• Baldridge, A. M., , Hook S. J. , , Grove C. I. , , and Rivera G. , 2009: The ASTER spectral library version 2.0. Remote Sens. Environ., 113, 711715, doi:10.1016/j.rse.2008.11.007.

    • Search Google Scholar
    • Export Citation
  • Bréon, F. M., , and Henriot N. , 2006: Spaceborne observations of ocean glint reflectance and modeling of wave slope distributions. J. Geophys. Res., 111, C06005, doi:10.1029/2005JC003343.

    • Search Google Scholar
    • Export Citation
  • Cao, C., , Weinreb M. , , and Sullivan J. , 2001: Solar contamination effects on the infrared channels of AVHRR. J. Geophys. Res., 106, 33 46333 469.

    • Search Google Scholar
    • Export Citation
  • Cao, C., , Xiong J. , , DeLuccia F. , , Liu Q. , , Blonski S. , , Lin G. , , Nishihama M. , , Progorzala D. , , and Oudtari H. , 2012: Visible/Infrared Imager Radiometer Suite (VIIRS) sensor data record (SDR): User's guide, version 1.1. NOAA NESDIS Tech. Rep. 142, 41 pp.

  • Cao, Y., , Yan D. , , Wang G. , , and You S. , 2012: De-striping algorithm in ALOS satellite imagery based on adaptive frequency filter. Image and Signal Processing for Remote Sensing XVIII, L. Bruzzone, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 8537), 853719, doi:10.1117/12.973674.

  • Chen, Y., , Han Y. , , and Weng F. , 2012: Comparison of two transmittance algorithms in the community radiative transfer model: Application to AVHRR. J. Geophys. Res., 117, D06206, doi:10.1029/2011JD016656.

    • Search Google Scholar
    • Export Citation
  • Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sunphotometer measurements. J. Atmos. Sci., 59, 461483.

    • Search Google Scholar
    • Export Citation
  • Corsini, G., , Diani M. , , and Walzel T. , 2000: Striping removal in MOS-B data. IEEE Trans. Geosci. Remote Sens., 38, 10131015.

  • Han, Y., , van Delst P. , , Liu Q. , , Weng F. , , Yan B. , , Treadon R. , , and Derber J. , 2005: JCSDA Community Radiative Transfer Model (CRTM)—Version 1. NOAA NESDIS Tech. Rep. 122, 33 pp.

  • Liang, X., , and Ignatov A. , 2011: Monitoring of IR Clear-Sky Radiances over Oceans for SST (MICROS). J. Atmos. Oceanic Technol., 28, 12281242.

    • Search Google Scholar
    • Export Citation
  • Liang, X., , and Ignatov A. , 2013: AVHRR, MODIS, and VIIRS radiometric stability and consistency in SST bands. J. Geophys. Res., 118, 31613171, doi:10.1002/jgrc.20205.

    • Search Google Scholar
    • Export Citation
  • Liu, Q., , and Weng F. , 2006: Advanced doubling–adding method for radiative transfer in planetary atmospheres. J. Atmos. Sci., 63, 34593465.

    • Search Google Scholar
    • Export Citation
  • Liu, Q., , Weng F. , , and English S. , 2011: An improved fast microwave water emissivity model. IEEE Trans. Geosci. Remote Sens., 49, 12381250.

    • Search Google Scholar
    • Export Citation
  • Liu, Q., , Cao C. , , and Weng F. , 2012: NPP VIIRS emissive band radiance calibration. Earth Observing Systems XVII, X. Gu, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 8510), 85101E, doi:10.1117/12.930201.

  • Miller, S. D., , Lee T. F. , , Fennimore R. L. , 2005: Satellite-based imagery techniques for daytime cloud/snow delineation from MODIS. J. Appl. Meteor.,44, 987–997.

  • NASA, 2011: Joint Polar Satellite System (JPPS) VIIRS geolocation algorithm theoretical basis document (ATBD). NASA JPSS Ground Project Code 474, 474-00053, 144 pp.

  • Poros, D. J., , and Peterson C. J. , 1985: Methods for destriping Landsat Thematic Mapper images—A feasibility study for an online destriping process in the Thematic Mapper Image Processing System (TIPS). Photogramm. Eng. Remote Sens., 51, 13711378.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , Smith T. M. , , Liu C. , , Chelton D. B. , , Casey K. S. , , and Schlax M. G. , 2007: Daily high-resolution blended analyses for sea surface temperature. J. Climate, 20, 54735496.

    • Search Google Scholar
    • Export Citation
  • Sayer, A. M., , Thomas G. E. , , and Grainger R. G. , 2010: A sea surface reflectance model for (A)ATSR, and application to aerosol retrievals. Atmos. Meas. Tech., 3, 813838, doi:10.5194/amt-3-813-2010.

    • Search Google Scholar
    • Export Citation
  • Simmer, C., 1994: Satellite Remote Sensing of Atmospheric Hydrological Parameters Using Microwaves (in German). Studies on Agroecology Series, Vol. 13, Verlag Dr. Kovač, 313 pp.

