Introduction

The Ocean Biology Processing Group (OBPG) completed a full-mission ocean color reprocessing of the Suomi-NPP Visible and Infrared Imager/Radiometer Suite (VIIRS) dataset in October 2014. This reprocessing is part of a multi-mission effort to update common algorithms, product suites, and data formats across all supported missions. Sensor-independent changes are detailed in the R2014.0 Ocean Color Reprocessing General Description. Here we describe the VIIRS sensor-specific details of the reprocessing, and provide an assessment of data quality and impact relative to the previous VIIRS reprocessing version R2013.1 VIIRS reprocessing.

Sensor-specific Processing Details

Source Data

As in previous reprocessings starting with R2012.2, this R2014.0 reprocessing uses the pseudo Level-1A files as derived by processing the standard (Level-0) RDRs with a static calibration to produce a set of partially calibrated Level-1 files in standard SDR format. See the R2012.2 documentation for details and file naming conventions. The Level-1A source files are unchanged for this reprocessing.

Instrument Calibration

In Level-0 to pseudo Level-1A processing, a calibration is applied based on results from the prelaunch characterization (e.g. spectral response, polarization sensitivity, response versus scan angle, etc.). The prelaunch radiometric calibration is assumed to be a linear function of the measured counts after dark current subtraction. In Level-1A to Level-2 processing, the prelaunch-calibrated radiances are multiplied by trending coefficients that track the on-orbit change of the radiometric gains. In the previous OBPG reprocessings of VIIRS, the temporal calibration was based entirely on measurements of the Sun through the solar diffuser (Eplee et al. 2013). These solar-based temporal trends were compared to the less frequent observations of the moon, but not modified. For the current reprocessing, this baseline VIIRS calibration approach was updated as follows.

Solar Calibration

It was discovered earlier in 2014 that the solar unit vectors used for processing the solar diffuser (SD) and solar diffuser stability monitor (SDSM) data were reported in the inertial true-of-date reference frame rather then the expected J2000 reference frame. Corrections to the solar unit vectors have been developed by the OBPG and the VIIRS Calibration Support Team (VCST), and VCST has generated a revised set of SD and SDSM attenuation screen transmission functions using these corrected unit vectors. The SD and SDSM data have been reprocessed using the corrected unit vectors and the updated screen transmission functions. The periodic residuals previously observed in the solar calibration time series no longer appear in the reprocessed time series.

For the solar calibration data, SDSM measurements are used to track changes in the bidirectional reflectance distribution function (BRDF) of the solar diffuser. The SDSM has 8 channels, with channels 1-7 corresponding in wavelength to VIIRS bands M1-M7 and with channel 8 at 935 nm (a wavelength where the diffuser BRDF is assumed to be unchanging) used to track unmodeled behavior in the SDSM measurements. The SDSM time series for channel 8 (C8) shows a trend with time, so to correct for this trend, the OBPG has fit the C8 data with an exponential function of time and a linear function of the solar beta angle. The exponential function is used to detrend the C8 data. The detrended C8 time series is then used to normalize SDSM channels 1-7, correcting for measurement noise and residual beta angle effects. The corrected SDSM measurements are interpolated to the sampling interval of the SD observations, and are then used to correct the SD measurements for the changing diffuser BRDF.

Lunar Calibration

For the lunar calibration data, the rapid degradation of the VIIRS primary mirror due to the Near Infrared Degradation Anomaly requires the use of time-dependent relative spectral response (RSR) functions for bands M1-M7 to properly track the radiometric response of VIIRS over time. The OBPG uses time-dependent RSR functions provided by VCST. The spectral shape of the lunar irradiance is redder than that of sunlight reflected by the solar diffuser or by the Earth, so the impact of the time-varying RSR affects the lunar data more so than the solar or Earth data. Since band M1 has a significant red light leak, the primary impact of the RSR is on the band M1 lunar data.

