Sensor performance comparison of HyperSpecTIR instruments 1 and 2

Abstract

SpecTIR Corporation has constructed a second copy of their HyperSpecTIR (HST) instrument, with modifications made to various mechanical, electrical, and optical systems. The first instrument (HST1) has been operating for several years aboard multiple platforms, and a sizable archive of imagery has been generated. Using this archive as a baseline, HST2 data have been evaluated to measure expected performance gains versus actual gains. The basic instrument specifications remain unchanged: 227 unique spectral channels from 450-2450nm with 8-12nm FWHM, 1 milliradian IFOV, 256 element cross-track scanning, up to 14 bit digitization, and beam steering optics for image stabilization. Notable changes in HST2 include AR coating of the SWIR FPA, miniaturization of the electronics, and integration of control and data processing computers within the sensor so that it may be used in a pod or UAV. Sufficiently clear data over a single study area does not exist, so data from the spectrally similar areas of Cuprite and Goldfield, Nevada are used to compare the performance of the two instruments. While AR coating of the SWIR focal plane and other improvements to HST2 have improved signal-to-noise performance, these gains are traded off for a shorter integration time allowing for faster and a greater volume of data collection. An attempt to objectively measure spectral image data quality using spectral similarity values and determining the inherent dimensionality of the data reveal similar spectral performance of the instruments under present operational modes.