using multispectral, multiangle remote sensing ... · ocean sciences meeting 2016, me51a-02 3...

Jacek Chowdhary1, Kirk Knobelspiesse2, Brian Cairns1,3

Using multispectral, multiangle remote sensing observations for ocean

color retrievals: Studies performed for the NASA/PACE mission

1 Columbia University, New York, USA2 NASA Ames Research Center, Moffett Field, California, USA

Ocean Sciences Meeting 2016, ME51A-02

3 NASA/Goddard Institute for Space Studies, New York, USA

PACE Mission

In autumn of 2011 NASA selected a science definition team (SDT) to provide (within 3 years) thescience justification and measurement, as well as mission requirements, for the PACE mission

In autumn of 2014 NASA selected a science team (ST) to pursue studies on inherent opticalproperties (IOPs) and Atmospheric Correction (AC)

Responding to the Challenge of Climate

and Env ironmental Change

National Aeronautics and Space Administration

The Pre-Aerosol, Clouds, and ocean Ecosystem (PACE) mission

will make essential global ocean color measurements, essentialfor understanding the carbon cycle and how it both affects and

is affected by climate change, along with polarimetry

measurements to provide extended data records on clouds and

aerosols.

Pre-Aerosol, Clouds, and ocean Ecosystem (PACE)

“The PACE mission will extend key climate data records whose future was in jeopardy

prior to the FY2011 budget request. Global ocean color measurements, essential forunderstanding the carbon cycle and how it affects and is affected by climate change, will

be made by a radiometer instrument on this mission. A polarimeter instrument will

extend data records on aerosols and clouds using this approach begun by the French

PARASOL mission and expanded upon by NASA’s Glory mission, as well as multi-spectral and multi-angle measurements made by NASA’s MODIS and MISR instruments

on NASA’s EOS platforms (MODIS on terra and Aqua, MISR on Aqua).”

NASA’s plan for a Climate-Centric Architecture for Earth Observations and Applications from Space

June 2010

• Primary mission: Ocean; Secondary mission: atmosphere

• Atmospheric Correction: polarization

o hyperspectral between 800-900

nm, 1-2 nm subsamples (O2 A)

PACE Mission

o 3 SWIR bands

(1240, 1640, 2130 nm)

o spatial resolution = 1 km2

o improved global coverage (1 day)

o spatial resolution better than

500m × 500m

Threshold Ocean Questions SQ 1-7

Goal Terrestrial Questions TSQ 1-3

OCI †

OCI/OG OCI Questions (SQ 1-7, TSQ 1-3)

Goal Coastal Questions CSQ 1-4

o Hyperspectral imager with 5 nm

resolution between 350-800 nm

o 2 NIR bands

(incl. 865 nm)

o OCI instrument capabilities

Option Science Threshold and Goal Questions Brief Instrument Description

Oce

an

Scie

nce o 3 SWIR bands

(1240, 1640, 2130 nm)

o spatial resolution = 1 km2

Note: Threshold Questions define required

research, i.e. they must be addressed

† OCI = Ocean Color Imager ‡ improved atmospheric correction, data continuity for POLDER products§ data continuity for MISR products, albeit with coarser spatial resolution¶ 3M = Multi-directional, Multi-polarization, Multi-spectral

OCI+

OCI-3M ¶

OCI/A

OCI/A-3M

OCI Questions (SQ 1-7, TSQ 1-3)

“Threshold” Atmosphere Question ASQ 1

OCI Questions (SQ 1-7, TSQ 1-3)

Goal Atmosphere Questions ASQ 4,5

OCI+ Questions (SQ 1-7, TSQ 1-3, ASQ 1)

Goal Atmosphere Question ASQ 2


o 3 additional NIR and SWIR

bands (940, 1378, 2250 nm)


o a 3M imager ‡ §

o OCI+ instrument capabilities

o selected atmospheric bands at

spatial resolution 250m × 250m

OCI-3M & OCI/A Questions (SQ 1-7,

TSQ 1-3, ASQ 2,4,5)

