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