MODIS/Aqua Ocean Reprocessing

Bryan Franz

Ocean Biology Processing Group

MODIS Science Team Meeting

22-24 March 2005

Introduction

• MODIS/Aqua Ocean Reprocessing 1 completed March 2005.– L1A-GEO-L1B-L2 completed less than 1 week (~150X)

– Also completed SeaWiFS Reprocessing 5

• This presentation will cover– Remaining issues from initial OBPG processing

– Changes implemented for this reprocessing

– Analysis of results

– Remaining issues & work in progress

– New products available for evaluation

Issues Following Initial OBPG Processing of MODIS/Aqua

Completed May 2004

Issues from Initial ProcessingCompleted May 2004

1 Inconsistency in the annual cycle of global water-leaving radiances

MODIS/AquaSeaWiFS R4.1

Deep-Water - mean of all common bins deeper than 1000 meters

Issues from Initial Processing

2 Long-term (3% per year) and seasonal (+/- 10%) deviations relative to SeaWiFS R4.1 global trends

Global Deep-Water nLw Ratio

+50%

-50%

~ 3% change/year

+10%

-10%

Issues from Initial Processing

3 Differences relative to SeaWiFS R4.1 which increase with latitude or solar zenith angle

35 North

25 North

15 North

nLw Ratios, Zonal Pacific, 150W-170W

+20%

-20%

Issues from Initial Processing

4 Very large differences relative to SeaWiFS R4.1 in global mean aerosol optical thickness retrievals

MODIS and SeaWiFS Deep-Water AOT Comparison

MODIS/Aqua

SeaWiFS

0.15

0.10

Issues from Initial Processing

5 Larger deviations in recent months due to out-dated MODIS instrument calibration look-up table (LUT)

Last SD Measurementfor Operational LUT

MODIS/SeaWiFS Deep-Water nLw Ratio

Reprocessing Changes

Changes for 1st Aqua Reprocessing

1 Updated instrument calibration• MCST V5.0.1 equivalent, OBC measurements to Nov 2004.

• Refit by OBPG as single exponential after removal of detector-dependent screen vignetting residuals.

Before LUT Change After LUT and Other Changes

Changes for 1st Aqua Reprocessing

2 Straylight Masking• Modeled point-spread function (PSF) indicates significant

sensitivity to straylight from adjacent sources.

• 7 x 5 masking around cloud and saturation pixels removes the most significant contamination, fixes AOT discrepancy.

Before SL Masking After SL Masking

MODIS Band 16 PSF MODIS & SeaWiFS Deep-Water AOT Comparison

0.15

0.10

Changes for 1st Aqua Reprocessing3 Changes to atmospheric correction

• Updated relative spectral response (RSR) functions: new Rayleigh tables and band-averaged coefficients.

• Revised Rayleigh pressure correction (M. Wang), to include solar zenith dependence (effect can be significant at high latitudes, for large deviations from standard pressure).

Deep-Water nLw Ratio 40-50N Pacific nLw Ratio

Impact of Change on MODIS nLw

+1%

-10%

Changes for 1st Aqua Reprocessing

4 Enhanced Fresnel correction for Lw normalization • include reflection/refraction to downwelling irradiance.

Fresnel Correction, Atmosphere to Ocean

SeaWiFS

Deep-Water nLw Ratios

+1%

40-50N Pacific nLw Ratios

SeaWiFS

+7%

Wind = 0 m/s

Wind = 20 m/sM. Wang Model

Changes for 1st Aqua Reprocessing5 Morel f/Q correction for Lw normalization

• accounting for non-isotropic sub-surface light field

Deep Water nLw RatioMODIS/SeaWiFS

After f/Q

Southern PacificMODIS/SeaWiFS

Before f/Q+20%

0%

-20%

+50%

0%

-50%

Changes for 1st Aqua Reprocessing5 Morel f/Q correction for Lw normalization

• accounting for non-isotropic sub-surface light field

After f/QBefore f/Q

Residual Scan Dependence in MODIS nLw(443)

