Studies will be described that define the capability of a spectrometer system to re-motely detect and characterize trace gases in a localized cloud; e.g., a stationary source effluent. The detection method utilizes all of the information contained in the observed spectral radiance contrast between cloud and background. It consists essentially of deter-mining the degree to which this spectrum is correlated with a computed (FASCOD1) reference spectrum. It is shown that trace gases can be reliably detected even when spectrum fea-tures are well below the noise level. The minimum detectable quantities (MDQ's) for various trace gases at one atmosphere total pressure are given. The MDQ's determine the combinations of gas column thickness and gas-background temperature difference that corres-pond to 95 percent detection probability and one percent false detection probability for a given system noise equivalent spectral radiance. The capability of the method to remotely infer the gas column thickness and temperature will also be discussed.