For the lunar calibration data, the disk-integrated lunar irradiances for each observation are processed through the USGS ROLO photometric model of the Moon to correct for changes in the lunar observing geometry. An additional year of lunar observations has allowed the OBPG to confirm that the periodic residuals observed in the output of the ROLO model arise from wavelength-dependent libration effects for the sub-spacecraft point on the lunar surface. The OBPG computes corrections for the librations of the sub-spacecraft longitude and latitude angles.

Merged Solar and Lunar Calibration

Since VIIRS observes the solar diffuser and the Moon at the same angle of incidence on the half-angle mirror, the OBPG uses a direct comparison of the solar and lunar calibrations to minimize the uncertainties in the derived instrument response over time. The solar calibration data track the instrument response each orbit on a per detector, per band, and per mirror side basis, but are impacted by the degradation of the diffuser BRDF. The lunar calibration data track the instrument response monthly, on a per band, per mirror side basis. Comparison of the solar and lunar time series time series on a point-by-point basis (using the solar calibration nearest to a given lunar calibration) shows statistically significant trends in the solar data relative to the lunar data for bands M1, M3, and M4. Accordingly, the OBPG has derived a correction for the solar trends in these bands. The calibration that is applied to the ocean data for the current reprocessing is derived from exponential plus linear functions of time fit to the solar time series, that have lunar-derived adjustments applied to bands M1, M3, and M4.

Striping Correction

The lunar and solar diffuser calibration methods described above resulted in images with obvious striping issues, particularly in the shorter wavelength bands. The OBPG derived a correction that adjusts the gain coefficients of the 16 VIIRS detectors relative to each other. The method was originally developed for MODIS and is described in Meister et al. 2009. From a global L2 data set of one day, detector runs are identified where all 16 detectors simultaneously measured a homogeneous ocean scene with minimal variations in atmospheric radiances. The measured parameters (water-leaving radiance and aerosol properties) are averaged for each run, and then propagated to the top-of-the-atmosphere radiance. The modeled radiance is ratioed to the measured radiance for each detector, and a global average of the ratios results in the gain corrections that are applied in this processing. This analysis is repeated for every month of the mission. A comparison of the striping in the previous version (plot below, left) and the new version (plot below, right) shows that for the global average, striping in the chlorophyll-a product has been reduced significantly:

R2013.2	R2014.0

An evaluation of selected scenes confirmed this finding. The sample scene below was collected over the U.S. West Coast on October 4, 2013. The top two images show chlorophyll concentration before and after the reprocessing; the bottom two images show the before/after difference (left) and ratio (right) of chlorophyll concentrations. (The full range of values in the sample images below is available as a set of histograms.)

Version 2013.1.1 chlorophyll concentration

Version 2014.0 chlorophyll concentration

Color scale for the 2 chlorophyll images

Version 2014.0 minus Version 2013.1.1

Version 2014.0 divided by Version 2013.1.1

Note that the current destriping coefficients are our current best effort and may be refined in the future. In order to further reduce striping, future reprocessings may involve the use of our cross-calibration approach, which provides more flexibility regarding a potential scan angle or polarization sensitivity dependence.

Vicarious_Calibration

The OBPG applies an additional vicarious calibration to VIIRS during 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, measurements from the Marine Optical Buoy (MOBY) near Lanai Hawaii (the same reference 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	0.9618	1.0028	1.0071	0.9751	1.0190	1.0422	1.0	1.0	1.0	1.0

Level-2 Processing

The algorithms employed and products produced from VIIRS are as described in the R2014.0 Ocean Color Reprocessing General Description. Sensor-specific center wavelength values used in processing and product naming are as shown in the table above.