Goal Atmosphere Question ASQ 3

o OCI/A instrument capabilities

o a 3M imager ‡ §

Atm

osp

he

re S

cie

nce

absorbingaerosols

CDOM

phytoplankton

pigments

functional groups

particle sizes

physiology

pigment

fluorescence

coastal biology

atmosphericcorrection

(clear ocean)


(coastal ocean)

SW

IRN

IRV

isib

leU

ltra

vio

let

5 n

m r

eso

luti

on

(3

50

-885

nm

)

26

re

qu

ire

d “m

ult

isp

ectr

al” b

an

ds

3 S

WIR

ba

nd

s

products

PACE Mission

NO2

UV-A Visible NIR SWIR

aerosol absorption

dark ocean

aerosol scattering

black oceanocean science products

Atmospheric Scattering (AS) estimate

NIR & SWIR band ratios: select aerosol scattering model

UV-A radiance: detect (& constrain?) aerosol absorption

AS

~ρw (412)

10–110–2 100 101

10–2

10–1

100

101

Sa

telli

te

In Situ

Santa Barbara Channel, CA

٭ MODIS

+ SeaWiFS

absorbingaerosols

CDOM

phytoplankton

pigments

functional groups

particle sizes

physiology

pigment

fluorescence

coastal biology


(clear ocean)


(coastal ocean)

SW

IRN

IRV

isib

leU

ltra

vio

let

5 n

m r

eso

luti

on

(3

50

-885

nm

)

26

re

qu

ire

d “m

ult

isp

ectr

al” b

an

ds

3 S

WIR

ba

nd

s

products

PACE Mission

NO2

UV-A Visible NIR SWIR

aerosol absorption

dark ocean

aerosol scattering

black oceanocean science products

Atmospheric Scattering (AS) estimate

NIR & SWIR band ratios: select aerosol scattering model

UV-A radiance: detect (& constrain?) aerosol absorption

AS

~ρw (412)

10–110–2 100 101

10–2

10–1

100

101

Sa

telli

te

In Situ

Santa Barbara Channel, CA

٭ MODIS

+ SeaWiFS

absorbingaerosols

CDOM

phytoplankton

pigments

functional groups

particle sizes

physiology

pigment

fluorescence

coastal biology


(clear ocean)


(coastal ocean)

SW

IRN

IRV

isib

leU

ltra

vio

let

5 n

m r

eso

luti

on

(3

50

-885

nm

)

26

re

qu

ire

d “m

ult

isp

ectr

al” b

an

ds

3 S

WIR

ba

nd

s

products

When is the ocean bright enoughto retrieve variations in CDOM?

When is the ocean dark enoughto retrieve variations in aerosolabsorption?

RT Comparison studies

Motivation

Results

1) Accuracy for Stokes parameters I, Q, Uemerging from various AOS is better than 10–6

2) Corresponding accuracy in degree of linear polarization better than 0.1%

Part I

RT comparison Studies

Motivation Polarization is an extremely useful tool to retrieve aerosol properties

Synthetic TOA data of I, Q, & U:

o Fine mode aerosol, τ = 0.2

(re = 0.4 μm; ve=0.2; m=1.45

o Rough ocean surface

(W = 7 m/s)

o Black water body

o μ0=0.8; μ=0.2, 0.4, 0.6, 0.8, 1.0

Δφ=60º & 120º

θ0 ≡ π – ϑ0

oce

an

atm

osp

her

e

x

y

zsun

k0

φ0 = 0º

view

k

ϑ

φ

AOS system

provided that it is measured with very high accuracies (0.2%–0.5%)

Source: Mishchenko and Travis, JQSRT 102:13,543-13,553 (1997)

Simulated aerosol retrieval from space-borne observation over ocean at 865 nm

Aerosol candidate models:

o Fine mode aerosol:

