VNP03DNB - VIIRS/NPP Day/Night Band Moderate Resolution Terrain-Corrected Geolocation 6-Min L1 Swath 750m

VNP03DNB is the short-name for the Suomi National Polar-orbiting Partnership (SNPP) platform-based NASA VIIRS Level-1 terrain-corrected geolocation product and contains the derived line-of-sight (LOS) vectors for the single panchromatic Day-Night band (DNB). The geolocation algorithm uses several inputs that include an Earth ellipsoid, geoid, and a digital terrain model along with the SNPP platform's ephemeris and attitude data, and knowledge of the VIIRS sensor and satellite geometry. It provides geodetic coordinates (latitude and longitude), and related parameters for each VIIRS Level-1 pixel. The VNP03DNB product includes geodetic latitude, longitude, surface height above the geoid, solar zenith and azimuth angles, lunar zenith and azimuth angles, sensor zenith and azimuth angles, land/water mask, moon illumination fraction and phase angle, and quality flag for every pixel location.

Geometric Correction Methods and Reference Data Sources
A primary tool used in the geometric correction process is a control-point matching (CPM) program, which uses a library of globally distributed ground-control point (GCP) chips as ground truth. These GCP chips are derived from highly accurate, cloud-free, high-contrast, 30 m, Landsat red-band geolocation sub-images that cover an area of approximately 20 km by 20 km. These datasets are used to simulate the coarser VIIRS image-resolution (I1) 375 m band-1 images with which they are correlated to produce a field of correlation coefficient values (CCV). Shifts occurring at the maximum CCVs form geolocation residuals, and a control-point file for these residuals is produced for every VIIRS 375 m I-band geolocation granule. These residual's mean values represent biases, which help construct look-up tables (LUT) that are used to correct geolocation errors. This CPM was originally developed for MODIS and later adapted for VIIRS. For additional details, consult the following papers: (Wolfe et al., 2013, Wolfe et al., 2012).

Geometric Accuracy of the Data
The overall geolocation accuracy for both the SNPP VIIRS and NOAA-20 VIIRS Level-1 data is nominally good. The nadir-equivalent geolocation uncertainty is within 75 m (1-sigma error), or 20% of an image-band pixel, in both the along-scan and along-track directions. These results are confirmed (Lin et al., 2022, Lin et al., 2020) for both the Collection-1 and Collection-2 reprocessing cycles for both missions. For instance, the Collection-2 mean is close to zero and the root-mean-square-error (RMSE) is less than 75 m. In general, each reprocessing cycle gradually improves the bias in mean errors (in both the along-track and along-scan directions) as well as the RMSEs. The on-orbit geometric calibration and validation activities for both VIIRS instruments aboard the SNPP and NOAA-20 platforms have continued for a decade. They have addressed nominal geolocation, risk-reduction strategies, and general improvements in how the VIIRS instrument sensor operations are managed. Following this decade-long involvement with studying and analyzing VIIRS' geolocation, sub-pixel geolocation accuracy — or a consistency in location and comparison of pixels from the same instrument — is achieved.