Impact and Quality Assessment

Impact of Reprocessing on Timeseries

To assess the impact of this R2014.0 reprocessing relative to the previous R2013.1 reprocessing, a comparative timeseries analysis was performed (see Franz 2009 for details on the evaluation approach). The impact of all calibration, algorithm, and ancillary data changes on remote sensing reflectance is shown in this comparison of the weekly mean Rrs for various globally-distributed geographic subsets. The updated instrument calibration significantly reduces positive trends in the blue spectral regime and lowers trends in the green, relative to previous reprocessing. Impact on the chlorophyll time-series is to remove a significant negative trend. The end result is a substantial improvement in interannual consistency of the chlorophyll trends (and the Rrs trends). See for example the temporal anomaly (relative to 2013 annual cycle) in clear-water chlorophyll before and after the 2014.0 reprocessing.

R2013.2	R2014.0

There is also a bias shift (increase) in the clear-water chlorophyll of approximately 0.01 mg/m3 on average, which is primarily an effect of the new chlorophyll algorithm employed.

Comparison with In Situ Measurements

Validation of the remote sensing reflectance retrievals was performed relative to all available match-ups from the Aerosol Robotic Network - Ocean Color (AERONET-OC), where only AERONET-OC data of quality level 2 were considered. Statistical analysis of the satellite to in situ match-ups as of this writing is provided below. Also shown is are scatter plots and frequency distribution comparisons. Results show a general negative bias in the Rrs retrievals relative to AERONET-OC, on the order of 10-20%, consistent with previous reprocessings.

Validation of the chlorophyll retrievals was performed relative to all available match-ups in SeaBASS. The number of match-ups is still quite small at 21, so it's dificult to draw any conclusions from this analysis. Validation data for all other standard VIIRS products are too few to make any assessment.

Update R2014.0.1

After the initial R2014.0 reprocessing a VIIRS, and error was discovered in the implementation of the OCI chlorophyll algorithm. Analysis of the impact prompted further refinement of the algorithm and quality flags. Reprocessing R2014.0.1 reflects these changes, and incorporates other minor refinements in processing software that will be reflected in the 2014.0 reprocessing of other sensors. This update also required a revised vicarious calibration, as shown here:

Wavelength (nm)	410	443	486	551	671	745	862	1238	1601	2257
Gain	0.9631	1.0043	1.0085	0.9765	1.0204	1.0434	1.0	1.0	1.0	1.0

The impact of these changes, relative to the original 2014.0 reprocessing of VIIRS, are provided in the comparative time-series analysis.

Update R2014.0.2

Following redevelopment of the VIIRS Level-0 to Level-1B processing software and formats by NASA, the full mission was once again reprocessed (in April 2016) to utilize the newly-developed 6-min Level-1A granule format and produce 6-min Level-2 files. This reprocessing did not involve any calibration or algorithm changes, and derived Level-3 products are identical within processing precision.

References

Gerhard Meister, Bryan A. Franz, Ewa J. Kwiatkowska, Robert E. Eplee, Jr., and Charles R. McClain (2009). Detector dependency of MODIS Polarization Sensitivity derived from on-orbit Characterization. Proceedings of SPIE, 7452, 7452-22.

E.J. Kwiatkowska, B.A. Franz, G. Meister, C.R. McClain, X. Xiong (2008). Cross-Calibration of ocean color bands from Moderate Resolution Imaging Spectroradiometer on Terra platform. Applied Optics, 47: (36) 6796-6810

Eplee, R., K. R. Turpie, G. Meister, F. S. Patt, G.F. Fireman, B. A. Franz, and C. R. McClain (2013). A synthesis of VIIRS solar and lunar calibrations, Proc. SPIE 8866, Earth Observing Systems XVIII, 88661L (October 15, 2013); doi:10.1117/12.2024069.

Franz, B. A., S. W. Bailey, P. J. Werdell and C. R. McClain (2007). Sensor-independent Approach to the Vicarious Calibration of Satellite Ocean Color Radiometry. Applied Optics, 46: (22) 5068-5082.

Franz, B. A. (2009). Methods for Assessing the Quality and Consistency of Ocean Color Products.