τ = 0.01 – 0.4, Δτ = 0.01

re = 0.01 – 0.8 μm, Δre = 0.01

m = 1.3 – 1.7, Δm = 0.01

ω = 0.78 – 1.00, Δω = 0.02

ve = 0.2

>350,000 aerosol candidate models


Motivation



τ = 0.01 – 0.4, Δτ = 0.01

re = 0.01 – 0.8 μm, Δre = 0.01

m = 1.3 – 1.7, Δm = 0.01

ω = 0.78 – 1.00, Δω = 0.02

ve = 0.2



ω=1.00 ω=0.98

ω=0.94 ω=0.92

optic

al th

ickn

ess

optic

al th

ickn

ess

refractive index refractive index

o radiance I,

o 9 viewing angles

o ΔI = 4%

o polarization Q/I and U/I,

o 9 viewing angles

o ΔP = 0.2%


Polarization is an extremely useful tool to retrieve aerosol properties



Motivation

o radiance I,

o 9 viewing angles

o ΔI = 6%


o 9 viewing angles

o ΔP = 0.8%

optic

al th

ickn

ess

optic

al th

ickn

ess


ω=1.00 ω=0.98

ω=0.94 ω=0.92







τ = 0.01 – 0.4, Δτ = 0.01

re = 0.01 – 0.8 μm, Δre = 0.01

m = 1.3 – 1.7, Δm = 0.01

ω = 0.78 – 1.00, Δω = 0.02

ve = 0.2



Motivation

o radiance I,

o 9 viewing angles

o ΔI = 8%


o 9 viewing angles

o ΔP = 2.0%

optic

al th

ickn

ess

optic

al th

ickn

ess


ω=1.00 ω=0.98

ω=0.94 ω=0.92







τ = 0.01 – 0.4, Δτ = 0.01

re = 0.01 – 0.8 μm, Δre = 0.01

m = 1.3 – 1.7, Δm = 0.01

ω = 0.78 – 1.00, Δω = 0.02

ve = 0.2



2015:

~1e-4

~1e-4

~1e-4

absolute difference

I

Q

UAOS system: molecular atmosphere above ocean surface

view angle

0 20 40 60 80

view angle

0 20 40 60 80



our forward RT computations need to match these accuracies!

Motivation


Results

θ0 ≡ π – ϑ0

oce

an

atm

osp

her

e

x

y

z

k0

φ0 = 0º

k

ϑ

φ

upper

lower

2 sun angles

13 viewing angles

4 azimuth angles

TOA

SRF

>100 scattering geometries x 2 altitudes

λ = 350 nm, 450 nm, 550 nm, 650 nm



I, Q, U ΔP ≤ 0.1%



models Ocean Body Ocean Surface Atmosphere

AOS-I

none

rough Gaussian isotropic No foam or shadowing

molecular Pure Rayleigh scattering

AOS-II pure water Pure Rayleigh scattering


none

AOS-III pure water Pure Rayleigh scattering



AOS-IV pure water & hydrosol Pure Rayleigh scattering Detritus-Plankton mix



AOS-V pure water & hydrosol Pure Rayleigh scattering Detritus-Plankton mix


molecular & aerosol Pure Rayleigh scattering Fine-mode aerosol

Results

10

–10

5

0

–5

–60–40–20 0 20 40 60

dI (x106) dQ (x106) dU (x106) dP (%)

SRF, θ0=60° SRF, θ0=60° SRF, θ0=60° SRF, θ0=60°10

–10

5

0

–5

–60–40–20 0 20 40 60

10

–10

5

0

–5

–60–40–20 0 20 40 60

0.2

–0.2

0.1

0.0

–0.1

–60–40–20 0 20 40 60

φ = 60° φ = 120°←║ φ = 60° φ = 120°←║ φ = 60° φ = 120°←║ φ = 60° φ = 120°←║

AOS-I (550 nm)

view angle view angle view angle view angle



GSFCJPLNRL

UCSDUMBC



models Ocean Body Ocean Surface Atmosphere

AOS-I

none



AOS-II pure water Pure Rayleigh scattering


none

AOS-III pure water Pure Rayleigh scattering



AOS-IV pure water & hydrosol Pure Rayleigh scattering Detritus-Plankton mix



AOS-V pure water & hydrosol Pure Rayleigh scattering Detritus-Plankton mix


molecular & aerosol Pure Rayleigh scattering Fine-mode aerosol

Results

10

–10

5

0

–5

–60–40–20 0 20 40 60

dI (x106) dQ (x106) dU (x106) dP (%)