  • van Delst, P., , and Wu X. , 2000: A high resolution infrared sea surface emissivity database for satellite applications. Proc. 11th Int. TOVS Study Conf., Budapest, Hungary, WMO, 407–411.

  • Vogel, R., , Liu Q. , , Han Y. , , and Weng F. , 2011: Evaluating a satellite-derived global infrared land surface emissivity data set for use in radiative transfer modeling. J. Geophys. Res., 116, D08105, doi:10.1029/2010JD014679.

    • Search Google Scholar
    • Export Citation
  • Yang, P., , Liou K. N. , , and Arnott P. W. , 1997: Extinction efficiency and single-scattering albedo for laboratory and natural cirrus clouds. J. Geophys. Res., 102 (D18), 21 82521 835.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Normalized histogram of the difference between the CRTM simulations and the VIIRS M12 observations over global oceans.

  • View in gallery

    A schematic graph for the VIIRS M band. The azimuth angle is defined as the ground projection (dashed lines) of light ray to the north.

  • View in gallery

    VIIRS M12 brightness temperature image (grayscale for 301–307 K) over a clear-sky oceanic image near Australia at 0615 UTC 25 Jan 2013. The sun-glint angle is less than 10°, and the surface wind speed is higher than 7 m s−1.

  • View in gallery

    VIIRS M12 brightness temperature image (294–299 K) over a clear-sky oceanic image near South America at 1800 UTC 25 Jan 2013. The sun-glint angle is more than 30°, and the sea surface wind speed is about 3.5 m s−1.

  • View in gallery

    Brightness temperature deviation (mean value subtracted) with scans (3 scans × 16 detectors = 48 scan lines). (top left) M12 band BT anomaly, showing VIIRS observation (solid line), CRTM simulation (dotted line), and CRTM simulation without solar radiation (dashed line). (bottom left) Reflectance anomaly in percentage for the VIIRS M1 (dotted line), M4 (dashed–dotted line), M10 (solid line), and M11 (dashed line). (right) Striped M12 band brightness temperature.

  • View in gallery

    The absolute sensor azimuth angle difference between detectors 1 and 16 for the same scan. The total pixels per scan for the VIIRS M band is 3200, and the pixel position at 1600 or 1601 is the closet point to nadir.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 48 48 14
PDF Downloads 36 36 7

Striping in the Suomi NPP VIIRS Thermal Bands through Anisotropic Surface Reflection

View More View Less
  • 1 Joint Center for Satellite Data Assimilation, and Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland
  • 2 NOAA/NESDIS/STAR/Satellite Meteorology and Climatology Division, College Park, Maryland
© Get Permissions
Full access

Abstract

The Visible Infrared Imaging Radiometer Suite (VIIRS) thermal emissive band (TEB) M12 images centered at 3.7 μm were analyzed and unexpected striping was found. The striping was seen from ascending orbit (daytime) over uniform oceans and has a magnitude of ±0.5 K aligned with the VIIRS 16 detectors in a track direction of 12 km. From the ocean surface, reflected solar radiation can significantly increase the M12 radiance under certain geometric conditions in which bidirectional reflectance distribution function (BRDF) becomes important. Using the Community Radiative Transfer Model (CRTM), developed at the U.S. Joint Center for Satellite Data Assimilation (JCSDA), M12 band image striping over a uniform ocean was found that was caused by the difference of sensor azimuthal angles among detectors and the contamination of solar radiation. By analyzing the VIIRS M10 and M11 bands, which are two reflective bands, similar striping images over the uniform oceans were found. The M10 and M11 radiance/reflectance can be used to determine the BRDF effect on the thermal emissive band M12, and eventually be used to remove the solar radiation contamination from the M12 band. This study demonstrated that the M12 image striping is a real instrument artifact. Whether to remove the striping or to utilize the striping information fully depends on the application.

Corresponding author address: Quanhua Liu, Joint Center for Satellite Data Assimilation, 5830 University Research Court, College Park, MD 20740-3822. E-mail: quanhua.liu@noaa.gov

Abstract

The Visible Infrared Imaging Radiometer Suite (VIIRS) thermal emissive band (TEB) M12 images centered at 3.7 μm were analyzed and unexpected striping was found. The striping was seen from ascending orbit (daytime) over uniform oceans and has a magnitude of ±0.5 K aligned with the VIIRS 16 detectors in a track direction of 12 km. From the ocean surface, reflected solar radiation can significantly increase the M12 radiance under certain geometric conditions in which bidirectional reflectance distribution function (BRDF) becomes important. Using the Community Radiative Transfer Model (CRTM), developed at the U.S. Joint Center for Satellite Data Assimilation (JCSDA), M12 band image striping over a uniform ocean was found that was caused by the difference of sensor azimuthal angles among detectors and the contamination of solar radiation. By analyzing the VIIRS M10 and M11 bands, which are two reflective bands, similar striping images over the uniform oceans were found. The M10 and M11 radiance/reflectance can be used to determine the BRDF effect on the thermal emissive band M12, and eventually be used to remove the solar radiation contamination from the M12 band. This study demonstrated that the M12 image striping is a real instrument artifact. Whether to remove the striping or to utilize the striping information fully depends on the application.

