User Uptake Workshop 16-17 November 2017 European Environment Agency, Copenhagen, Denmark Dimitry Van der Zande - RBINS Clémence Goyens – VITO Quinten Vanhellemont – RBINS Liesbeth De Keukelaere - VITO Introduction to iCOR and ACOLITE atmospheric correction for Landsat-8 and Sentinel-2
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Dimitry Van der Zande - RBINS Clémence Goyens – VITO
Quinten Vanhellemont – RBINS Liesbeth De Keukelaere - VITO
Introduction to iCOR and ACOLITE atmospheric correction for Landsat-8
and Sentinel-2
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ATMOSPHERIC CORRECTION
Ozone layer
Absorption by aerosols
& molecules
Scattering by aerosols & molecules
Scattering from the
target
Rayleigh
Aerosols
θ𝑠𝑠 θ𝑣𝑣
Scattering from the
background
Atmosphere
Δφ
𝐿𝐿 𝑇𝑇𝑇𝑇𝑇𝑇 = 𝐿𝐿𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝐿𝐿𝑡𝑡𝑡𝑡𝑎𝑎−𝑝𝑝𝑡𝑡𝑡𝑡𝑝 + 𝐿𝐿𝑏𝑏𝑡𝑡𝑏𝑏𝑏𝑏𝑡𝑡𝑡𝑡−𝑝𝑝𝑡𝑡𝑡𝑡𝑝
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
AC in Ocean Color
Courtesy of PJ Werdell
𝐿𝐿 𝑇𝑇𝑇𝑇𝑇𝑇 = 𝐿𝐿𝑤𝑤𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝐿𝐿𝑡𝑡𝑡𝑡𝑎𝑎−𝑝𝑝𝑡𝑡𝑡𝑡𝑝 + 𝐿𝐿𝑏𝑏𝑡𝑡𝑏𝑏𝑏𝑏𝑡𝑡𝑡𝑡−𝑝𝑝𝑡𝑡𝑡𝑡𝑝 In the NIR
Gordon and Wang, 1994 Black Pixel Assumption
Ignored
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
AC in Ocean Color
Courtesy of PJ Werdell
𝐿𝐿 𝑇𝑇𝑇𝑇𝑇𝑇 = 𝐿𝐿𝑤𝑤𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝐿𝐿𝑡𝑡𝑡𝑡𝑎𝑎−𝑝𝑝𝑡𝑡𝑡𝑡𝑝 + 𝐿𝐿𝑏𝑏𝑡𝑡𝑏𝑏𝑏𝑏𝑡𝑡𝑡𝑡−𝑝𝑝𝑡𝑡𝑡𝑡𝑝
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ATMOSPHERIC IMAGE CORRECTION
- Use a NIR modelling scheme and remove ρW(λNIR) from ρt(λ) → apply GW94 e.g., Siegel et al. (2000), Stumpf et al. (2003), Bailey et al. (2010), Wang et al. (2012) - Fix aerosol contribution from nearby pixels and model to determine the ρW(λNIR) e.g., Hu et al. (1999) and Ruddick et al. (2000, 2006)
- Using Short Wave Infrared (SWIR) bands in highly turbid waters e.g., Wang and Shi (2007) and Wang et al. (2009)
- Coupled Ocean-Atmosphere optimization/inversion e.g., Gordon et al. (1997) (spectral matching); Chomko and Gordon (1998), Chomko et al.
(2003) and Kuchinke et al. (2009) (spectral optimization method); Jamet et al. (2005) and Brajard et al. (2012) (spectral optimization including NN); Doerffer and Schiller (2008, 2007), Schroeder et al.(2007) (NN inversion); Steinmetz et al. (2010) (polynomial atmospheric path radiance).
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ICOR, ATMOSPHERIC IMAGE CORRECTION
- Sensor generic (S2, L8, hyperspectral, … ) - Scene generic (Land and Water) - Focus on operationability - Handle land and water pixels - Focus on coastal and inland waters Previously known as OPERA
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
𝐿𝐿 𝑇𝑇𝑇𝑇𝑇𝑇 = 𝐿𝐿𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 + 𝐿𝐿𝑡𝑡𝑡𝑡𝑎𝑎−𝑝𝑝𝑡𝑡𝑡𝑡𝑝 + 𝐿𝐿𝑏𝑏𝑡𝑡𝑏𝑏𝑏𝑏𝑡𝑡𝑡𝑡−𝑝𝑝𝑡𝑡𝑡𝑡𝑝
De
Keu
kela
ere
et a
l. (s
ubm
itted
)
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
• Using external sources OR image based • Image based
– Guanter et al., 2005 – Derived above land – Spatially extended over water pixels – Fixed rural aerosol model
Latm−path & AOT retrieval
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
As seen by the satellite
Rayleigh scattering
ρw(λ) ?
