seadas training ~ nasa ocean biology processing group 1 introduction to ocean color satellite...

SeaDAS Training ~ NASA Ocean Biology Processing Group

1

Introduction to ocean color satellite calibration

NASA Ocean Biology Processing Group

Goddard Space Flight Center, Greenbelt, Maryland, USA

SeaDAS Training Material


2

scope of the calibration paradigm:

to meet the accuracy goals, top-of-the-atmosphere

radiances need to have uncertainties lower than 0.5%

uncertainties are present in

* instrument characterization and calibration

* atmospheric and in-water data processing algorithms

Ocean color calibration


3

direct calibration

pre-launch: sensor is calibrated in a laboratory (thermal vacuum)

on-orbit: regular solar, deep-space, and lunar observations track changes in sensor response (possible additional on-board calibrators)

vicarious calibration

on-orbit: force instrument + atmospheric correction system to agree with sea-truth data (e.g., in situ measurements)

Instrument calibration stages


4

photons to data

each stage in this sequence contributes to uncertaintiesevery element needs:

to be well characterizedits calibration parameters derived

radiant source (Earth surface and atmosphere)scanning mirrorcalibratorsoptics (aperture, mirrors, beam splitters, objectives)filtersdetectorselectronicsanalog to digital (A/D) convertersdata formatters and data recordersground receiving antennadigital count to radiance conversion

Elements of instrument operation


5

SeaWiFS (12 noon descending orbit)

• Rotating telescope• 8 bands: 412, 443, 490, 510, 555, 670, 765,

865 nm• 12 bit digitization truncated to 10 bits on

spacecraft• 4 focal planes, 4 detectors/band, 4 gain

settings, bilinear gain configuration• Polarization scrambler: sensitivity at

0.25% level (for fully polarized light)• Solar diffuser (SD) daily observations• Monthly lunar views at 7° phase angle via

pitch maneuvers

MODIS-Aqua (1:30 pm ascending orbit)

• Rotating mirror• 9 OC bands: 412, 443, 488, 531, 551,

667, 678, 748, 869 nm• 12 bit digitization• 2 VIS-NIR focal planes, 10 to 40

detector arrays depending on band resolution, 0.25 to 1 km

• No polarization scrambler: sensitivity up to 6% at 412 nm

• Spectral Radiometric Calibration Assembly (SRCA)

• Solar diffuser (observations every two weeks), Solar Diffuser Stability Monitor (SDSM)

• Monthly lunar views at 55° phase angle via space view port

NPP/VIIRS (1:30 pm descending orbit)

• SeaWiFS-like rotating telescope• MODIS-like focal plane arrays• No polarization scrambler• Solar diffuser with stability monitor• 7 OC bands: 412, 445, 488, 555, 672, 746,

865 nmdifferences in sensor designdifferences in orbits

Example sensor specifications


6

MODIS instrument design


7

MODIS pre-launch characterization concerns

solar diffuser characterization• bidirectional reflectance factor (BRF)

impact on calibration• Earth shine effect – sunlight reflecting

off the Earth and onto the diffuser and adding to the solar irradiance

• attenuation screen characterization through vignetting function

• SDSM uncertainty in monitoring SD reflectance changes

mirror degradation, response vs. scan-angle (RVS), two mirror sides

detector calibration changes

polarization sensitivity

in-band and out-of-band response

instrument and focal plane temperature effects

electronic cross-talk

stray-light contamination

solar diffuser stability

stray-light contamination• photons in the optical path from

Earth coming from bright sources, i.e. clouds, land, and sun glitter (characterized by point spread function)


8

* MODIS solar diffuser calibrations performed at the Pole every 2 weeks* North Pole for Terra and South Pole for Aqua* at the dark side of the terminator to limit the stray light entering the instrument

