Xuexia Chen

Improving the characterization and performance of the Suomi-NPP VIIRS solar diffuser stability monitor

The VIIRS instrument on Suomi-NPP performs its primary radiometric calibration using the Solar Diffuser, which degrades with exposure to UV light. The Solar Diffuser is monitored by the Solar Diffuser Stability Monitor. In this paper, we evaluate potential improvements to the algorithms that generate the resulting H-factors, including updates to the screen transmission functions and new methodologies to increase the amount of useful data. We also track the on-orbit degradation of the SDSM detectors and predict the long-term performance of the sensors.

Determination of the SNPP VIIRS SDSM Screen Relative Transmittance From Both Yaw Maneuver and Regular On-Orbit Data

The Visible Infrared Imaging Radiometer Suite aboard the Suomi National Polar-orbiting Partnership (SNPP) satellite performs radiometric calibration of its reflective solar bands primarily through observing a sunlit onboard solar diffuser (SD). The SD bidirectional reflectance distribution function (BRDF) degradation factor is determined by an onboard SD stability monitor (SDSM), which observes the Sun through a pinhole screen and the sunlit SD. The transmittance of the SDSM pinhole screen over a range of solar angles was determined prelaunch and used initially to determine the BRDF degradation factor. The degradation-factor-versus-time curves were found to have a number of very large unphysical undulations likely due to the inaccuracy in the prelaunch determined SDSM screen transmittance. To refine the SDSM screen transmittance, satellite yaw maneuvers were carried out. With the SDSM screen relative transmittance determined from the yaw maneuver data, the computed BRDF degradation factor curves still have large unphysical ripples, indicating that the projected solar horizontal angular step size in the yaw maneuver data is too large to resolve the transmittance at a fine angular scale. We develop a methodology to use both the yaw maneuver and a small portion of regular on-orbit data to determine the SDSM screen relative transmittance at a fine angular scale. We determine that the error standard deviation of the calculated relative transmittance ranges from 0.00030 (672 nm) to 0.00092 (926 nm). With the newly determined SDSM screen relative transmittance, the computed BRDF degradation factor behaves much more smoothly over time.

Determination of the SNPP VIIRS SDSM screen transmittance from both yaw maneuver and regular on-orbit data

The Visible Infrared Imaging Radiometer Suite (VIIRS) aboard the Suomi National Polar-orbiting Partnership (SNPP) satellite carries out radiometric calibration of its reflective solar bands primarily through observing a sunlit onboard solar diffuser (SD). The SD bidirectional reflectance distribution function (BRDF) degrades over time. The degradation factor is determined by an onboard solar diffuser stability monitor (SDSM) which observes the Sun through a pinhole screen and the sunlit SD. The transmittance of the SDSM pinhole screen over a range of solar angles was determined prelaunch and used initially to determine the BRDF degradation factor. The degradation factor versus time curves were found to have a number of very large unphysical undulations likely due to the inaccuracies in the prelaunch determined SDSM screen transmittance. To validate and if necessary to refine both the SD and the SDSM screen transmittances, satellite yaw maneuvers were carried out. With the yaw maneuver data determined SDSM screen transmittance, the computed BRDF degradation factor curves still have large unphysical ripples, indicating that the projected solar horizontal angular step size in the yaw maneuver data is too large to resolve the transmittance at a fine angular scale. We develop a methodology to use both the yaw maneuver and regular on-orbit data to determine the SDSM screen transmittance at a fine angular scale with a relative error standard deviation from 0.00029 (672 nm; detector 5) to 0.00074 (926 nm; detector 8). With the newly determined SDSM screen transmittance, the computed BRDF degradation factor behaves much more smoothly over time.

Suomi-NPP VIIRS Day-Night Band On-Orbit Calibration and Performance

Abstract. The Suomi national polar-orbiting partnership Visible Infrared Imaging Radiometer Suite (VIIRS) instrument has successfully operated since its launch in October 2011. The VIIRS day–night band (DNB) is a panchromatic channel covering wavelengths from 0.5 to 0.9  μm that is capable of observing Earth scenes during both daytime and nighttime at a spatial resolution of 750 m. To cover the large dynamic range, the DNB operates at low-, middle-, and high-gain stages, and it uses an on-board solar diffuser (SD) for its low-gain stage calibration. The SD observations also provide a means to compute the gain ratios of low-to-middle and middle-to-high gain stages. This paper describes the DNB on-orbit calibration methodology used by the VIIRS characterization support team in supporting the NASA Earth science community with consistent VIIRS sensor data records made available by the land science investigator-led processing systems. It provides an assessment and update of the DNB on-orbit performance, including the SD degradation in the DNB spectral range, detector gain and gain ratio trending, and stray-light contamination and its correction. Also presented in this paper are performance validations based on Earth scenes and lunar observations, and comparisons to the calibration methodology used by the operational interface data processing segment.

