VIIRS-NPP Initial Processing for Evaluation

VIIRS-NPP Initial Processing for Evaluation

1. Introduction

The Suomi National Polar-orbiting Partnership (NPP) spacecraft successfully launched 27 October 2011 bearing several Earth observing instruments, including the Visible-Infrared Imager Radiometer Suite (VIIRS). VIIRS is being used by NOAA to operationally generate normalized water-leaving radiance, inherent optical properties, and chlorophyll a concentration. However, it has been determined that these operational products would not meet NASA'science objectives because of the limited product set and lack of reprocessing (Turpie et al. 2011). Thus, as an evaluation of whether the VIIRS instrument could support NASA'science objectives, the NASA Ocean Biology Processing Group (OBPG) is applying calibration methods and algorithms based on lessons learned from SeaWiFS and MODIS with the goal of producing the standard suite of NASA ocean color products from VIIRS that are consistent with heritage sensors.

This document is intended explain how the NASA-VIIRS ocean color products were derived, and how they may differ from the standard NOAA operational products. At this early stage of the effort, users should be aware that the products being distributed are likely to change, without notice, as knowledge of the instrument calibration evolves. We are not tracking versions, and we are not at this time claiming that these products are valid for ocean color research. We are simply making the products available for community evaluation.

2. Instrument Calibration

A primary difference between the NOAA-VIIRS ocean color products and the NASA-VIIRS ocean color products is in the instrument calibration: Level-0 (RDR) to Level-1B (SDR) processing. NASA is deriving a continuous temporal calibration based on the on-board calibration measurements for the visible bands (M1-M7, 412-865), and then reprocessing the full mission to produce a continuously calibrated SDR product. The calibration of the NASA-VIIRS ocean color products is based on results from the prelaunch characterization (e.g. spectral response, polarization sensitivity, response versus scan, etc.), and on-orbit temporal calibration (lunar measurements and solar diffuser measurements). Prelaunch characteristics and the solar diffuser screen transmission function were obtained using results from the NASA VIIRS Calibration Support Team (VCST).

The basic approach of the OBPG calibration for VIIRS is a combination of the approaches for SeaWiFS and MODIS. The prelaunch gains (derived from measurements of VIIRS looking into a calibrated spherical integrating sphere (SIS)) are multiplied by trending coefficients that track the on-orbit change of the radiometric gains (SeaWiFS approach). The solar diffuser is currently the main source for gain trending, which is independently validated with lunar measurements. Trending is currently performed over time with respect to the first light measurement, across detectors (detector destriping), and between mirror sides (mirror-side correction). The measurements of a Solar Diffuser Stability Monitor (SDSM) are used to correct for reflectance degradation of the solar diffuser (MODIS approach). The gains are assumed to be a linear function of the measured counts after dark current subtraction. The response versus scan is corrected based on prelaunch measurements. Temperature correction may be added at a later date. The gains are calculated independently for each mirror side, gain state, and detector.

The figure below shows the change in radiometric response for bands M4-M7 (551-865) as derived from the OBPG solar diffuser analysis. The filled circles show the independently-derived changes from analysis of lunar observations, which clearly confirm the large and rapid degradation observed in the longer wavelengths. The rapid degradation is due to tungsten oxide contamination of the primary mirror, when exposed to ultraviolet light. Prior to 2 January 2012, the instrument was stowed several times in an effort to assess this effect, which is the reason for the plateaus in the degradation curve. After 2 January 2012, the instrument was fully powered-up (including thermal bands) and continuously exposed. The degradation after that time is smooth and tractable, but clearly the long-term viability of the instrument for ocean color is a concern, given that radiometric responsivity has dropped more than 30% in the first year.

viirs_figs1.png

To process VIIRS Level-0 (RDR) files to Level-1B (SDR) files, the OBPG uses the calibration described above in conjunction with the ADL code (Algorithm Development Library, provided by the University of Wisconsin, developed by Raytheon). The ADL code follows the official IDPS code used by NOAA to process VIIRS data, and thus the format of the SDRs produced by NASA is identical to the NOAA SDR format. In its current implementation (June 2012), however, ADL cannot apply gain coefficients from a look-up table (LUT) that vary with time: only the last entry into the gain LUT is used for processing. To incorporate a smoothly varying temporal calibration into the RDR to SDR processing, the OBPG is generating a new LUT for each VIIRS RDR granule, based on the granule start time and the continuous calibration derived from the solar diffuser analysis. This granule-specific, ADL-compatible LUT is generated prior to running ADL, and discarded immediately after.