+2%

-4%

+10%

0%

-10%

Changes for 1st Aqua Reprocessing

6 MOBY data reprocessing, final vicarious calibration• largest effect in 412 and 551-nm bands

Deep-Water Coastal

Impact of MOBY Change on MODIS nLw

+2%+4%

Reprocessing Results

• New LUT• Straylight rejection• RSR and pressure corrections• Fresnel• f/Q• New MOBY vicarious cal

Reprocessing Results

• New LUT• Straylight rejection• RSR and pressure corrections• Fresnel• f/Q• New MOBY vicarious cal

Also applied in SeaWiFS Reprocessing 5

Initial Processing Reprocessing 1

Results from Reprocessing 1nLw Comparison with In situ

SeaWiFS R5 MODIS/Aqua R1

Results from Reprocessing 1nLw Comparison with In situ

SeaWiFS R5 MODIS/Aqua R1

Results from Reprocessing 1Chlorophyll Comparison with In situ

Results for Reprocessing 1

Consistency in annual cycle is approaching SeaWiFS

MODIS/Aqua R1SeaWiFS R5

Results from Reprocessing 1nLw Comparison to SeaWiFS R5

• Sensor agreement to within 7% for global mean deep-water nLw retrieval. Better for clear water.

• Some bias & trend is still evident between instruments.

Initial Processing Reprocessing 1

+10%

0%

-10%

Results from Reprocessing 1

35 North

25 North

15 North

nLw Ratio to SeaWiFS R5, Zonal Pacific, 150W-170W

Initial Processing Reprocessing

+/- 20%

+/- 10%

+/- 5%

Laitu

de (S

olar Z

enith

)

Results from Reprocessing 1Deep-water AOT and Chlorophyll Comparison to SeaWiFS

AOT (865) Chlorophyll

Solid Line: MODIS/Aqua R1Dashed Line: SeaWiFS R5

<Cmodis> = 0.174 mg m-3

<Cseawifs> = 0.171 mg m-3

Seasonal Chlorophyll Images

0.01-64 mg m-3

Summer 2004

Winter 2004

SeaWiFS R5MODIS/Aqua R1

Winter 2004

Summer 2004

Remaining Issues & Work in Progress

• Bias & trends in relative agreement with SeaWiFS – instrument calibration (SD fit, lunar vs solar, RVS change)

– effect of aerosols on polarization (analysis in progress)

– straylight correction (reduce masking losses)

– apparent earth shine in dark counts

• Residual cross-scan trends• Residual striping

Residual Cross-scan Trends

• Global mean residual RVS at +/- 3% in nLw• Already Level-3 masking at 60° sensor zenith

nLw(412) nLw(551)+20%

-20%

+10%

-10%

60° Sensor Zenith60° Sensor Zenith

Residual RVS & Striping in Red Bands

nLw(667) nLw(678)+20%

-20%

+10%

-10%

0%

• Fractionally large cross-scan artifacts in red, no explanation yet.

Residual Striping

• Global mean residual striping at +/- 2% in nLw• Consistent over life of mission (problem with SD cal?)

nLw(412) nLw(551)+10%

-10%

+5%

-5%

Progress on Striping Corrections

MCST Lunar Analysis ()

• derived by averaging all Lunar calibrations on a per-detector basis

OBPG TOA Analysis (,)

• derived by collecting 20-pixel along-track runs (10 detectors x 2 mirror sides) in clear atmosphere/low polarization/low chlorophyll conditions, correcting for relative change due to atmosphere, and averaging per detector, per mirror-side.

Red & NIR bands saturate on Moon

Effect of OBPG Striping CorrectionnLw 412 nm

Before Correction

After Correction

New Productsfor evaluation

New or Revised Productsimplemented for evaluation

Carder IPAR & ARP• ipar - instantaneous photosynthetically available radiation• arp - instantaneous absorbed radiation by phytoplankton

Chlorophyll Fluorescence and Supporting Products• flh - fluorescence line height• cfe - chlorophyll fluorescence efficiency

Clark Empirical Algorithms• chl_clark - chlorophyll concentration• poc_clark - particulate organic carbon• tsm_clark - total suspended matter

Gordon & Balch Calcite• calcite_2b - calcite concentration, 2-band algorithm• calcite_3b - calcite concentration, 3-band algorithm

