VIIRS Ocean Color Reprocessing 2018.0
The Ocean Biology Processing Group (OBPG) intiated a full-mission ocean color reprocessing of the Suomi-NPP Visible and Infrared Imager/Radiometer Suite (VIIRS) dataset in November 2017. This reprocessing is part of a multi-mission effort to update the instrument calibrations and vicarious calibration. Sensor-independent changes are detailed in the R2018.0 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 R2014 VIIRS reprocessing.
Sensor-specific Processing Details
The Level-1A data was regenerated to resolve some minor issues with missing packets and incomplete meta-data.
For this reprocessing, there were two major changes to instrument calibration: 1) the relative temporal calibration was updated to take advantage of additional solar and lunar measurements since the R2014 reprocessing, and to reflect additional advancements in the interpretation and modeling of those observations as described below, and 2) the absolute calibration of all spectral bands was updated based on observations of the sun through the solar diffuser. Prior to this reprocessing, the absolute calibration employed for ocean color processing relied on the prelaunch absolute calibration alone, with solar and lunar measurements employed only for the relative change from first observation. Due to the rapid degradation of the near-infrared (NIR) and short-wave infrared (SWIR) channels on SNPP-VIIRS at the start of the mission, this update to the absolute calibration is needed to resolve changes between the prelaunch calibration and the first solar calibration measurement.
For the on-orbit calibration of SNPP VIIRS, the two calibration sources are the solar diffuser and the Moon. The solar diffuser is the initial calibration source, with lunar observations serving as the long-term radiometric reference. Solar diffuser (SD) observations, obtained once per orbit, yield radiometric gains over time (F-factors) on a per band, per detector, per gain, and per mirror side basis. Figures 1 and 2 show the solar time series for bands M1-M7 through June 2017. The primary source of uncertainty in the solar-derived radiometric gains is in the ability of the SD stability monitor to track and correct for degradation of the solar diffuser reflectance. Lunar observations, obtained on a monthly basis, yield radiometric gains on a band-averaged basis for high gain and both mirror sides. Figures 3 and 4 show the lunar time series for bands M1-M7 through June 2017. The primary source of uncertainty in the lunar calibration time series is residual libration effects, which must be corrected before any residual radiometric trends can be evaluated.
|Figure 1||Figure 2|
|Figure 3||Figure 4|
For ocean color reprocessing, our approach to on-orbit calibration of SNPP VIIRS has been to use both the solar and lunar calibration time series to achieve an on-orbit calibration of the instrument that optimizes the stability of top-of-the-atmosphere radiances. In Reprocessing 2014 the OBPG performed a direct comparison of the solar calibration time series and the libration-corrected lunar calibration time series, computing a pointwise difference in the two trends for each band (M1-M7). Fits to the trend differences were used to adjust the solar radiometric trends. The F-factors used for Reprocessing 2014 were derived from fits to the lunar-adjusted solar radiometric gains. Uncertainties in the libration corrections of the lunar data and scatter in the lunar observations themselves directly contributed to uncertainties in the solar and lunar trend comparisons and thus to uncertainties in the lunar adjustments to the solar radiometric trends.
In Reprocessing 2014, the lunar observations did not have the solar-radiometric trends applied during the calibration of the lunar images. This approach made the implicit assumption that the radiometric responses for all the detectors in a given band decay at approximately the same rate. This assumption turns out to be incorrect. Consequently, there is striping the the lunar images and an increased scatter in the lunar time series. To address this issue for Reprocessing 2018, the lunar images were calibrated on a per-detector basis using the solar-derived radiometric gains. The calibrated, libration-corrected lunar time series for bands M1-M7 are shown in Figures 5 and 6. Deviations from 1.0 represent the lunar correction to the solar-derived temporal calibration.
|Figure 5||Figure 6|
Model fits to the radiometric trends over time in the calibrated, libration-corrected lunar time series are used to adjust the solar radiometric trends. Figure 7 shows the derived lunar adjustment for band M1, as developed for Reprocessing 2014 (green) and Reoprocessing 2018 (blue and red to show each mirror-side). The direct comparison of the solar and lunar time seriesThe previous approach resulted in an over correction compared to the revised analysis of the calibrated time seriesfor this Reprocessing.
In general, the lunar adjustments derived for R2018 are smaller than those derived for R2014. For R2014 lunar-derived adjustments were applied to all of bands M1-M7, though the adjustments to bands M5 through M7 were small. For Reprocessing 2017, lunar adjustments are applied to bands M1-M4 alone, as the adjustments bands M5 through M7 are not statistically significant.
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 based on prelaunch measuremnts and absolute calibration to the solar diffuser. Band M6 (748nm) is then 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).
A major change in this reprocessing, as detailed above, is the first use of absolute calibration to the Sun, which effectively replaces the prelaunch absolute calibration in all spectral bands. To first order, this change in absolute instrument calibration will be directly reflected as an equivalent but opposite change in the vicarious calibration of the visible bands. A secondary impact is that it resolves any changes in the calibration of the 862nm band (and SWIR bands) between prelaunch chracterization and first solar calibration.
Another major change in this reprocessing is to incorporate the updated MOBY time-series, which was reprocessed by the MOBY Operations Team (MOT) to incorporate advancements in instrument calibration and characterization as discussed in the R2018 General Description.
The updated vicarious gains for this reprocessing are reported in the table below.
The product algorithms, product suites, and data formats are identical to those employed in the R2014.0 viirs reprocessing.
Impact and Quality Assessment
Impact of Reprocessing on Timeseries
To assess the impact of this R2018.0 reprocessing relative to the previous R2014 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 mean Rrs for various globally-distributed geographic subsets. In these plots, VT66 is the R2018 reprocessing configuration, and VT59 is the R2014 configuration. On average, the combination of revised MOBY time-series and revised absolute instrument calibration contributes to a significant bias shift in all spectral bands, of order 2%-5% on the global deep-water mean (largest in blue), with resulting decrease in chlorophyll of order 10%. See summary tables here. The updated instrument calibration also significantly reduces a suspect negative trend in the 443nm reflectance, relative to previous reprocessing, with associated impact to blue-water chlorophyll trends.
Comparison with In Situ Measurements
Validation of the remote sensing reflectance (Rrs) and chlorophyll retrievals was performed relative to all available in situ match-ups from SeaBASS and the Aerosol Robotic Network - Ocean Color (AERONET-OC), where only AERONET-OC Rrs data of quality level 2 were considered. Statistical analysis and scatter plots of the satellite to in situ match-ups are provided below, with results from R2014 also shown for reference. As can be seen, Rrs retrieval accuracy is significantly improved in the shortest wavelengths, with a notable reduction in bias and negative radiances, which can be attributed primarily to the refined MOBY data used in the vicarious calibration. Chlorophyll retrievals are also improved, with median absolute percent differences reduced from 44% to 30% and median ratio reduced from 1.4 to 1.1.
Rrs Validation Statistics for R2014 Reprocessing
Rrs Validation Statistics for R2018 Reprocessing
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. (2009). Methods for Assessing the Quality and Consistency of Ocean Color Products.