The current on-orbit calibration of multispectral sensors, such as MODIS and VIIRS, is a multi-step process involving numerous sophisticated components. In this report, we calls attention to various issues that have presented challenges or caused misunderstandings that are worthy to be addressed in order to make better forward progress. Primary among the issues presented is the prevalent misunderstanding of calibration strategy, its complexity and even its fundamental premise. More specific issues include systematic radiometric drift versus error residual, impact to overall radiometric accuracy by different calibration components, the use and the interpretation of uncertainty specification and the use of different evaluation and monitoring tools. Special emphasis is made for NOAA-20 and future VIIRS and the importance of having a good handle on this collection of issues.
The radiometric stability of selected thermal emissive bands (TEBs) of Terra and Aqua MODIS is evaluated via a direct intersensor comparison with SNPP VIIRS. The analysis adapts a “nadir-only” refinement of the simultaneous nadir overpasses (SNOs) analysis to evaluate radiometric deviations in MODIS Collection 6.0 and 6.1 data releases, the latter having been implemented with a version of crosstalk effect mitigation for Terra MODIS Bands 27−30. The comparison time series shows radiometric deviation in Terra MODIS Band 29 (8.55 mm) up to 6%, or 3 K in brightness temperature (BT), in Collection 6.0, and -1.5%, or -0.7 K in BT, in Collection 6.1. In particular, the safe mode event of February 2016 has significantly worsened the effect. This study includes the inter-TEB comparisons of MODIS bands B31 and B32 with SNPP VIIRS, and also describes a procedure using both radiance-to-radiance comparison and BTdifference time series to estimate the BT-difference uncertainty specifications as well as the radiance-to-radiance time series baselines. The Terra MODIS Band 29 result highlights the continual impact of crosstalk effect on numerous TEBs of MODIS Collection 6.1 that is not fully mitigated, which can further impact cloud masks, sea surface temperatures and other product retrievals.
The radiometric performance of the reflective solar bands (RSBs) of NOAA-20 VIIRS, recently launched on 18 November 2017, is evaluated through an intercomparison with Aqua MODIS in a continuation of previous year’s effort. The analysis adapts a “nadir-only” refinement of the simultaneous nadir overpass (SNO) to generate comparison time series for assessment of the on-orbit calibration of NOAA-20 VIIRS RSBs using the official sensor data records (SDRs). Result shows improved radiometric stability in the past year but upward radiometric drift in Bands M1 and M2 is possibly emerging, indicative of the active solar diffuser effect that is known for SNPP VIIRS. NOAA-20 VIIRS RSBs continues to maintain a 2 to 8% radiometric deficit relative to SNPP VIIRS RSBs.
The on-orbit calibration performance of the Ocean and Land Colour Imager (OLCI) onboard the Sentinel-3A satellite, launched on 16 February 2016, is evaluated via a radiometric intersensor comparison referencing to SNPP VIIRS. Among the 21 OLCI bands, which are reflective solar bands (RSBs), seven of the bands match up spectrally with the seven shortest wavelength SNPP VIIRS bands. The radiometric comparison utilizes a “nadir-only” refinement of the simultaneous nadir overpass (SNO) approach for analysis. The time series result for OLCI bands Oa02, Oa03, Oa08 and Oa17, which well spreads out over the spectral range of OLCI, shows a two-year stability at the level of 0.3% that supports nominally correct on-orbit calibration for Sentinel-3A OLCI. The result for Oa08, Oa09 and Oa10, the three spectrally adjacent bands matching to SMPP VIIRS Band M5, demonstrates the impact of spectral mismatch - different radiometric ratio baselines and seasonally modulating patterns.
The radiometric performance of the reflective solar bands (RSBs) of NOAA-20 VIIRS, recently launched on 18 November 2017, is evaluated through an intercomparison with Aqua MODIS. The analysis adapts a “nadir-only” refinement of the simultaneous nadir overpass (SNO) to generate comparison time series for assessment of the on-orbit calibration of NOAA-20 VIIRS RSBs using the official sensor data records (SDRs). The comparison result reveals an unstable and varying early radiometric performance upward of 5%. SNPP VIIRS, the precursor VIIRS, is also used to generate a comparison time series against Aqua MODIS. The result shows that NOAA-20 VIIRS RSBs have a 2 to 8% radiometric deficit relative to SNPP VIIRS RSBs.
