Citation:

Williams, D J., B. Feldman, T. Williams, M. Winters, D Pilant, AND L D. Worthy. DETECTION AND IDENTIFICATION OF TOXIC AIR POLLUTANTS USING AIRBORNE LWIR HYPERSPECTRAL IMAGING. Presented at International Society for Optical Engineering (SPIE), Remote Sensing of the Atmosphere, Ocean, Environment, and Space Symposium 2004, Honoluu, HI, November 8-12, 2004.

Impact/Purpose:

The objectives of this task are to:

Assess new remote sensing technology for applicability to landscape characterization; Integrate multiple sensor systems data for improved landscape characterization;

Coordinate future technological needs with other agencies' sensor development programs;

Apply existing remote sensing systems to varied landscape characterization needs; and

Conduct remote sensing applications research for habitat suitability, water resources, and terrestrial condition indicators.

Description:

Gaseous releases from petrochemical, refinery, and electrical production facilities can contribute to regional air quality problems. Fugitive emissions or leaks can be costly to industry in terms of lost materials and products. Ground-based sampling and monitoring for leaks are time consuming and costly as well, and do not accurately characterize total facility releases over time. Hyperspectral remote sensing in the LWIR spectral region allows for synoptic and repeatable monitoring of important air pollutants at potentially lower cost than ground sampling. This spectral region is optimal for spectrally identifying many organic chemical species. LWIR hyperspectral imagery was collected over several petrochemical facilities in Houston, Texas,
USA in April 2004. The sensor used was the Airborne Hyperspectral Imager (AH!), designed and built by the University of Hawaii. This system has 256 spectral bands from 7.5 to 11.7 micrometers and can be flown on most camera ready twin engine aircraft platforms. Ground- based gas sampling was accomplished during the AHI overflight to relate the airborne observations with ground truth. Several computational methods were used to analyze the AHI data including the N-FINDR algorithm and matched filter techniques. The results of the data analysis, together with the ground-based chemical measurements, were assimilated into a GIS database to determine gas emission detection probabilities and chemical compound identification accuracy. This methodology holds great promise for accurately and efficiently monitoring gaseous emissions at the numerous and varied chemical industries that exist in many metropolitan areas.

Record Details: