAR Lu : 380, 412, 443, 490, 510, 560, 665 nm Tilt and compass sensors Chla, backscattering, CDOM