A GEOLOCATION ERROR ASSESSMENT OF SNPP VIIRS I5 USING LANDSAT THERMAL BAND

Abstract:

The Visible Infrared Imaging Radiometer Suite (VIIRS) on board the Suomi National Polar orbiting Partnership (NPP) satellite launched in 2011 has been extensively calibrated and validated and has mature Sensor Data Records (SDR). The VIIRS geolocation cal/val activities are based on a 15-year heritage dating back to the launch of the MODerate resolution Imaging Spectrometer (MODIS). The geolocation accuracy of the VIIRS Reflective Solar Bands (RSB) is based on calibration of I1 red (0.555 µm) radiance while all other bands are registered to the I1 using band-to-band registration (BBR) implemented as part of the pre-launch activities. On-board assessment of BBR is conducted using the moon during monthly lunar maneuvers. While band-to-band registration is reported to have an accuracy of ~40 m, degradation due to sensor thermal variation can occur and a more frequent alternate assessment is needed. The geolocation accuracy of the RSB has been well documented, however the accuracy of the thermal bands has not been quantified. A geolocation error assessment of the VIIRS I5 (11.45 µm) thermal band is presented here. The methodology presented is an adaptation of the visible band approach previously mentioned. While the visible band approach uses Landsat red band based Ground Control Points (GCP) or “chips” that are resampled to VIIRS I1 red band, the thermal approach uses Landsat band 10 (10.89 µm) radiance based “thermal chips” resampled to VIIRS I5 band. The geolocation error is quantified by comparing the VIIRS radiance with simulated Landsat derived VIIRS radiance. Thermal chips can take advantage of data collected in the VIIRS night-time descending orbits however there are limited Landsat night-time data, this results in a mismatch of relative thermal characteristics (i.e. water is warmer than land at night) therefore the daytime and nighttime chips cannot be compared. A novel solution is presented that makes use of an “edge detection” technique eliminating this time-dependence. Results of the analysis illustrate the daily along-scan and –track errors and are compared with SNPP geolocation band-to-band registration.