Solar calibration


9

Moon acts as an external diffuserMoon is viewed at specific lunar phase angles

Lunar calibration


10

Lunar calibration


11

MODIS absolute radiometric accuracy

reflective solar bands (0.41–2.1m): ±2% in reflectance and ±5% in radiance

MODIS relative accuracy over time

reflective solar bands (0.41–2.1m): ±0.2% in reflectance

Direct calibration uncertainty limits


12

Vicarious calibration approach

on-orbit calibration

temporal change through the mission

vicarious calibration

single radiometric gain adjustment

NIR band calibration NIR band calibration

calibration of the combined instrument + algorithm system


13

cloud-free air mass with low optical thickness (e.g., AOT(865) < 0.1)

spatially homogeneous Lw() ~ or, Lw(NIR) = 0 for NIR calibration)

limited solar and sensor geometries, wind speed, stray-light and glint contamination

VIS calibration

NIR calibration

Criteria for vicarious calibration


14

TARGET

SATELLITE

TOP OF ATMOSPHERE

from the satellite

+ Lr , td , …

Lttarget

Criteria for vicarious calibration


15

provides a relative calibration between

the two NIR bands

based on assumptions of the most

probable maritime atmosphere

assumptions

open ocean is black in the NIR, i.e. Lw(748) and Lw(869) = 0

vicarious gain of band 869-nm is fixed at 1 based on on-orbit calibration only

maritime aerosol with 90% humidity (M90) is chosen over the calibration sites

band 869-nm defines the amount of aerosol, AOT(869)

aerosol radiance is tabulated for M90 and any geometry

MODIS Band

MODIS (nm)

8 412 9 443 10 488 11 531 12 551 13 667 14 678 15 748 16 869

NIR {

NIR vicarious calibration


16

Lw( , 0 )cos( 0 ) t( , 0 )

Visible band vicarious calibration

the Marine Optical Buoy (MOBY)

alternatives:

ocean surface reflectance model

alternative buoy

accumulated field campaigns


17

target data extract 5x5 box

locate L1A files

extract 101x101 pixel box

process to L2

Vicarious calibration


18


identify flagged pixels:

LAND, CLDICE, HILT,

HIGLINT, ATMFAIL,

STRAYLIGHT, LOWLW

require 25 valid pixels

calculate gpixel for each pixel in

semi-interquartile range; then:

gscene = gpixel / npixel

limit to scenes with average values:

< 0.20 Ca

< 0.15 (865)

< 60 sensor zenith

< 75 solar zenith

locate L1A files


process to L2

(1) calculate gains for each matchup



19


identify flagged pixels:

LAND, CLDICE, HILT,

HIGLINT, ATMFAIL,

STRAYLIGHT, LOWLW

require 25 valid pixels

calculate gpixel for each pixel in

semi-interquartile range; then:

gscene = gpixel / npixel

limit to scenes with average values:

< 0.20 Ca

< 0.15 (865)

< 60 sensor zenith

< 75 solar zenith

limit to gscene within

semi-interquartile range

visually inspect all scenes g = gscene /

nscene

locate L1A files


process to L2

(1) calculate gains for each matchup

(2) calculate final, average gain



20



21



22



23

changes in g with increasing sample size …

standard error of g decreases to 0.2%

overall variability (min vs. max g) approaches 0.5%

provides insight into temporal calibration, statistical choices



24

future ruminations …

… statistical and visual exclusion criteria influence g only slightly, yet

… they reduce the standard deviations … can uncertainties be quantified

… for the assigned thresholds?

… how do the uncertainties of the embedded models (e.g., f / Q, the NIR-

… correction, etc.) propagate into the calibration?

… what are the uncertainties associated with Lwtarget?



25

Franz et al., Appl. Opt. (2007) ~ vicarious calibration approach, using MOBY

Werdell et al., Appl. Opt. (2007) ~ vicarious calibration using an ocean surface reflectance model

Bailey et al., Appl. Opt. (in press) ~ vicarious calibration using alternative in situ data sources (e.g., NOMAD, BOUSSOLE)

Vicarious calibration references

seadas training ~ nasa ocean biology processing group 1 introduction to ocean color satellite...

Documents

calibration parameters

calibration paradigm

modis instrument design

instrument characterization

direct calibration prelaunch

solar irradiance attenuat

rotating telescope modis

nm differences