Error analysis of spectral remote sensing by CO2 4.3 μ m band in various temperature profiles

Abstract

The inversion error of temperature profiles reconstructed by the spectral remote sensing (SRS) technique is analyzed by using CO24.3 μm band. Total of 14 narrow bands averaged over 25 cm-1 spectral width is selected as the low-resolution sensing spectra from 2050 to 2400 cm-1. The HITRAN line-by-line database enhanced with hot line data is utilized as the tool to generate the reference spectral intensity. The medium is assumed to be filled with nonisothermal pure CO2 gas at 1 atm. Four temperature profiles are tested in this work: three boundary layer types, and one prabolic type. Continuous temperature distribution is assumed by specifying three pointwise temperatures. Through this work, it is investigated how much inversion error of temperature is induced per unit relative error in true and measured intensities when changing the number of sensing narrow bands, combination of narrow bands, path length, temperature profile shape, temperature level, and temperature gradient. The error analyses show that the SRS inversion with this band gives accurate inversion of temperature within 1% error when both of the measured and the reference intensities have 1% error for any temperature profile up to 2500 K and up to about 10 m path length. It is also found that a cold layer behind a hot layer is not accurately detected for higher path length, so the sensor is better to be located at the colder side. ？ 2000 Elsevier Science B.V.