Edward Hyer Naval Research Laboratory JCSDA Symposium 11 October 2012

NPP VIIRS Aerosol Products:Overview, early cal/val results, and data assimilation results from NAVDAS-AOD

Edward HyerNaval Research Laboratory

JCSDA Symposium11 October 2012

In this talk

• Overview of NPP VIIRS aerosol products• Status of NPP Cal/Val effort• Comparison of MODIS and VIIRS products• Preparation of DA-ready obs from VIIRS• Impact of VIIRS obs in NAAPS• Future Work• MODIS DA-ready AOD now available from LANCE

3

NPP Visible Infrared Imaging Radiometer Suite (VIIRS)

0130, 1330 local time orbits“Early afternoon slot”

Credit: NASA

October28 2011

• OLS

Heritage Operational and R&D Sensors

4

AVHRR

SeaWiFSMODIS

R&D Sensors

Ocean ColorBands

RotatingTelescope

Band Selection/Continuity Thin Cirrus Band Solar Diffuser Earthshine Lessons Learned

Operational Sensors High Spatial Resolution Day/Night Band Minimize Resolution

Growth Over Scan

VIIRS

275 kg 22 bands

OMM EM

Radiometric Accuracy

SST Band Continuity

33 kg 5 bands

45 kg 8 bands

220 kg 36 bands

OLS

74 kg 2 bands

Credit: Northrup Grumman & Raytheon

Heritage Capabilities

5

0.4 140.6 0.7 0.8 0.90.5 1.2 1.8 6 82.4 4 10 12

[micron]

OLS

AVHRR

SeaWiFS

MODIS

VIIRS DNB

Low Light Imagery

Long wave IRMid wave IRVisible Breaks in scale

Credit: Northrup Grumman & Raytheon

Simplified VIIRS Aerosol Process

SDR: Calibrated TOA Radiances / Reflectances

VIIRS Cloud Mask

Aerosol Intermediate

Product IVAOT

Met. Fields (NCEP)(NOGAPS backup)

NAAPS Aerosol Optical Depth

Aerosol EDR VAOOO (contains no NAAPS)

VIIRS Aerosol EDR Product

• ~6km resolution– 8x8 ~750m pixels

• Retrieved: – AOT – Angstrom Exponent– Suspended Matter Type

• Many Quality Flags: – Cloud (incl. adjacent)– Glint and other geometry– Input data quality

• Retrieval is based on LUTs generated with 6S

• Heritage is MOD09 Surface Reflectance

• Over-ocean similar to MOD04– Fine-coarse mixture retrieved

explicitly

• Over-land uses different aerosol models– Single model (fine+coarse) is

retrieved via LUT

Aerosol EDR VAOOO

Status of Aerosol Cal/Val

• Aerosol EDR is sum of Parts:– VIIRS SDR

radiances/reflectances: These are still Beta, Provisional status will be considered later this fall

– VIIRS Cloud Mask: basic tuning complete, product is still Beta

• Aerosol EDR is recently partially advanced to ‘Beta’ maturity:– Global AOD– Over-ocean Aerosol Size

Parameter– NOT over-land Aerosol

Size Parameter– NOT Suspended Matter

EDR

MODIS and VIIRS AOD

MODIS-Aqua VIIRS-NPP

Credit: Min Oo, Bob Holz, Atmospheres PEATE, University of Wisconsinhttp://peate.ssec.wisc.edu/flo/npp/gridding?date=2012%2F09%2F24

MODIS and VIIRS AOD

MODIS-Aqua VIIRS-NPP

Credit: Min Oo, Bob Holz, Atmospheres PEATE, University of Wisconsinhttp://peate.ssec.wisc.edu/flo/npp/gridding?date=2012%2F09%2F24

GA

P

NO

GA

P

VIIRS has wider swath– complete overlap even at equator

MODIS and VIIRS AOD

MODIS-Aqua VIIRS-NPP

Credit: Min Oo, Bob Holz, Atmospheres PEATE, University of Wisconsinhttp://peate.ssec.wisc.edu/flo/npp/gridding?date=2012%2F09%2F24

VIIRS and MODIS show broad agreement in large features

MODIS and VIIRS AOD

MODIS-Aqua VIIRS-NPP

Credit: Min Oo, Bob Holz, Atmospheres PEATE, University of Wisconsinhttp://peate.ssec.wisc.edu/flo/npp/gridding?date=2012%2F09%2F24

VIIRS and MODIS have different cloud screening– VIIRS is still being characterized. Both products must avoid clouds for accurate retrieval.

MODIS and VIIRS AOD

MODIS-Aqua VIIRS-NPP

Credit: Min Oo, Bob Holz, Atmospheres PEATE, University of Wisconsinhttp://peate.ssec.wisc.edu/flo/npp/gridding?date=2012%2F09%2F24

VIIRS and MODIS show significant differences over land

VIIRS vs. AERONET AOD over ocean

• Best QA only• Low mean bias• Not enough data to

clearly diagnose bias at higher AOD

VIIRS vs. AERONET AOD over land

• Best QA only• Significant positive

mean bias• Positive bias for all AOD

ranges observed

AOD Pre-processing for DA

MODIS• NRL & University of North

Dakota developed method starting with MOD04 Level 2 data

• Extensive filtering for cloud, surface, etc.

• Empirical Correction• Prognostic Uncertainty

Model• Now a product produced in

NRT at NASA LANCE

NPP VIIRS• Method under

development• Currently only filters

using built-in QA• No corrections• Aggregated to NAAPS 1-

degree resolution

NAVDAS-AOD Testbed

Start with best analysis from research run (includes assimilated MODIS AOD)

NAVDAS-AOD Testbed

Assimilate NPP VIIRS AOD (+/- 3 hours)

NAVDAS-AOD Testbed

Initialize 72-hour forecasts from MODIS and MODIS+NPP analyses

NAVDAS-AOD Testbed

Compare forecast with obs to evaluate propagation of differences

MODIS obs do not show high AOD over ocean west of Hawaii

NAVDAS-AOD Testbed

Compare forecast with obs to evaluate propagation of AOD differences

MODIS and VIIRS obs do not show high AOD over ocean west of Hawaii

Much More Work To Do

• Quantitative Evaluation of NAAPS-MODIS vs NAAPS-MODIS+NPP analysis and forecast

• Determination of impact on forecast error of including over-ocean VIIRS AOD

• Construction and evaluation of prognostic error model for VIIRS AOD

• Construction of higher-resolution products

The NRL-UND MODIS Level 3 Aerosol Optical Depth Product

14 April 2011 Hyer ISRSE34 Sydney 23 of 25

• 0.5 degree, 6-hourly, available within 4 hours of overpass, HDF format with full metadata

• Available now: nrt1.modaps.nascom.nasa.gov

Thank you

• Thanks to:– NPP VIIRS Cal/Val teams– Atmospheres PEATE (U. Wisconsin)– NRL Aerosol and Radiation Section– Jianglong Zhang and group at UND

• Sponsors:– NOAA (NPP Cal/Val)– ONR (AOD Preprocessing and QA/QC)– NASA AQAST (transition of DA-ready AOD to LANCE)