Corresponding author address: Quanhua Liu, Joint Center for Satellite Data Assimilation, 5830 University Research Court, College Park, MD 20740-3822. E-mail: quanhua.liu@noaa.gov

1. Introduction

Satellite image striping can be caused by imperfect calibration, different spectral response, and various geometric-looking directions among detectors. The image striping due to the imperfect calibration relates to instrumental design and calibration algorithm, for example, mirror side, gain stage, electronic unit, polarization, and stray light leaking. The striping noise needs to be removed for a quantitative image analysis and satellite products (Cao et al. 2012; Corsini et al. 2000). In this study, we focus on the striping related to the differences in spectral response and geometry among detectors, because such a striping is a real instrument artifact. Both spectral response difference and geometric difference can be considered in radiative transfer calculations. Removing the striping may cause the inconsistency between measurements and radiative transfer calculations, which can be an issue in direct radiance assimilation.

The successful launch of the Suomi National Polar-Orbiting Partnership (SNPP) spacecraft on 28 October 2011 with the Visible Infrared Imaging Radiometer Suite (VIIRS) starts a new generation of capabilities for operational environmental remote sensing for weather, climate, ocean, and other environmental applications. The VIIRS was turned on on 8 November 2011 for measuring reflective band radiance. The VIIRS Cryo-cooler door was opened on 18 January 2012, starting the observations for thermal emissive bands. The VIIRS employs 32 detectors for image (I) bands and 16 detectors for moderate-resolution (M) bands of a moderate resolution. In contrast to conventional imaging radiometers (Miller et al. 2005), which feature a single detector scanning each line of the scene, a single physical scan for the VIIRS M bands (16 detectors aligned in the along-orbit-track direction) results in the imaging of a 16 × 0.75 (=12) km cross-track strip. This allows for a slower scan rate than that of a single detector and, therefore, enhances the spatial resolution without losing the signal-to-noise ratio (SNR). Unfortunately, two artifacts, “bow tie” deletion and image striping, emerge as a trade-off to this multidetector arrangement that must be dealt with (Miller et al. 2005). The image striping for the Moderate Resolution Imaging Spectroradiometer (MODIS) is an issue for analyzing snow and clouds (Miller et al. 2005). The software MODIS Adaptive Processing System (MODAPS), developed by the Space Science and Engineering Center (SSEC) at the University of Wisconsin–Madison (http://modaps.nascom.nasa.gov/cgi-bin/PCR_detail.cgi?file=/Atmos/PCR06-001.dat), may be used to improve the image quality.

During the VIIRS sensor–intensive calibration, the National Oceanic and Atmospheric Administration (NOAA) sea surface temperature team first found and reported the VIIRS image striping of M12 centered at 3.7 μm. A smoothing software may be used for the VIIRS to improve the image quality (Poros and Peterson 1985). However, the objective of this study is to investigate the root cause of the striping and for quantitative radiance assimilations and satellite product retrievals. The message this study delivers is that we need to distinguish satellite image striping as imperfect calibration from the difference in sensor spectral response and geometry among detectors. The former striping needs to be removed, and the latter has to be handled in radiative transfer calculations for satellite radiance assimilation. It depends on the application whether the striping has to be removed. If image quality matters—for example, for nowcasting applications or in the case of deriving polar atmospheric motion vectors—then the removal of any striping is desired. In section 2, the VIIRS data are introduced. Section 3 describes the Community Radiative Transfer Model. Section 4 analyzed the VIIRS images and detailed the image striping and radiative transfer calculations. Conclusions and discussion are given in section 5.

2. VIIRS data

VIIRS (Cao et al. 2012) is designed to provide moderate-resolution, radiometrically accurate images of the globe twice daily. It is a wide-swath (3040 km) scanning radiometer with spatial resolutions of 0.375 and 0.75 km at nadir for I and M bands, respectively. It has 22 spectral bands in total covering the spectrum between 0.412 and 12.01 μm, including 14 reflective solar bands (RSB), 7 thermal emissive bands, and 1 day–night band (DNB). The VIIRS bands are further divided into radiometric bands (M bands) and imaging bands (I bands). The M bands have better signal-to-noise ratio and accuracy, which are better suited for quantitative applications, while the imaging bands have a high spatial resolution with broader spectral response. VIIRS does not have any infrared atmospheric sounding bands equipped with MODIS.

VIIRS uses six dual-gain reflective bands to provide the high radiometric resolution needed for ocean color applications, at the same time without saturating the sensor when observing high reflectance surfaces such as land and clouds. The dynamic range of the dual-gain bands in high gain is comparable to that of the MODIS ocean color bands, while the dynamic range in the low-gain state is comparable to those of the similar MODIS land bands. The dynamic ranges across all other bands are similar to their MODIS counterparts. VIIRS has one dual-gain thermal emissive band (M13) to measure fire temperature at a low gain and normal surface temperature at a high gain.

VIIRS uses a unique approach of “bow-tie deletion” through pixel aggregation (NASA 2011) that controls the pixel oversampling and noise toward the end of the scan edge—a problem that exists for MODIS, the Advanced Very High Resolution Radiometer (AVHRR), and other cross-scan instruments. A feature of the VIIRS flight software that trims the M and I bands in the along-track direction as a function of along-scan aggregation zone will be enabled. This will result in some of the samples in the overlap area being excluded from the data that are delivered to the ground. It may be worth mentioning that the VIIRS sensor data record (level 1 product) uses a fill value for the bow-tie deletion pixel. The VIIRS spatial resolutions for nadir and edge-of-scan data are comparable. The VIIRS spatial resolution is 0.375 and 0.75 km at nadir for imaginary and moderate bands, respectively, and 0.8 and 1.6 km at the scan edge for imaginary and moderate bands, respectively. This does introduce visual artifacts in the raw image due to the aggregation and removal of duplicated pixels beyond midscan on each side. The artifact can be removed when the image is displayed through interpolation.

Reflected and emitted radiation from the earth enters the sensor through the rotating telescope assembly (RTA) and is reflected from a rotating half-angle mirror (HAM) into a stationary aft-optics subsystem. The light is then spectrally and spatially separated by dichroic beam splitters and directed to three separate focal plane arrays (FPAs): the visible/near-infrared (Vis/NIR) FPA, the shortwave/midwave infrared (SW/MWIR) FPA, and the longwave infrared (LWIR) FPA. The light is detected and converted to analog electrical signals in these FPAs and further processed prior to analog-to-digital (A/D) conversion with 12-bit quantization. The digital signals are then processed and multiplexed into the instrument output data stream. Housekeeping data in the form of instrument health, safety, and engineering telemetry are also generated from measurements of internal temperatures, voltages, and currents. These telemetry measurements are reported for every scan.

The panchromatic DNB band measures night lights, reflected solar, and/or moon lights with a large dynamic range of 45 000 000:1, which allows the detection of reflected signals from as low as quarter-moon illumination to the brightest daylight. To achieve this large dynamic range, it uses a three-stage focal plane. The sensor maintains a nearly constant 0.75-km resolution over the entire 3000-km swath using an onboard aggregation scheme.

This study concentrated on VIIRS thermal emissive band calibration and data validation (Liu et al. 2012). The VIIRS thermal emissive band has two image emissive bands and five moderate-resolution bands. The seven emissive bands are centered at 3.74, 11.45, 3.75, 4.05, 8.55, 10.76, and 12.01 μm. The two emissive image bands are mainly for cloud imagery and precise geolocation. The five moderate-resolution emissive bands are used to determining surface temperature and cloud-top pressure. The only dual-gain thermal emissive band (TEB) M13 is used for is determining surface temperature at low radiance, and fire detection at high radiance. The VIIRS sensor specification and prelaunch performance are given in Table 1 (Cao et al. 2012). The spectral range for the VIIRS bands is updated using the actual VIIRS spectral response data. The postlaunch performance changes with time because of instrumental degradation—in particular, the VIIRS main mirror set degradation due to tungsten contamination on mirrors.

Table 1.

VIIRS sensor specifications and prelaunch assessment. NEdT is the noise-equivalent delta temperature. EDR is the environmental data record. NDVI is the normalized difference vegetation index.

Table 1.

3. Radiative Transfer Model

The Community Radiative Transfer Model (CRTM; Han et al. 2005; Liu and Weng 2006) provides forward, tangent-linear, adjoint, and K-matrix functions to compute radiance (also microwave and infrared brightness temperatures) and sensitivities of radiance to atmospheric/surface parameters. The CRTM is a sensor-based radiative transfer model. It supports more than 100 sensors, including most meteorological satellites and some remote sensing satellites. For a new sensor, the CRTM team can generate spectral and transmittance coefficient files as long as the sensor response data of the new sensor are available. The CRTM adopts the Optical Path Transmittance (OPTRAN) algorithm for H2O, CO2, O3, N2O, CO, CH4, and the rest are treated as “dry part” (Chen et al. 2012). It contains a precalculated lookup table for the optical parameters of clouds (Simmer 1994; Yang et al. 1997) and aerosols (Chin et al. 2002).

The surface emissivity/reflectance models are divided into water, land, ice, and snow gross types for visible, infrared, and microwave sensors. Each gross type is further divided into subtypes, for example, new and old snows. For the infrared and visible portions of the spectrum, the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) spectral library (Baldridge et al. 2009) data are applied for land infrared emissivity and the surface is assumed as Lambertian, and the emissivity equals one minus reflectivity (Vogel et al. 2011). We also developed utility codes for users to use an emissivity atlas that depends on month, latitude, and longitude. The infrared water emissivity is based on the Wu–Smith model (van Delst and Wu 2000). Microwave water emissivity can be calculated from surface temperature, wind vector, and salinity (Liu et al. 2011).