Lbackgr.−path over water
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Lbackgr.−path over water
NIR similarity spectrum Ruddick et al. (2006 )
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Lbackgr.−path over water
Lake Marken - 15/09/16 Lake Marken - 15/09/16
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Lake Marken - 15/09/16
Transect – 865 nm (9km)
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Mantua Lakes - 23/09/14
No SIMEC SIMEC
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Lbackgr.−path over land
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
iCOR in SNAP
Let’s try it during 11:00 Practical hands-on session: iCOR atmospheric correction using the SNAP toolbox
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE Atmospheric Correction for High-Resolution Satellites
To develop AC algorithms for Sentinel-2, Landsat-8 taking TOA radiance data and auxiliary data as input and provide bottom of
atmosphere water reflectance as output.
L8 20140716
TOA radiances Rayleigh corrected radiances water reflectance
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE aims at
ρ Rayleigh
TOA
ρ aerosol
Rayleigh corrected
Retrieve water-leaving radiance reflectance, ρw, for turbid waters - use dark SWIR pixels to establish estimate of aerosol reflectance - SWIR threshold delineates water/non water - exponential extrapolation using 1.6/2.2 SWIR ratio (smoothed) - no direct AOT/Ta estimation
Sensitive to adjacency effects (inland waters), water surface effects (glint)
water reflectanc
e
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE processor
L1 Imagery
Rayleigh corrected imagery
Rayleigh correction: LUT based on 6SV and sun and sensor geometry
Land/cloud masking
ρc 6 > 0.0215
Cloud shadow detection -in development-
Rayleigh corrected imagery with water
pixels
Height of lake
Elevation automatically extracted from DE
Select clear water pixels
SWIR (1609,2201)
ε spatially
fixed or variable
NIR (655,865nm)
ε spatially
fixed
SWIR (865,1609nm)
ε spatially
fixed
SWIR (865,2201nm)
ε spatially
fixed
Aerosol type (ε) selection
ρw=𝟏𝟏𝒕𝒕
(𝝆𝝆𝒄𝒄 − 𝜺𝜺. ρam)
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Currently at version ‘2070718’ • Binary distribution of L8/S2 AC processor for Windows, Linux and MAC • Simple and fast processing of L8/S2 images for marine and inland water applications • AC is image based, no need for external Inputs!! • ε constant over scene (crop), aerosol multiple scattering reflectance varies per pixel • Use of settings file (created by acolite) batch processing
‘idl –rt=acolite.sac –args image=imagelist.txt sttings=acolite_settings_file’
ACOLITE software
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Standard Features • Atmospheric correction of L8/S2 • RGB image generation (TOA, RCO, pan sharpened), RGB scaling • extraction of rectangular regions • Mapping options: next to NetCDF, GeoTIFF, PNG now also KMZ • L2 processing products:
ACOLITE software
Rhow
CHL_OC2, CHL_OC3
SPM_NECHAD 561, 655, 865
SPM_Balaton
RTOA, LTOA DEM T_NECHAD_645
RRC, LRC FAI (floating algae index)
T_DOGLIOTTI (red, NIR, mix)
RHOAM NDVI, NDVI_TOA, NDVI_RCO
QAA_a QAA_bb QAA_Kd
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Qualitative inspection of ACOLITE products for L8-OLI
ACOLITE products for L8
Lake Balaton (20140716)
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Qualitative inspection of ACOLITE products for L8-OLI
ACOLITE products for L8
Lake Balaton (20140716)
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2
S2 10m resolution 16/1/2016
Ice
Ice
Thin cloud
Submerged ice
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2
S2 10m resolution 16/1/2016 Submerged
ice
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2
S2 665nm 10m resolution 16/1/2016
Cloud shadow
High reflectance by ice (cracks)
Striping in RGB bands
Water visible through ice?
Clouds flagged
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2 Cyanobacterial bloom in Lake Mulwala (Australia)
705 nm reference
665 nm chla absorption
9x9 pixels
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2 Cyanobacterial bloom in Lake Mulwala
9x9 pixels
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8 & S2 AC
Lake Balaton (20140716) *WISP-ASD-RAMSES-HyperSAS
Lake Mantua (20140923) *ASD, SE Curonian Lagoon (20140902)
*WISP
Inter-comparison exercise of AC procedures for L8 • In situ remote sensing reflectance gathered by CNR (and INFORM consortium) in
Balaton Lake, Kis, Lake Mantua and Curonian Lagoon • Match up time difference max 2h • Every match up QC for cloud, cloud shadow, glint
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Inter-comparison exercise of AC procedures • In situ Rrs data was QC and processed by individual partners responsible for data
acquisition 1 hyperspectral Rrs measurement per station
• Resampled to L8-OLI bands and S2-MSI to enable direct comparison (using SRF)
Validation of L8 & S2 AC
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Balaton 20160721 L8-OLI
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8-OLI 443nm
490nm
ACOLITE
iCOR
ACOLITE Slope = 0.927 R² = 0.921 rmse = 0.004
iCOR / iCOR-SIMEC Slope = 1.075 / 1.114 R² = 0.934 / 0.939 rmse = 0.004 / 0.004
iCOR-SIMEC
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of S2-MSI
ACOLITE Slope = 1.166 R² = 0.752 rmse = 0.003
iCOR / iCOR-SIMEC Slope = 0.920 / 0.630 R² = 0.950 / 0.812 rmse = 0.0002 / 0.001
ACOLITE ACOLITE
iCOR iCOR-SIMEC
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
• Atmospheric correction could be considered the most important step in the generation of L2W products.
