Diagnosis of Saturation Effects in Infrared Channels of VIIRS Using the NOAA Nightfire Product

As the first of the next generation of operational Earth-observing instruments, the Visible Infrared Imaging Radiometer Suite (VIIRS) offers many improvements over its predecessors. VIIRS delivers satellite data record (SDR) and environmental data record (EDR) data at higher spatial resolution than legacy instruments, including the MODerate Resolution Imaging Spectroradiometer (MODIS), the Advanced High Resolution Radiometer (AVHRR), and the Operational Line Scanner (OLS). One marked difference between VIIRS and MODIS is that VIIRS employs on-board processing to reduce downlink bandwidth requirements and preserve spatial resolution across the swath. This on-board processing, which involves averaging of multiple detectors to obtain single SDR “pixel” radiances, can have unintended and poorly understood consequences in cases of detector saturation. To examine the potential impact of detector saturation, we used the NOAA Nightfire product, which provides radiance data and extensive metadata for a pre-screened selection of nighttime pixels displaying anomalies in the near-infrared and infrared wavelengths. Nearly four months of data were compiled and analyzed to detect saturation in the infrared bands. The 3.7 µm band (M12) was the most frequently-saturating band, with 0.15% of pixels displaying a hard saturation limit indicating saturation of all component detectors. However, the M12 radiance distribution displayed a strong relationship with the viewing angle, reflecting the effects of sub-pixel (detector-based) averaging. The three different detector aggregation schemes were separated in the data and analyzed, showing how the effects of detector saturation vary based on viewing angle. A simulation was constructed to verify that the distributions we observed in the M12 data were consistent with sub-pixel detector saturation. This partial saturation affects a much larger percentage of pixels with reported radiances well below the hard saturation limit. In addition, we show the implication of saturation for retrieved fire properties such as fire radiative power (FRP). The impact of detector-level saturation must be considered for applications which require quantitative use of the VIIRS M12 band, and further thought must be given to the implementation of on-board processing in future instruments.