The infrared water bidirectional reflectance distribution function (BRDF) model is used for the CRTM direct reflectance to compute reflected solar radiation (Sayer et al. 2010). Using UV and visible spectral refractive indices of water, the BRDF model in the CRTM is also used for UV and visible measurements over water. The BRDF model is described by Fresnel reflection coefficients for a rough surface (Bréon and Henriot 2006). The surface roughness has a Gaussian distribution and is a function of wind speed. Sea foam needs to be included for large wind speed. The solar radiation over a sun-glint area may contribute 30 K or more to the infrared channel at 3.7 μm. The direct reflectance of the solar radiation strongly depends on geometry of the sensor and sun (Cao et al. 2001). Therefore, the geometric difference among detectors can affect the reflected solar radiation. We have compared the AVHRR, MODIS, and VIIRS shortwave infrared band brightness temperatures at 3.7 μm and the CRTM simulations with and without the BRDF model. For a sun-glint angle smaller than 15°, the CRTM simulated brightness temperature without considering that the BRDF can be 20 K less than the observations. The difference between the VIIRS observations and the CRTM simulations with the BRDF is about 1 K (see Fig. 1) and is monitored daily at NOAA (Liang and Ignatov 2011). The sun-glint angle θg for a calm ocean surface can be calculated from the following equation:
e1
where θr and ϕr are sensor-viewing zenith and azimuth angles, and θs and ϕs are sun-viewing zenith and azimuth angles. It can be seen from Eq. (1) that the sun-glint angle depends on the relative azimuth angle between the satellite and sun. For many applications, the sun-glint angle is used to exclude the data affected by solar radiation.
Fig. 1.
Fig. 1.

Normalized histogram of the difference between the CRTM simulations and the VIIRS M12 observations over global oceans.

Citation: Journal of Atmospheric and Oceanic Technology 30, 10; 10.1175/JTECH-D-13-00054.1

4. Results

VIIRS uses 16 detectors for each M band and 32 detectors for each I band. For each VIIRS M band, 16 detectors aligned in the track direction and have the same scan angle or same viewing zenith angle. As shown in the schematic Fig. 2, the 16 detectors observe 16 pixels on the ground, respectively. The 16 pixels have the same viewing zenith angle. Azimuth angle α for pixel 1 is obviously different from azimuth angle β for pixel 16, where the azimuth angle is defined as the ground projection (dashed lines) of the light of ray to the north. The sun is far away from the ground pixels and can be treated as parallel light and has an azimuth angle γ for all pixels. Therefore, for each physical scan of the VIIRS, the azimuthal angle difference among detectors results in a different sun-glint angle that changes the solar radiation component.

Fig. 2.
Fig. 2.

A schematic graph for the VIIRS M band. The azimuth angle is defined as the ground projection (dashed lines) of light ray to the north.

Citation: Journal of Atmospheric and Oceanic Technology 30, 10; 10.1175/JTECH-D-13-00054.1

The VIIRS users reported the issue in image quality—in particular, image striping due to different detector characteristics (Cao et al. 2012). We analyzed the VIIRS spectral response for each detector. Although we do not have an identical spectral response for any two detectors, the effect due to the difference in the spectral response is generally small for window bands. For the VIIRS M12, the brightness temperature difference due to the various spectral responses of the detectors is smaller than 0.05 K.

We analyzed the VIIRS M12 daytime images over oceanic areas largely affected by the BRDF and less affected by BRDF individually. We selected a uniform oceanic area based on the VIIRS M2, M4, and M5 true-color image and the M15 brightness temperature image. Figure 3 is the VIIRS M12 brightness temperature image (grayscale for 301–307 K) over a clear-sky oceanic image near Australia at 0615 UTC 25 January 2013. The sun-glint angle is less than 10°, and the surface wind speed is higher than 7 m s−1. The white area there represents strong solar contamination. The string along-track direction displays a periodic change with every 16 detectors. Figure 4 is the VIIRS M12 brightness temperature image (294–299 K) over a clear-sky oceanic image near South America at 1800 UTC 25 January 2013. The sun-glint angle is larger than 30°, and the sea surface wind speed is about 3.5 m s−1. The BRDF effect is very small. The daytime M12 image without significant BRDF effect and the nighttime image do not display the striping. The two images suggest that only the area strongly affected by BRDF may have the image striping.

Fig. 3.
Fig. 3.

VIIRS M12 brightness temperature image (grayscale for 301–307 K) over a clear-sky oceanic image near Australia at 0615 UTC 25 Jan 2013. The sun-glint angle is less than 10°, and the surface wind speed is higher than 7 m s−1.