The NASA SDR product differs from the NOAA SDR product in two significant ways. First, the NASA calibration is derived as a continuous degradation from the start of the mission while the NOAA calibration is changed at discrete intervals (currently weekly, but less frequent in early mission). Second, the NOAA calibration includes an additional absolute calibration to the Sun based on the solar diffuser, which effectively supercedes the prelaunch absolute calibration. This last change was done by NOAA in an effort to resolve absolute and spectral-relative calibration errors, which the OBPG is currently resolving through a vicarious calibration that is applied in Level-2 processing. The result is SDRs between NASA and NOAA that are temporally, spectrally, and absolutely inconsistent.

3. Vicarious Calibration

As noted above, the OBPG applies an additional vicarious calibration during SDR to Level-2 processing (Franz et al., 2007). Band M7 (862nm) is assumed to be correctly calibrated from prelaunch measurements. Band M6 (748nm) is adjusted using match-ups from the South Pacific Gyre, to force the aerosol type retrievals to match, on average, the aerosol type observed at the Tahiti AERONET site. The calibration of bands M1-M5 (410nm to 671nm) is then adjusted to produce retrievals that match, on average, the water-leaving radiances calculated from a sea surface reflectance model and a climatology of chlorophyll-a concentration (Werdell et al. 2007). When sufficient match-ups are available, a switch will be made to use measurements from the MOBY buoy at Hawaii (currently used for SeaWiFS and MODIS).

The vicarious calibration gains derived in this manner are:

Wavelength (nm)

410

443

486

551

671

745

862

1238

1601

2257

Gain

0.9767

1.0202

1.0273

0.9936

1.0257

1.047

1.0

1.0

1.0

1.0

It must be emphasized that these gains are only applicable to the NASA SDR products and the standard NASA atmospheric correction algorithm. If NOAA SDRs are processed through the NASA algorithms, it is recommended to set the vicarious calibration to unity.

No attempt has yet been made to calibrate the shortwave infrared (SWIR) bands.

4. Processing Algorithms and Derived Products

For the NASA SDR to Level-2 processing, the OBPG is using the standard NASA atmospheric correction, i.e., the same algorithms and software that are currently used in standard processing of MODIS, SeaWiFS, and other ocean color sensors. The VIIRS relative spectral response functions (fused RSRs) were used to derive the nominal center wavelengths, bandpass-specific quantities, and out-of-band corrections following the same bandpass integration procedures and sources used for all other NASA'supported ocean color sensors, VIIRS-specific Rayleigh and aerosol tables were generated using the same vector radiative transfer code and methods (Ahmad et al., 2010).

The OBPG is currently producing a subset of the standard NASA Level-2 ocean color product suite. This includes the water-leaving "remote sensing" reflectance (Rrs) in each visible spectral band, aerosol optical thickness at 862nm (aot_862), aerosol Angstrom exponent at 443nm relative to 862nm (angstrom), chlorophyll concentration (chlor_a) based on the OC3 algorithm, and marine diffuse attenuation (Kd_490) based on the KD2 algorithm. Note that the chlor_a and Kd_490 algorithms were specifically tuned for the nominal center wavelengths listed above. Expansion of the standard product suite to include photosynthetically available radiation (PAR) and particulate organic and inorganic carbon (poc and pic) is anticipated. Generation of additional evaluation products, e.g., inherent optical properties, is also likely as confidence is gained in the quality of the VIIRS Rrs retrievals.

5. File Formats and Naming Conventions

The NASA VIIRS SDR and Level-2 products are distributed through the Level-1/2 Browser and the via the online the Ocean Color Archive for direct access and bulk download. As previously noted, the SDR products produced by NASA follow the same format as the standard NOAA SDR products. Each file spans 85-seconds of observation time, thus there are over 1000 granules in a day. This granularity may eventually be changed, but the OBPG is currently constrained to match the granularity of the source (RDR) data. Each band within an SDR is stored in a separate HDF5 file, as is the geolocation data. The NASA processing code requires all 16 M-bands and the geolocation file. To simplify the data distribution and ensure that all bands are present, the OBPG is distributing the SDR as a tar file that contains the full suite of bands for the granule (excluding higher resolution I-bands). The tar file mimics the standard naming convention used for other NASA ocean color missions: Vyyyydddhhmmss.L1B_NPP.tar, where yyyy is year, ddd is day, hhh is hour, mm is minute, and ss is second, and the date/time indicates the observation time of the first line of the granule. An example is shown below.