Using Ground Targets to Validate S-NPP VIIRS Day-Night Band Calibration

In this study, the observations from S-NPP VIIRS Day-Night band (DNB) and Moderate resolution bands (M bands) of Libya 4 and Dome C over the first four years of the mission are used to assess the DNB low gain calibration stability. The Sensor Data Records produced by NASA Land Product Evaluation and Algorithm Testing Element (PEATE) are acquired from nearly nadir overpasses for Libya 4 desert and Dome C snow surfaces. A kernel-driven bidirectional reflectance distribution function (BRDF) correction model is used for both Libya 4 and Dome C sites to correct the surface BRDF influence. At both sites, the simulated top-of-atmosphere (TOA) DNB reflectances based on SCIAMACHY spectral data are compared with Land PEATE TOA reflectances based on modulated Relative Spectral Response (RSR). In the Libya 4 site, the results indicate a decrease of 1.03% in Land PEATE TOA reflectance and a decrease of 1.01% in SCIAMACHY derived TOA reflectance over the period from April 2012 to January 2016. In the Dome C site, the decreases are 0.29% and 0.14%, respectively. The consistency between SCIAMACHY and Land PEATE data trends is good. The small difference between SCIAMACHY and Land PEATE derived TOA reflectances could be caused by changes in the surface targets, atmosphere status, and on-orbit calibration. The reflectances and radiances of Land PEATE DNB are also compared with matching M bands and the integral M bands based on M4, M5, and M7. The fitting trends of the DNB to integral M bands ratios indicate a 0.75% decrease at the Libya 4 site and a 1.89% decrease at the Dome C site. Part of the difference is due to an insufficient number of sampled bands available within the DNB wavelength range. The above results indicate that the Land PEATE VIIRS DNB product is accurate and stable. The methods used in this study can be used on other satellite instruments to provide quantitative assessments for calibration stability.

Validation of S-NPP VIIRS Day-Night Band and M Bands Performance Using Ground Reference Targets of Libya 4 and Dome C

This paper provides methodologies developed and implemented by the NASA VIIRS Calibration Support Team (VCST) to validate the S-NPP VIIRS Day-Night band (DNB) and M bands calibration performance. The Sensor Data Records produced by the Interface Data Processing Segment (IDPS) and NASA Land Product Evaluation and Algorithm Testing Element (PEATE) are acquired nearly nadir overpass for Libya 4 desert and Dome C snow surfaces. In the past 3.5 years, the modulated relative spectral responses (RSR) change with time and lead to 3.8% increase on the DNB sensed solar irradiance and 0.1% or less increases on the M4-M7 bands. After excluding data before April 5th, 2013, IDPS DNB radiance and reflectance data are consistent with Land PEATE data with 0.6% or less difference for Libya 4 site and 2% or less difference for Dome C site. These difference are caused by inconsistent LUTs and algorithms used in calibration. In Libya 4 site, the SCIAMACHY spectral and modulated RSR derived top of atmosphere (TOA) reflectance are compared with Land PEATE TOA reflectance and they indicate a decrease of 1.2% and 1.3%, respectively. The radiance of Land PEATE DNB are compared with the simulated radiance from aggregated M bands (M4, M5, and M7). These data trends match well with 2% or less difference for Libya 4 site and 4% or less difference for Dome C. This study demonstrate the consistent quality of DNB and M bands calibration for Land PEATE products during operational period and for IDPS products after April 5th, 2013.

Analysis of S-NPP VIIRS RSB bands detector saturation status and its change with time.

The Visible Infrared Imaging Radiometer Suite (VIIRS) on the S-NPP satellite has been in successful operation for more than six years. One of the key performance parameters of the instrument detectors is the saturation of their digital counts (DN). For VIIRS, when the scene spectral radiance level is high enough, before reaching the digital maximum that can be accommodated by the number of bits, the DN for the detector stops increasing and often decreases with increasing scene radiance. Consequently, for some high scene radiances, the pixel spectral radiance calculated from the DN does not accurately reflect the true radiance. To inform the data product user that the calculated pixel radiance may not be accurate, a quality flag is used to indicate that the pixel may be inaccurate. In the current S-NPP VIIRS L1B product, the DN saturation flag is turned on once the DN exceeds a fixed threshold level. In this study, the VIIRS Reflective Solar Band (RSB) detectors’ true threshold levels are characterized by studying their responses to high radiance scenes. The long-term trending of these true threshold levels for each detector is analyzed to examine whether the threshold levels are time dependent. Some saturation effects that may be amenable to correction are also investigated, resulting in more useful data. The results from this study will improve data quality with more accurate DN saturation flags.