Citation: Journal of Atmospheric and Oceanic Technology 30, 10; 10.1175/JTECH-D-13-00054.1

Fig. 4.
Fig. 4.

VIIRS M12 brightness temperature image (294–299 K) over a clear-sky oceanic image near South America at 1800 UTC 25 Jan 2013. The sun-glint angle is more than 30°, and the sea surface wind speed is about 3.5 m s−1.

Citation: Journal of Atmospheric and Oceanic Technology 30, 10; 10.1175/JTECH-D-13-00054.1

We also analyzed the M12 striping image reported by the NOAA sea surface temperature team. The image is over a clear-sky uniform ocean near Australia at 0555 UTC 12 March 2012. As shown on the right image of Fig. 5, M12 band brightness temperature displayed the striping every 16 scan pixels corresponding to 16 detectors of the VIIRS M band. We checked the relative spectral response and can exclude the cause as the difference in the sensor response among detectors. There was a discussion about the possibility that solar radiation may be causing the striping, which would be bad if this is the case because such an effect depends on time and location, and is hard to remove. For each orbit at daytime, solar radiation transmits to the VIIRS through the solar diffuser screen and the solar diffuser stability screen for the calibration of the VIIRS reflective spectral bands. The solar radiation could leak onto other parts of the instrument (Liu et al. 2012). Fortunately, the striping was not observed from the daytime image scene, where it was not or less affected by BRDF (see Fig. 4). The result may exclude the root cause because of the leaking of solar radiation onto the sensor.

Fig. 5.
Fig. 5.

Brightness temperature deviation (mean value subtracted) with scans (3 scans × 16 detectors = 48 scan lines). (top left) M12 band BT anomaly, showing VIIRS observation (solid line), CRTM simulation (dotted line), and CRTM simulation without solar radiation (dashed line). (bottom left) Reflectance anomaly in percentage for the VIIRS M1 (dotted line), M4 (dashed–dotted line), M10 (solid line), and M11 (dashed line). (right) Striped M12 band brightness temperature.

Citation: Journal of Atmospheric and Oceanic Technology 30, 10; 10.1175/JTECH-D-13-00054.1

Figure 5 already shows that the brightness temperature is quasi linear with detector numbers, which again excludes the main root cause from spectral response differences among detectors, unless the detectors are specifically aligned for this linear purpose. Using the detector-dependent sensor response function and the CRTM, we can estimate the difference of measurements among detectors. The European Centre for Medium-Range Weather Forecasts' (ECMWF) 6-h forecasting for atmospheric/surface state variables is used. We evaluate the difference caused by detectors. Generally, the brightness temperature difference due to the difference in detector response is less than 0.05 K. In the top-left panel of Fig. 5, we plot the brightness temperatures' deviation from their mean for VIIRS M12 observations (solid line), the CRTM simulations (dashed line), and the CRTM simulations without solar radiation (dotted line). The dotted line shows a very small variation because of a slight difference in the detectors' spectral characteristics. Both the VIIRS measurements and the CRTM simulations agree in the shape, which indicates that the radiative transfer modeled striping may be able to explain the root cause of the striping. To understand the root cause, we may express the satellite-measured radiance under a clean-sky condition as
e2
where and are the sensor-viewing and sun zenith angles, respectively; and are the sensor-viewing and sun azimuthal angles, respectively; is atmospheric transmittance; is solar spectral irradiance at the top of the atmosphere; and are atmospheric downward and upward radiation, respectively; is the Planck radiance at the surface temperature Ts; is the surface emissivity; and is the surface reflectance of the direct solar radiation.

The first two terms on the right side of Eq. (2) are the top of the atmosphere (TOA) radiance (referred to as A in Table 2) without solar radiation. The last term (the third term) on the right side of Eq. (2) is from surface-reflected solar radiation (referred to as B in Table 2). Using the CRTM modeling, we found that the striping during daytime was caused by the difference in azimuth angles for the 16 detectors, because from the surface reflected solar radiation depends on azimuth angle (see Table 2). Ocean has a strong effect of bidirectional reflectance distribution function. It largely depends on the sensor azimuth angle. The VIIRS's 16 detectors for M bands align along the track direction. For the same scan position or pixel, sensor zenith angles for the 16 detectors are the same. However, the azimuth angles of the 16 detectors are not the same, different from what we expect. Both solar and sensor azimuth angles are relative to the VIIRS pixels and are measured from the local north toward east.

Table 2.

One case study of M12 for 16 detectors over ocean, where R is the total radiance at TOA; see text for A and B. The transmittance and reflectance are unitless.

Table 2.

Figure 6 shows the absolute azimuth angle difference between detectors 1 and 16 for the same scan line. The difference on the left side is positive, while the difference becomes negative on the right side. The azimuth angle between the first and the last detectors has the largest difference near nadir and the difference decreases with the zenith angle. Table 3 lists the azimuth angles of the 16 detectors at different sensor zenith angles or pixels for one scan. The last row in Table 3 is for detector 1 from an adjacent scan. The azimuth angle difference between two scan lines for the same detector is much smaller than the difference between the first and the last detectors for the same scan.