SRF, θ0=60° SRF, θ0=60° SRF, θ0=60° SRF, θ0=60°10

–10

5

0

–5

–60–40–20 0 20 40 60

10

–10

5

0

–5

–60–40–20 0 20 40 60

0.2

–0.2

0.1

0.0

–0.1

–60–40–20 0 20 40 60

φ = 60° φ = 120°←║ φ = 60° φ = 120°←║ φ = 60° φ = 120°←║ φ = 60° φ = 120°←║

AOS-I (550 nm)

view angle view angle view angle view angle

view angle

0 20 40 60 80

~1e-4

dI

~1e-4

dQ

view angle

0 20 40 60 80

~1e-4

dU

view angle

0 20 40 60 80

o Benchmarked >magnitude better

o Satisfies polarization accuracy



GSFCJPLNRL

UCSDUMBC



UV sensitivity Studies

Motivation

Results

1) TOA polarized reflectance is more sensitive to variations in aerosol height than in ocean or aerosol microphysical properties

2) It varies by (much) more than 0.001 per km change in aerosol height in boundary layer

Part II

UV Sensitivity Studies

Motivation


Reflection r = p I

S0 cosq0

60° 40° 20°

60° 60°

120°120°

q =

Azimuth angle

View angle

Sideward scattering

Backward scattering

Specular reflection

Sunglint region

Anti-solar point for q0

(i.e. backscattering direction)

j−j0 = 0°

180°

UV polarized reflected light is nearly insensitive to variations in the ocean color

But is this light sensitive to the presence and variations in aerosols?

rS0

TOA

SRF

385 nm

4 km


j−j0 = 180°

j−j0 = 0°

rocean (%) (385 nm)

3 3.5 4 4.5 5 7 12 5 13 13.56 ><

contribution of ocean to TOA reflectance: >13%

TOA

SRFrocean (%)

rocean, ,

P(%)

rocean (%) ≡ rocean / r TOA

Case I (open) ocean

Rayleigh scattering

Bodhaine et al. (1999)

Aerosol scattering

re = 0.15 μm, ve = 0.15, τ550 = 0.10

contribution of ocean to TOA polarized reflectance: <5%

average oligotrophic ocean

AOT(550) = 0.1, θ0 = 30°

Chow

dhary

et al. (2012)

4 km

RT results

385 nm

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°


j−j0 = 180°

j−j0 = 0°

ΔrTOA (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

TOA

SRF

4 km

DrTOA (%)

DrTOA, ,

P(%)

DrTOA (%) ≡ DrTOA / r TOA

Case I (open) ocean

Rayleigh scattering

Bodhaine et al. (1999)

Aerosol scattering

re = 0.15 μm, ve = 0.15, τ550 = 0.10

max CDOM variation

AOT(550) = 0.1, θ0 = 10°

TOA polarized reflectance change: <2%

380 nm

Mediterranean

South Pacif ic

Chl [mg/m3]

DKbio = 0.2433

bio-optical model

Kd

≡ K

w+

Kbio

[1/m

]

More

l et al. (2007)

DKd ≈ 1.04(md)−1 (Da + bb)

CDOM

RT results

TOA reflectance change: ~8%

385 nm

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°


j−j0 = 180°

j−j0 = 0°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

ΔrTOA (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

ΔrTOA (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

TOA

SRF

4 km

DrTOA (%) ≡ DrTOA / r TOA

DrTOA (%)

DrTOA, ,

P(%)

remove aerosol

DrTOA (%)

DrTOA, ,

P(%)

max CDOM variation


AOT(550) = 0.1, θ0 = 30°

RT results


385 nm

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

TOA polarized reflectance change: >6%


RT results

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

ΔrTOA (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

max CDOM variation

DrTOA (%)

DrTOA, ,

P(%)

ΔrTOA (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

change aerosol z change aerosol z change aerosol z

SRF

4 km

6 km

change aerosol z change aerosol z

8 km

10 km

change aerosol z

4 km

2 km

SRF



60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

TOA polarized reflectance change: >6%


60° 40° 20°

300° 60°

120°240°

q =60° 40° 20°

300° 60°

120°240°

q = 60° 40° 20°

300° 60°

120°240°

q =

q =60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

q =60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

DrTOA (%)

DrTOA, ,

P(%)