New or Revised Productsimplemented for evaluation

IOP Models• gsm01 - Garver-Siegel-Maritorena Semi-Analytical• carder - Carder Semi-Analytical• qaa - Quasi-Analytical Algorithm *

IOP-based Products• Kd_lee - Z.P.Lee diffuse attenuation *

* Coming Soon, Thanks to P. Martinolich at NRL Stennis

In The Works

Daily-Average PAR (R. Frouin)• PAR - SeaWiFS PAR updated for MODIS

Seasonal Chlorophyll Images

0.01-64 mg m-3

Summer 2004

Winter 2004

SeaWiFS R5MODIS/Aqua R1

Winter 2004

Summer 2004

Extra Slides

Residual RVS & Striping in Red Bands

nLw(667) nLw(678)

+10%

-10%

+5%

-5%

+20%

-20%

+10%

-10%

New or Revised Productsimplemented for evaluation

Garver-Siegel-Maritorena Semi-Analytical Bio-Optical Model• chl_gsm01 - chlorophyll concentration• a_nnn_gsm01 - total absorption• bb_nnn_gsm01 - total backscatter• bbp_nnn_gsm01 - particulate backscatter • aph_nnn_gsm01 - absorption due to phytoplankton• adg_nnn_gsm01 - absorption due to gelbstof and detrital material

Carder Semi-Analytical Bio-Optical Model• chl_carder - chlorophyll concentration• a_nnn_carder - total absorption• bb_nnn_carder - total backscatter • bbp_nnn_carder - particulate backscatter• aph_nnn_carder - absorption due to phytoplankton• adg_nnn_carder - absorption due to gelbstof and detrital material• flags_carder - product-specific flags

nnn - sensor wavelength

Coming Soon

Quasi-Analytic Algorithm (Z.P. Lee)• a_nnn_qaa - total absorption• bb_nnn_qaa - total backscatter• bbp_nnn_qaa - particulate backscatter • aph_nnn_qaa - absorption due to phytoplankton• adg_nnn_qaa - absorption due to gelbstof and detrital material

IOP-based Diffuse Attenuation (Z.P. Lee)• Kd_nnn_lee - diffuse attenuation at sensor wavelength nnn

nnn - sensor wavelength

Thanks to P. Martinolich at NRL Stennis

On the Subject of Flags

• Each derived product should include tests for:1 poor or questionable quality, sets the Level-2 CHLWARN flag bit

2 algorithm failure, sets the Level-2 CHLFAIL flag bit

• The failure and warning flags can be masked at Level-3. The binner masking is product independent, all products binned at the same time are masked the same.

• Additional flags or supporting products can be associated with a derived product, for QC and evaluation purposes at Level-2. No impact to binning.

Flag Name Bit Position DescriptionCFLAG_FAIL 0x0001 failure (bad inputs)CFLAG_DEFAULT 0x0002 using default (empirical)CFLAG_BLEND 0x0004 blending with default algorithmCFLAG_UNPKG 0x0010 unpackaged caseCFLAG_PKG 0x0020 packaged caseCFLAG_HIPKG 0x0040 highly packaged caseCFLAG_GLOBAL 0x0080 global caseCFLAG_LO412 0x0100 low input 412 reflectanceCFLAG_LO555 0x0200 low input 55x reflectanceCFLAG_CHLINC 0x0400 large inconsistency with default

Carder Model Product-Specific Flag Definition

flags_carder

In situ Match-ups for MODIS/Aqua

55% of Radiance Match-ups

nLw: MODIS/Aqua vs SeaWiFS Deep-Water

Chlor-a

551 vs 555

488 vs 490

Simulated Center-Wavelength Differences(after band-pass correction)

Deep-Water Mean

Relative to SeaWiFS, the 488 and 551 bands should be elevated at chlorophylls typical of deep water (~0.2 mg/m3).

MODIS/AquaVariation of Geometry Along-Track

Eastern Scan Edge

Scattering Angle

Variation of Rayleigh Radiance Along-Track

NadirEastern Edge Western Edge

ScatteringAngle

RayleighRadiance412 nm

0.05 mW/cm2 um sr or ~2% of nLw

Cross-Scan Variation of Sensor View ZenithMODIS/Aqua

Coastal

Deep

Clear