We describe a new variant of the on-orbit calibration methodology for the reflective solar bands (RSBs) of SNPP VIIRS using the entire illumination interval of the solar diffuser (SD), instead of the smaller “sweet spot” within the illumination interval used by the standard procedure, to compute the calibration coefficients, or F-factors. The instrument response from the full-illumination profile per orbit over the whole mission is directly used to carry out a step-by-step fitting and characterization analysis to arrive at the new F-factors. The new F-factor result is compared with that of standard, lunar-based calibration, and Earth-scattered light approach, expectedly demonstrating very good agreement for bands of longer wavelength but also discrepancies for Bands M1 to M4, the four shortest wavelength bands. The difference is attributed to the angular-dependence in the degradation of the SD that is manifested by the different approaches having different angles of incidence of light to the SD. Thus this result demonstrates the inherent systematic and worsening error for all SD-based on-orbit RSB calibration methodologies to mitigate. On its own, the full-profile approach achieves remarkable stability and robustness on the level of 0.1%, making it a very competitive or better alternative to the current methodology.
We update the effort in the radiometric evaluation of the two different versions of the calibrated sensor data records (SDRs) for the SNPP VIIRS reflective solar bands (RSBs) through direct comparison analysis with Aqua MODIS. The two SDR versions of interest are the official version generated by the Interface Data Process Segment (IDPS) system and the independent version calibrated by the NOAA Ocean Color (OC) Team. The key finding is the continual drift in the IDPS-generated radiance data for five short-wavelength RSBs (443 nm to 1238 nm), contrasting the multi-year stable result for the OC version. SNPP VIIRS M1 (410 nm) versus Aqua MODIS B8 (412 nm) continues to show drift for the IDPS-based and the OC-based comparison time series, pointing to the continual and worsening inaccuracy in Aqua MODIS B8 radiometric data that had been established by preceding studies in the previous year. The SNPP VIIRS M7 (862 nm) versus Aqua MODIS B2 (859 nm) comparison exhibits recent discontinuity in both IDPS-based and OC-based time series - the discontinuity for IDPS-generated SNPP VIIRS M7 is estimated at 0.5% upward near early October 2016, and for Aqua MODIS B2 it is 1.7% upward near mid-July 2016. The other important first-time result is the SNPP VIIRS M11 (2257 nm) versus Aqua MODIS B7 (2130 nm) comparison time series using the land-based snowy scenes of Antarctica, which demonstrates multi-year stability of SNPP VIIRS M11 and the agreement between the IDPS and the OC version of the VIIRS SDRs at this spectral range.
We present a new variant to the on-orbit calibration methodology for the reflective solar bands (RSBs) of SNPP VIIRS using solar diffuser (SD). Instead of following the standard method using full solar illumination through SD port that occurs during terminator crossing, we use light scattered off Earth scenes coming through the nadir port as the source of illumination for the SD. We describe the methodology, present the result, and compare with standard result. The scattered light signal for each orbit is summed up and a 16-day average is taken to build up the calibration coefficient trend. The preliminary result for SNPP VIIRS Bands M1 through M7 shows the scatter light-based approach to be viable, with its calibration coefficient result very well match the standard RSB calibration result over the entire mission to date. The preliminary result does show more variation, on the level of 2%. The methodology is applicable to other instruments employing a similar SD-based calibration strategy. Specifically for the twin MODIS in which the on-orbit RSB calibration is effectively identical, this approach is directly applicable as is.
The inter-sensor radiometric comparison of the 2257 nm channel (M11) of SNPP VIIRS and the 2130 nm channel (B7) of Aqua MODIS curiously demonstrates a wildly varying time series, in stark contrast to the very stable results of other band pairs. A link is found between the few statistically robust outcomes in the time series and the land-based snowy scenes, specifically of Antarctica. With a refined procedure including a selection of land-based snowy scenes, we are able to generate a viable comparison time series of radiance of Aqua MODIS B7 over that of SNPP VIIRS M11. The refined time series shows multi-year stability and hovers around 0.39. The connection between the observed behaviors and their non-overlapping relative spectral responses (RSRs) is discussed.