ΔrTOA,P (absx1000) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

ΔrTOA, P (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

ΔrTOA (%) (385 nm)

0.5 1 1.5 2 2.5 3.5 4 5 6 93 ><

DrTOA, ,

P(abs) 60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 180°

RT results

SRF

4 km

6 km

8 km

10 km

4 km

2 km

SRF

change aerosol z change aerosol z change aerosol z

TOA change: >1%

TOA change: >0.001

4 km

6 km

385 nm 385 nmTOA

SRF

TOA

SRF

2 km


RT results

AOT(550) = 0.1, θ0 = 30°

Changing height of aerosols in lower 5 km

backscattering direction

ΔrTOA, P per km vertical change (abs x 1000)

0.5 1 1.5 2 2.5 3.5 4 5 6 83 ><


Change in TOApolarized reflectance

per km shift in aerosol height: >0.001

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

j−j0 = 180°

4 km

6 km

385 nm 385 nmTOA


0.5 1 1.5 2 2.5 3.5 4 5 6 83 ><


Change in TOApolarized reflectance

per km shift in aerosol height: >0.001

SRF

TOA

SRF

2 km

60° 40° 20°

300° 60°

120°240°

q =

j−j0 = 0°

j−j0 = 180°

60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

j−j0 = 180°


RT results Changing height of aerosols in lower 5 km

case for absorbing aerosol remains ~same !

absorbing aerosol: ω550 = 0.91

RT Comparison studies 1) Accuracy for Stokes parameters I, Q, Uemerging from various AOS is better than 10–6

2) Corresponding accuracy in degree of linear polarization better than 0.1%

Conclusions

Hydrosol models 1) Approximate, but not a substitute for, real ocean color spectra

1) Allow for simultaneous retrieval of aerosol and ocean color spectra from polarimetric data

Part I

Part II

Backup slides

UV sensitivity Studies 1) TOA polarized reflectance is more sensitive to aerosol height than to aerosol microphysical properties

2) It varies by (much) more than 0.001 per km change in aerosol height in boundary layer

Backup Slides

aerosol model variations

q = 60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

q = 60° 40° 20°

300°

240°

j−j0 = 0°

40° 20°

60°

120°

j−j0 = 0°

q = 60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

q = 60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

q = 60° 40° 20°

300°

240°

j−j0 = 0°

40° 20°

60°

120°

j−j0 = 0°

q = 60° 40° 20°

300° 60°

120°240°

j−j0 = 0°

pola

rized r

efle

cta

nce c

hange

tota

l refle

cta

nce c

hange

ω(550) = 0.91 re = 0.10 μm τ(550) = 0.2reference aerosol


0.5 1 1.5 2 2.5 3.5 4 5 6 83 ><

ΔrTOA per km vertical change (abs x 1000)

Changing height of aerosols in boundary layer by 1 km


Backup Slides

Hydrosol models

ACROSS Update hydrosol model couple atmosphere & ocean in atmospheric correction

Retrieve aerosol model/properties while accounting for underwater light

scattering requires an ocean model

o NIR total radiance

o NIR polarized radiance

o VIS total radiance

o VIS polarized radiance

o UV total radiance

o UV polarized radiance

o Ocean model

assumptions (§)

“ –” below surf

)IOPsview,(,,model w, UQI=r(§)

rr=r retrievew,atmsrf-atmTOA t

Retrieve water-leaving radiance in the VIS part of the spectrum

o VIS total radiance (¶)

(¶)

“ +” above surf

aerosol properties

Note:r modelw, : approximates the ocean in aerosol retrieval

r retrievew, TOAr: extracted from using the retrieved aerosol

rDr wmodelw,r retrievew,

ocean model

spectral residual fit

retrieval

spectral residual in fit of

cannot be resolved with aerosolTOAr

Hydrosol models

Update hydrosol model couple atmosphere & ocean in atmospheric correction

Validation ocean retrieval

RSP ocean retrieval

Hydrosol models

ACROSS

Update hydrosol model couple atmosphere & ocean in atmospheric correction

IOCCG5 model

(back-up slides)

Hydrosol models

ACROSS

using multispectral, multiangle remote sensing ... · ocean sciences meeting 2016, me51a-02 3...

Documents