Using the Short-Wave Infrared for Nocturnal Detection of Combustion Sources in VIIRS Data

Mikhail Zhizhin,Christopher D Elvidge,Feng-Chi Hsu,Kimberly E Baugh

doi:10.7125/apan.35.6

Abstract

Night-time images from the SNPP satellite VIIRS scanning radiometer in visible and infrared spectral bands provide invaluable data for detection and characterization of natural and technological combustion sources on the surface of the Earth, such as forest fires, gas flares, steel mills or active volcanoes. The presence of sub-pixel hot infrared (IR) emission sources can be readily detected at night in 1.6 micron near-infrared M10 channel. Their temperature and radiant heat intensity can be estimated by fitting of the Planck black-body spectral curve to the observed radiances of VIIRS infrared M-channels out to 4 um. VIIRS instrument is sensitive to the IR sources over a wide range of temperatures. This method can discriminate low temperature sources such as volcanoes and forest fires from the high temperature gas flares with 300 m average location error. The processing includes correction for panoramic “bow-tie” effect and filtering of the false detections resulting from sensor bombardment by the cosmic rays, especially at the aurora rings and at the South Atlantic anomaly. False detections can be largely removed by correlating of the observed bright spots in M10 channel with other infrared and the visible day-night band. NGDC NOAA provides global daily detection products for thousands of IR sources as KMZ vector maps and as CSV tables.

Highlights

Detection of the infrared combustion sources on the Earth surface from satellites is important to the study of biomass and gas flare burning, which result in greenhouse gas emissions volumes large enough to affect ecology and global climate [1,2]
In this paper we present a novel multi-spectral method for detection of nocturnal IR sources on the Earth and determine their temperature, portion of pixel, radiant heat intensity and radiant heat using up to six spectral bands spanning the visible to mid-infrared
Nocturnal detection of the short-wave IR radiated from combustion sources is performed at the NOAA National Geophysical Data Center (NGDC) in Boulder nightly on a regular basis using global coverage UTC day VIIRS dataset received from the NOAA Comprehensive Large Array Stewardship System (CLASS) [7]

Summary

Introduction

Detection of the infrared combustion sources on the Earth surface from satellites is important to the study of biomass and gas flare burning, which result in greenhouse gas emissions volumes large enough to affect ecology and global climate [1,2]. It is possible to detect some combustion sources during the day, nocturnal observations are advantageous since the radiant emissions can be observed without solar contamination. Thermal channels originally designed to identify cloud-contaminated pixels and evaluate the radiant contribution from water vapor showed the ability to detect pixels with high temperature sources. The method only works for pixels where the hot source is detected in both spectral bands, which is a rare occurrence in the 11 um channel. Identifying pixels with hot sources based on the brightness temperature difference between a 4 um and an 11 um channel emerged as the primary method for generating fire products from both AVHRR and MODIS data [5,6,7]

Methods

Results

Conclusion