• iCOR (SNAP) & ACOLITE (stand alone) open access software provides AC for water applications with land-sensors
• iCOR & ACOLITE show good results in many cases but limitations have been identified (e.g. noisy SWIR bands S2)
• Limitations due to sensor characteristics have been mentioned in the recommendations for future Copernicus missions
• BEL-Ponder: development of new AC approach (multi-band dark pixel fitting) to deal with AC issues for current sensors
Summary
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Bonus slides
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8 with MOW1 data
Validation of ACOLITE using BE AERONET
Match-up analysis Zeebrugge-MOW1 station (prelim results): • AERONET-DATA from Febr 2014-Febr
2016 + Aug-Sept 2016
• Level 1 13 match ups • Level 1.5 1 match up • Level 2 1 match up
• ACOLITE v20160602v2 (SubF_Unity) • Actual pixel value • < 1h time difference
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8 with MOW1 data
T5.1.3 Validation of ACOLITE
443nm
490nm
560nm
665nm
865nm
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of S2-MSI with AERONET-OC data
T5.1.3 Validation of ACOLITE
• AERONET-DATA 2016 MOW1: since Aug 2016 Cpower: full S2-MSI mission
• Level 1.5 16 match ups (15 Cpower, 1 MOW1) • 8 Cpower and 1 MOW1 matchup remained after quality control (clouds,
haze, wave effects, glint, …) • ACOLITE v20161123v1.0 (SWIRvar_Unity)
• 300m by 300m (31x31 pixels)
• < 1h time difference
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
T5.1.3 Validation of ACOLITE
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
T5.1.3 Validation of ACOLITE
Increased scatter towards shortest blue band (bug in ACOLITE?)
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
T5.1.3 Validation of ACOLITE
665nm: Increased scatter in clear water match-ups due to differences between SRFs of narrow AERONET-band and
wide S2-MSI band
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
T5.1.3 Validation of ACOLITE
Difficulties to measure a very low NIR signal in clear waters show the limits of the sensor and atmospheric correction
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8 AC –ACOLITE- 560nm 443nm
490nm 665nm
865nm
MAPE
ACOLITE
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8 AC –iCOR- 560nm 443nm
490nm 665nm
865nm
MAPE
iCOR
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of L8 AC –iCOR SIMEC- 560nm 443nm
490nm 665nm
865nm
MAPE
iCOR-SIMEC
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of S2 AC –ACOLITE- 560nm 443nm
490nm 665nm
865nm
MAPE
ACOLITE
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of S2 AC –iCOR- 560nm 443nm
490nm 665nm
865nm
MAPE
iCOR
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of S2 AC –iCOR SIMEC- 560nm 443nm
490nm 665nm
865nm
MAPE
iCOR-SIMEC
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Validation of S2-MSI
490nm
ACOLITE Slope = 0.952 R² = 0.767 rmse = 0.003
iCOR / iCOR-SIMEC Slope = 0.920 / 0.630 R² = 0.950 / 0.812 rmse = 0.0002 / 0.001
ACOLITE
iCOR
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
Inter-comparison exercise of AC procedures • In situ remote sensing reflectance gathered by CNR (and INFORM consortium) in
Balaton Lake, Kis, Lake Mantua and Curonian Lagoon, Lake Garda, Lake Iseo, Lake Marken
Validation of L8 & S2 AC
ACOLITE (RBINS) –Based on approach of Ruddick et al. (2000) • Input: target altitude; sun-target geometry • Aerosol model and aerosol reflectance estimated from spectral imagery • No adjacency correction • Water applications
iCOR (VITO) – Based on MODTRAN • Input: target altitude; sun-target geometry; • Visibility with land approach • Fixed rural aerosol model • Spatially extended over water pixels • procedure for adjacency effects • Land and water applications
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2 Lake Garda
17/8/2016 12/9/2015 1/7/2016
Rrs664
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2
S2 443nm 60m resolution 16/1/2016
Mis-alignment of lenses for 443 band?
User Uptake Workshop
16-17 November 2017
European Environment Agency, Copenhagen, Denmark
ACOLITE products for S2
S2 497nm 10m resolution 16/1/2016
NO Mis-alignment of lenses for 497 band
Related Documents
