Abstract
Annual VIIRS global night-time lights traditionally report mean radiance levels after filtering to remove moonlit and cloudy data followed by outlier removal. The mean is the first statistical moment, but there are three other statistical moments (variance, skew and kurtosis) that are currently not produced globally. In this study, we generated VIIRS day/night band multiyear and annual statistical moments for a widely dispersed set of test areas. The moments were calculated from 15 arc second nightly temporal profiles spanning 2012-2020, filtered to exclude cloudy and sunlit data, with radiance adjustments to reduce view angle and lunar illuminance effects. The moment data were examined in two ways: (1) Geospatial grids-which reveal zonation and temporal changes present in urban areas and (2) Scattergrams of moment pairs. The variance versus mean scattergram exhibits several distinct data clusters used to define five zones: core lighting, dark-erratics, mid-erratics, bright-erratics and bright and seteady. The results indicate that the information content of VIIRS night-time lights can be augmented if all four statistical moments are calculated.
Original language | American English |
---|---|
Pages (from-to) | 7778-7802 |
Number of pages | 25 |
Journal | International Journal of Remote Sensing |
Volume | 45 |
Issue number | 21 |
DOIs | |
State | Published - 2023 |
NREL Publication Number
- NREL/JA-2C00-85565
Keywords
- day-night band
- DNB
- night-time lights
- remote sensing
- urban mapping
- VIIRS