The use of a specially manufactured solar diffuser (SD) is at the heart of the on-orbit calibration of the reflective solar bands (RSBs) for many important satellite sensors. This includes the Visible Infrared Imaging Radiometer Suite (VIIRS) onboard the Suomi National Polar-orbiting Partnership (SNPP) satellite, and the Moderate-resolution Imaging Spectroradiometer (MODIS) onboard the Terra and Aqua satellites. Within the current standard calibration procedure is an implicit assumption of an idealized degradation of SD in which its angular dependence remains the same functional form with the overall degradation level characterized by a single parameter, the SD degradation factor. This permits the measurement of the SD reflectance performance, measured by the SD stability monitor (SDSM) at a given outgoing angle with respect to the SD, to be used as a valid substitute for the SD reflectance performance toward the RSB direction that is at a different outgoing angle. Recent in-depth studies have uncovered evidence to contradict this assumption, and due to this difference in the outgoing angles between the RSBs and SDSM, the RSB calibration coefficients inherit growing bias. In this exposition, we will explicitly show the evolving angular dependence in SD degradation for SNPP VIIRS and Terra/Aqua MODIS. By examining the angular dependence of the available detector response within each calibration event we are able to build a historical trend clearly demonstrating evolving angular dependence. We refer to this phenomenon as the “SD degradation nonuniformity effect”. Our finding lays out a very basic mismatch between the use of the SD and the current official RSB calibration methodology that will be an important issue to be addressed.
The Visible Infrared Imaging Radiometer Suite (VIIRS) in the Suomi National Polar-orbiting Partnership (SNPP) satellite has been on orbit for nearly five years since its launch on 28 October 2011. The NOAA Ocean Color (OC) Team through the investigations of Sun and Wang has recently achieved robust calibration of the VIIRS reflective solar bands (RSBs) and generated its own version of the sensor data records (SDRs) with accuracy sufficient for ocean color applications. For the purpose of making a direct evaluation of the SDR performance, for both the OC version and the official Interface Processing Data Segment (IDPS) version, we utilize an inter-sensor radiometric comparison of SNPP VIIRS against the MODerate-resolution Imaging Spectroradiometer (MODIS) onboard the Aqua satellite for the spectrally matching RSBs. The VIIRS RSBs M1–M8, from 410 to 1238 nm in the spectral range, are tested. Except for the VIIRS M1 versus MODIS Band 8 result, the radiance comparison time series shows that the OC SDRs demonstrate good agreement with Aqua MODIS and overall better results than the IDPS SDRs, such as less variation, no large discrepancy at the beginning of the VIIRS mission, and no long-term drift. The VIIRS M1 versus MODIS Band 8 trend is the lone exception showing a drift in the OC SDR-based trends, but eventually a downward drift of 1% in Aqua MODIS Band 8 is identified to be the cause. It is readily concluded that the inter-comparison result directly demonstrates the OC SDRs to be correct within statistics, especially considering that the ocean color products derived from the OC SDRs have already matured and demonstrated good agreement with in situ data. On the other hand, the IDPS SDR results demonstrably expose the known inherent growing bias in RSB calibration that affects any versions of the SNPP VIIRS SDRs not using the correctly mitigated calibration baseline. The inter-comparison of two moderate resolution sensors is also an exercise in statistics, and we briefly discuss key points of the pixel-based analysis that establishes the precision and the reliability of the result
Bands 13 (667 nm) and 14 (678 nm) of the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Terra and Aqua satellites are built with a time delay and integration (TDI) circuit and a high-low dual-gain setting intended for the monitoring of the ocean environment. An important on-orbit performance issue for the high-gain output of Terra Band 14 and Aqua Band 13 and 14 is that these three bands saturate during standard on-orbit calibrations using the onboard solar diffuser (SD). Consequently, their calibration procedure requires a derivation different from other reflective solar bands (RSB). A high-low scaling-ratio approach is employed for these bands to obtain the high-gain coefficients via the low-gain output 13L and 14L through a calculated constant ratio derived from other sources. Currently, earth view (EV) scenes over the North Atlantic Ocean are used as the primary sources. As it also has been observed that the response versus scan-angle (RVS) exhibits noticeable changes with time for RSB, a time-dependent RVS calibration algorithm has been implemented based on the SD and lunar calibration results. This paper reviews the electronic design and operation of Band 13 and 14, discusses the SD calibration and the saturation issue of the high-gain output, and the time-dependent RVS algorithms. Long-term performance of the calibration coefficient and RVS are presented. The long-term trend shows that the gain-change for Terra MODIS Band 13 and 14, for both high- and lowgain bands, is significant at 6% and 23%, but is more modest for Aqua MODIS Band 13 and 14 at 6% and 2%. The gain ratios 13H/13L and 14H/14L are very stable at the level of 0.01% for both MODIS instruments. The long-term gain difference due to RVS effect is up to 3% for Terra MODIS and about 2% and below for Aqua MODIS.