Suomi NPP VIIRS DNB and RSB M bands detector-to-detector and HAM side calibration differences assessment using a homogenous ground target.

Near-nadir observations of the Libya 4 site from the S-NPP VIIRS Day-Night Band (DNB) and Moderate resolution Bands (M bands) are used to assess the detector calibration stability and half-angle mirror (HAM) side differences. Almost seven years of Sensor Data Records products are extracted from the Libya 4 site center over an area of 32×32 pixels. The mean values of the radiance from individual detectors per HAM side are computed separately. The comparison of the normalized radiance between detectors indicates that the detector calibration differences are wavelength dependent and the differences have been slowly increasing with time for short wavelength bands, especially for M1-M4. The maximum annual average differences between DNB detectors are 0.77% in 2017 at HAM-A. For the M bands, the maximum detector differences in 2017 are 1.7% for M1, 1.8% for M2, 1.3% for M3, 1.2% for M4, 0.67% for M5, 0.75% for M7, 0.57% for M8, 13% for M9, 0.63% for M10, and 0.66% for M11. The average HAM side A to B difference in 2017 are 0.00% for DNB, 0.22% for M1, 0.17% for M2, 0.15% for M3, 0.09% for M4, -0.07% for M5, 0.02% for M7, 0.01% for M8, 1.4% for M9, 0.01% for M10, and 0.03% for M11. Results for M6 are not available due to the signal saturation and M9 results are not accurate because of the low reflectance from the desert site and the strong atmospheric absorption in this channel. The results in this study help scientists better understand each detector’s performance and HAM side characteristics. Additionally, they provide evidence and motivation for future VIIRS calibration improvements.

Initial radiometric calibration status and performance of NOAA-20 VIIRS reflective solar bands

The Earth-observing Visible Infrared Imaging Radiometer Suite (VIIRS) on the NOAA-20 satellite (formerly the Joint Polar Satellite System-1) is the follow-on sensor to the early launched VIIRS on the Suomi National Polar-orbiting Partnership (SNPP) satellite. The on-orbit radiometric calibration of its reflective solar bands (RSBs) is regularly performed primarily through observations of an onboard sunlit solar diffuser (SD). The on-orbit change of the SD bidirectional reflectance distribution function (BRDF) value, denoted as the H-factor, is determined by an onboard solar diffuser stability monitor (SDSM). The scene spectral radiance is calculated by a quadratic polynomial of the background subtracted detector digital number for most of the RSBs and a cubic polynomial for the M8-11 bands. A numerical factor, denoted as the F-factor, provides an on-orbit adjustment to the prelaunch polynomial coefficients through observations of the sunlit SD. The accuracy and change in the F-factor directly affect the sensor radiometric performance. The accuracy of the F-factor is proportionally affected by the accuracy in the H-factor. In this paper, we show the time trends of the Hand F-factors and the SDSM detector gain, and also compare the trends with those for the previous VIIRS instrument on the Suomi National Polar-orbiting Partnership satellite. We derive the Earth view signal-to-noise ratio at the typical spectral radiance level and estimate the calibration bias between the two VIIRS instruments through observations of the Moon and pseudo-invariant Earth sites.

Positional dependence of the SNPP VIIRS SD BRDF degradation factor

The Visible Infrared Imaging Radiometer Suite (VIIRS) aboard the Suomi National Polar-orbiting Partnership (SNPP) satellite is a passive scanning radiometer and an imager. The VIIRS regularly performs on-orbit radiometric calibration of its reflective solar bands (RSBs) through observing an onboard sunlit solar diffuser (SD). The reflectance of the SD changes over time and the change is denoted as the SD bidirectional reflectance distribution function degradation factor. The degradation factor, measured by an onboard solar diffuser stability monitor, has been shown to be both incident sunlight and outgoing direction dependent. In this Proceeding, we investigate the factor’s dependence on SD position. We develop a model to relate the SD degradation factor with the amount of solar exposure. We use Earth measurements to evaluate the effectiveness of the model.