% tar tf V2012145214452.L1B_NPP.tar

  1. README.txt
  2. GMTCO_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  3. SVM01_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  4. SVM02_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  5. SVM03_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  6. SVM04_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  7. SVM05_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  8. SVM06_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  9. SVM07_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  10. SVM08_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  11. SVM09_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  12. SVM10_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  13. SVM11_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  14. SVM12_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  15. SVM13_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  16. SVM14_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  17. SVM15_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5
  18. SVM16_npp_d20120524_t2144540_e2146182_b02972_obpg_ops.h5

The Level-2 files produced by NASA follow the same general naming convention (e.g., V2012145214452.L2_NPP_OC), and the format is the same HDF4 format used for all other NASA ocean color sensor products. The OBPG is also distributing VIIRS Level-3 products derived from the NASA Level-2 processing. The Level-3 products are binned to 4.6-km spatial resolution (identical to MODIS), composited to daily, 8-day, monthly, and seasonal products, and distributed in the same binned and mapped HDF4 formats as other NASA'sensors. The Level-3 binned and mapped products are available from the evaluation tab of the Level-3 Browser and directly from the Ocean Color Archive.

NOTE: To support evaluation activities by the NASA NPP Science Team, the OBPG is also distributing a set of Level-3 products derived from the standard NOAA Level-2 (EDR) products. These files are clearly identified on the Level-3 browser, and the filenames contain NPPE in place of NPP.

6. Results

Initial results from the NASA processing of NPP VIIRS have been compared against field measurements and the MODIS and SeaWiFS record. In situ match-ups with AERONET-OC water-leaving reflectances (Rrs) show good agreement, with some bias at longer wavelengths. The first figure below shows a comparison of the chlorophyll trend of VIIRS and MODIS-Aqua for global deep water, with the historical range of SeaWiFS chlorophyll seasonal variability shown as the red lines. The second figure shows a comparison of chlorophyll and Kd(490) images from the two sensors for the Winter 2012 season.

viirs_figs3.png viirs_figs5.png

The agreement between the sensors is quite good for this stage of the mission, and it will likely improve as additional vicarious calibration data and instrument radiometric calibration knowledge is accumulated. Note that trending differences relative to MODIS-Aqua may in fact indicate a problem with MODIS, as late-mission temporal trends in MODIS are known to suffer increased uncertainty due to uncharacterized variability in the temporal calibration. In general, however, the plots and images show that MODIS and VIIRS are producing derived geophysical properties that are consistent in geographic distribution with similar dynamic range. More detailed analysis of VIIRS relative to MODIS/Aqua indicates larger discrepancies in the highest and lowest chlorophyll range (VIIRS low chlorophylls are too high and the highs are too low), but again this is likely to change as vicarious calibration is improved.

7. Timeline of Changes

  • January 2012: Production of NASA L2 products from NOAA SDRs begins.
  • 28 March 2012: First full mission L1 reprocessing is started using ADL and NASA calibration LUTs.

  • 3 April 2012: First full mission L2 & L3 reprocessing begins.

  • 8 May 2012: L3 reprocessed to turn off masking based on CHL_WARN flag.

  • 29 May 2012: Second full mission L1 reprocessing is started using updated F-Table based on new screen tau tables, new PolarWander file, 4 Geolocation LUT updates.

  • 6 June 2012: Updated F-Table (covers 2 Jan 2012 - 2 Jun 2012) put in forward stream.

  • 21 June 2012: Updated F-Table (covers 2 Jan 2012 - 20 Jun 2012) with relative detector-to-detector and mirror side corrections. New PolarWander file.

  • 25 June 2012: Third full mission reprocessing.
  • 2 October 2012: Partial reprocessing starting from May, to update calibration

8. Additional Information

  1. Documentation of NOAA operational algorithms and product formats

  2. Access to NOAA operational data distribution (CLASS)