MODIS has 20 reflective solar bands (RSB) with a total of 330 individual detectors. Currently, there are two nearly identical MODIS instruments operating in space: one on the Terra spacecraft launched in December 1999 and another on the Aqua spacecraft launched in May 2002. MODIS reflective solar bands (RSB) are calibrated on-orbit by a system that consists of a solar diffuser (SD) and a solar diffuser stability monitor (SDSM). Near-monthly lunar observations are also used to track the sensor response versus scan angle (RVS) change on-orbit. On-orbit observations show that the changes in the detector response are wavelength, scan-angle and mirror side dependent. This paper provides a comprehensive assessment of the detector-to-detector calibration differences in the MODIS VIS/NISR spectral bands using the on-board calibrators and earth-view (EV) level 1B (L1B) data products. Different EV targets are analyzed to accommodate the high-gain ocean bands which tend to saturate over land surfaces. The results from this study highlight the necessity of the constant monitoring of the detector-level gain and its change as a function of scan-angle, as implemented in the MODIS Collection 6 (C6) products.
Since launch, Terra and Aqua MODIS have performed more than 12 and 10 years of scientific measurements of the
Earth’s surface. MODIS has 36 spectral bands, among which 20 are Reflective Solar Bands (RSB), covering a spectral
range from 0.41 μm to 2.1 μm. MODIS was developed with stringent requirements for calibration and uncertainty and is
equipped with a set of on-board calibrators (OBC) that facilitate a constant monitoring and update of its on-orbit
calibration coefficients. The RSB are calibrated on-orbit using a Solar Diffuser (SD) and a Solar Diffuser Stability
Monitor (SDSM), with help from the lunar observations via a Space View (SV) port and an onboard Spectroradiometric
Calibration Assembly (SRCA). The algorithms to accurately characterize the sensor’s gain change and the on-orbit
change in the response versus scan-angle (RVS) have been applied to improve the quality of the Earth-view
measurements. Various improvements to the calibration algorithms have been incorporated since launch and the
following paper will discuss the calibration algorithms and enhancements developed for MODIS Collection 6 (C6)
processing. In addition, to supplement the measurements from the on-board calibrators, pseudo-invariant desert targets
are also used to track the on-orbit response change for selective RSB. Discussions of the on-orbit calibration uncertainty
and the Level 1B (L1B) Uncertainty index (UI) product are also included. A comprehensive assessment of the impact on
the L1B product in comparison to Collection 5 (C5) is also discussed. Significant improvements are observed in the case
of VIS bands wherein the long-term bias observed in C5 products is eliminated to provide a more accurate radiometric
product.
This study uses observed brightness temperatures (BT) over clean ocean to track the stability of the onorbit
response versus scan angle (RVS) for the thermal emissive bands (TEB) of the MODIS (the
Moderate Resolution Imaging Spectroradiometer) instrument onboard Aqua and Terra spacecraft. The
stability is examined by tracking the BT difference between those obtained at each scan angle and the
scan angle of the view for the onboard blackbody. Over two thousands granules over the Atlantic Ocean
for Terra and Aqua MODIS are used. Cloudy pixels within each granule are excluded to improve trending
quality and consistency. The BT trends are derived at 13 angles of incidence (AOI) over the entire Aqua
and Terra missions. Results show that the relative changes in BT for the long-wave infrared bands are
within a few tenths of a degree, equivalent to 0.5 to 1.0%, while those for a few middle-wave infrared
bands still show large fluctuations due to their high sensitivity to atmospheric water vapor contents. A
comparison with trends obtained using measured BT for instrument interior cavity shows that both results
are in good agreement. The BT differences between two mirror sides at each scan angle are also derived
to examine the consistency of the RVS results.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.