Fig. 6.
Fig. 6.

The absolute sensor azimuth angle difference between detectors 1 and 16 for the same scan. The total pixels per scan for the VIIRS M band is 3200, and the pixel position at 1600 or 1601 is the closet point to nadir.

Citation: Journal of Atmospheric and Oceanic Technology 30, 10; 10.1175/JTECH-D-13-00054.1

Table 3.

Azimuth angles of 16 detectors for various zenith angles/pixels. Both azimuth and zenith angles are given in a unit of degree. The last row is for detector 1 adjacent to the 16 detectors. The azimuth angle difference between two adjacent scan lines for detector 1 is much smaller than the difference between detectors 1 and 16 from the same scan line.

Table 3.

We further examine visible and near-infrared bands. The bottom-left panel of Fig. 5 shows that M1 reflectance (dotted line) does not display the significant surface signal because of dominant molecular scattering. Since molecular scattering decreases dramatically with wavelength, M4 (dashed–dotted line) displays some of the surface BRDF effect. As the atmospheric scattering becomes smaller with the increase of sensor band wavelength, surface-reflected solar radiation becomes significant. M10 (solid line) and the M11 reflectance (dashed line) anomaly displayed the same feature as the M12 band striping.

We also analyzed the VIIRS M12 image at nighttime. The striping was not seen from nighttime images, which excludes the cause due to the difference in the detectors' spectral response. We have analyzed M12 images over clear and uniform ocean scenes and confirmed their finding. In this study, we analyzed the VIIRS M10 and M11 images. The two VIIRS bands are reflective bands and mainly measure reflected solar radiation. The thermal emissive radiation for the two bands is negligible. M10 and M11 radiances displayed the similar striping as the M12, which indicates the striping is related to reflected solar radiation. Such a striping occurred over oceans, where BRDF effect is large. Using the CRTM, we found that M10, M11, and M12 band image striping over a uniform ocean was caused by the difference of sensor azimuthal angles among detectors. The CRTM can simulate the similar striping feature as the M12 measurements.

5. Conclusions and discussion

For a sensor having multiple detectors, the azimuth angle among detectors may be different. The difference can result in satellite image striping if the surface BRDF effect is significant. Both imperfect calibration and geometric difference among detectors can cause the satellite image striping. The imperfect calibration needs to be improved for removing the striping noise. The geometric striping is real and should be considered in satellite product retrievals and in radiative transfer calculations for the instrument monitoring (Liang and Ignatov 2011, 2013) and radiance assimilation.

This study demonstrated that the VIIRS M12 image striping over clear-sky uniform oceans was caused by the azimuth angle difference among the 16 detectors. The azimuth angle difference between the first and the last detectors for a physical scan line is about 2°–3°. Similar striping due to the geometric difference was also found in the VIIRS M10 and M11. In comparison to the blue band M1, the molecular scattering is less than 0.5% and 0.1% for M10 and M11, respectively. This finding can help us to remove solar radiation contamination from the VIIRS M12 band and use the M12 band for sea surface temperature retrievals (Reynolds et al. 2007) at daytime.

Whether to remove and how to remove the image striping depends on the application. The above conclusions are for satellite radiance assimilations and physical retrievals. If image quality instead of radiance matters, then a smoothing software (Poros and Peterson 1985) and MODAPS software are desirable. For example, to derive atmospheric motions from sequential satellite images, the pattern correlation, instead of radiance itself, is the most important. For a qualitative analysis, images without removing the stripe may be acceptable.

Acknowledgments

The authors thank Dr. Alexander Ignatov and the NOAA sea surface temperature team for reporting the VIIRS M12 image striping. The views expressed in this publication are those of the authors and do not necessarily represent those of NOAA.

REFERENCES

  • Baldridge, A. M., , Hook S. J. , , Grove C. I. , , and Rivera G. , 2009: The ASTER spectral library version 2.0. Remote Sens. Environ., 113, 711715, doi:10.1016/j.rse.2008.11.007.

    • Search Google Scholar
    • Export Citation
  • Bréon, F. M., , and Henriot N. , 2006: Spaceborne observations of ocean glint reflectance and modeling of wave slope distributions. J. Geophys. Res., 111, C06005, doi:10.1029/2005JC003343.

    • Search Google Scholar
    • Export Citation
  • Cao, C., , Weinreb M. , , and Sullivan J. , 2001: Solar contamination effects on the infrared channels of AVHRR. J. Geophys. Res., 106, 33 46333 469.

    • Search Google Scholar
    • Export Citation
  • Cao, C., , Xiong J. , , DeLuccia F. , , Liu Q. , , Blonski S. , , Lin G. , , Nishihama M. , , Progorzala D. , , and Oudtari H. , 2012: Visible/Infrared Imager Radiometer Suite (VIIRS) sensor data record (SDR): User's guide, version 1.1. NOAA NESDIS Tech. Rep. 142, 41 pp.

