REMOTE SENS, ENVIRON. 50:346-350 (1994~ Short Communication A Precaution Regarding Cirrus Cloud Detection from Airborne Imaging Spectrometer Data Using the 1.38/am Water Vapor Band E. Ben-Dor* A case study that shows a malfunction of the water ods are known to be able to assess cirrus clouds from vapor channel at 1.38 ~am to detect cirrus clouds is orbit. For the most part, these methods are performed presented in this article. A combination of elevation, from the Infrared spectral region [2.7 /am--Barton vegetation coverage, water vapor content, and albedo (1983) and 8, 11, 12 /am--Ackerman et al. (1990)]. characteristics (mostly governed by the terrain) are the Recently, Gao et al. (1993) proposed a method for major factors affecting cirrus cloud detection. Using the detecting cirrus clouds from AVIRIS data using the criteria of relative low radiance and high signal-to-noise water vapor band at 1.38 /am. They concluded that ratio amongst several targets and across the 1.84-1.92 this channel has advantages over the other channels /am spectral region, the 1.8489-/am channel was found to suggested by Barton (1983) and Ackerman et al. (1990) more effectively mask ground signals than the 1.3827-/am and suggested allocation of future satellite sensors at channel. Over targets having moderate elevation, dry around 1.38/am for "daytime cirrus cloud detection." conditions, minimal vegetation, and high albedos, both Green et al. (1993) noted that under conditions of ex- spectra regions present significant ground signals that tremely low humidity and high altitude, the stronger can mistakenly be attributed to cirrus cloud particles. It water vapor at 1.88/am (or at 2.50/am) may provide more is strongly recommended that for accurate cirrus cloud unambiguous cirrus cloud detection. Nevertheless, as detection, both spectral regions around 1.38/am and 1.88 was reported by Gao et al. (1993), even a dry day over /am be examined along with the above-mentioned factors. Nevada, having approximately 0.4 cm of water vapor, still results in zero radiance around 1.38/am, which led them to speculate that this channel is a preferred band for daytime cirrus cloud detection. INTRODUCTION The scope of this article is to present a unique case study that shows, in one scene, several factors that cause The study of cirrus cloud conditions has attracted the the malfunction of the 1.38/am to detect cirrus cloud attention of many workers (e.g., McDougal, 1993). The areas and to put forth a precaution regarding use of extensive coverage, persistence, and radiative effects of water vapor bands across AVIRIS spectra coverage (0.4- cirrus clouds play a major role in the climate system 2.5 #m) for monitoring cirrus clouds. (Starr, 1987). Nevertheless, cirrus clouds are difficult to detect from most satellite sensors, and only a few meth- AVIRIS DATA AVIRIS provides 224 spectral channels across the VIS- *Center for the Study of Earth from Space (CSES), Cooperative SWIR (0.4-2.5 /am) spectral region (visible-VIS and Institute for Research in EnvironmentalScience (CIRES), University short-wave infrared- SWIR) for everv 20 m ground of Colorado, Boulder Address correspondence to E. Ben-Dor, Geography, TeI-Aviv pixels with a sampling interval of about 10 nm. The University, Ramat-Aviv,Tel-Aviv69978, Israel. spectrometer onboard an ER-2 aircraft at an altitude of Received 20 April 1994; revised 23 July 1994. 20 km consists of a swath width of 12 km (Van, 1987). 0034-4257 / 94 / $ 7. O0 346 @Elsevier Science Inc., 1994 6.55 Avenue of the Americas, New York, NY 10010