  • Cao, Y., , Yan D. , , Wang G. , , and You S. , 2012: De-striping algorithm in ALOS satellite imagery based on adaptive frequency filter. Image and Signal Processing for Remote Sensing XVIII, L. Bruzzone, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 8537), 853719, doi:10.1117/12.973674.

  • Chen, Y., , Han Y. , , and Weng F. , 2012: Comparison of two transmittance algorithms in the community radiative transfer model: Application to AVHRR. J. Geophys. Res., 117, D06206, doi:10.1029/2011JD016656.

    • Search Google Scholar
    • Export Citation
  • Chin, M., and Coauthors, 2002: Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sunphotometer measurements. J. Atmos. Sci., 59, 461483.

    • Search Google Scholar
    • Export Citation
  • Corsini, G., , Diani M. , , and Walzel T. , 2000: Striping removal in MOS-B data. IEEE Trans. Geosci. Remote Sens., 38, 10131015.

  • Han, Y., , van Delst P. , , Liu Q. , , Weng F. , , Yan B. , , Treadon R. , , and Derber J. , 2005: JCSDA Community Radiative Transfer Model (CRTM)—Version 1. NOAA NESDIS Tech. Rep. 122, 33 pp.

  • Liang, X., , and Ignatov A. , 2011: Monitoring of IR Clear-Sky Radiances over Oceans for SST (MICROS). J. Atmos. Oceanic Technol., 28, 12281242.

    • Search Google Scholar
    • Export Citation
  • Liang, X., , and Ignatov A. , 2013: AVHRR, MODIS, and VIIRS radiometric stability and consistency in SST bands. J. Geophys. Res., 118, 31613171, doi:10.1002/jgrc.20205.

    • Search Google Scholar
    • Export Citation
  • Liu, Q., , and Weng F. , 2006: Advanced doubling–adding method for radiative transfer in planetary atmospheres. J. Atmos. Sci., 63, 34593465.

    • Search Google Scholar
    • Export Citation
  • Liu, Q., , Weng F. , , and English S. , 2011: An improved fast microwave water emissivity model. IEEE Trans. Geosci. Remote Sens., 49, 12381250.

    • Search Google Scholar
    • Export Citation
  • Liu, Q., , Cao C. , , and Weng F. , 2012: NPP VIIRS emissive band radiance calibration. Earth Observing Systems XVII, X. Gu, Ed., International Society for Optical Engineering (SPIE Proceedings, Vol. 8510), 85101E, doi:10.1117/12.930201.

  • Miller, S. D., , Lee T. F. , , Fennimore R. L. , 2005: Satellite-based imagery techniques for daytime cloud/snow delineation from MODIS. J. Appl. Meteor.,44, 987–997.

  • NASA, 2011: Joint Polar Satellite System (JPPS) VIIRS geolocation algorithm theoretical basis document (ATBD). NASA JPSS Ground Project Code 474, 474-00053, 144 pp.

  • Poros, D. J., , and Peterson C. J. , 1985: Methods for destriping Landsat Thematic Mapper images—A feasibility study for an online destriping process in the Thematic Mapper Image Processing System (TIPS). Photogramm. Eng. Remote Sens., 51, 13711378.

    • Search Google Scholar
    • Export Citation
  • Reynolds, R. W., , Smith T. M. , , Liu C. , , Chelton D. B. , , Casey K. S. , , and Schlax M. G. , 2007: Daily high-resolution blended analyses for sea surface temperature. J. Climate, 20, 54735496.

    • Search Google Scholar
    • Export Citation
  • Sayer, A. M., , Thomas G. E. , , and Grainger R. G. , 2010: A sea surface reflectance model for (A)ATSR, and application to aerosol retrievals. Atmos. Meas. Tech., 3, 813838, doi:10.5194/amt-3-813-2010.

    • Search Google Scholar
    • Export Citation
  • Simmer, C., 1994: Satellite Remote Sensing of Atmospheric Hydrological Parameters Using Microwaves (in German). Studies on Agroecology Series, Vol. 13, Verlag Dr. Kovač, 313 pp.

  • van Delst, P., , and Wu X. , 2000: A high resolution infrared sea surface emissivity database for satellite applications. Proc. 11th Int. TOVS Study Conf., Budapest, Hungary, WMO, 407–411.

  • Vogel, R., , Liu Q. , , Han Y. , , and Weng F. , 2011: Evaluating a satellite-derived global infrared land surface emissivity data set for use in radiative transfer modeling. J. Geophys. Res., 116, D08105, doi:10.1029/2010JD014679.

    • Search Google Scholar
    • Export Citation
  • Yang, P., , Liou K. N. , , and Arnott P. W. , 1997: Extinction efficiency and single-scattering albedo for laboratory and natural cirrus clouds. J. Geophys. Res., 102 (D18), 21 82521 835.

    • Search Google Scholar
    • Export Citation
Save