Application of Oxygen A-band Equivalent Width for Cloud Optical Depth Measurement

The oxygen A-band near 760 nm has a long history of use in measuring atmospheric properties. Because it uses a uniformly mixed atmospheric constituent, the amount of absorption is a direct measure of the radiation path length. We have developed a novel application of A-band absorption to resolve the inherent ambiguity in the relationship between the spectral radiance of scattered sunlight from clouds overhead and cloud optical depth. In optically thin clouds, the brightness of the cloud increases as the optical depth increases; in optically thick clouds, the brightness decreases as the optical depth increases. This means that there are two possible cloud optical depths that can produce a measured spectral radiance. The A-band equivalent width (defined as 1 minus transmittance integrated over the band) has the advantage of being a monotonic function of the cloud optical depth, although it does depend on the average physical thickness and altitude of the cloud as well as other factors. We have found that most of the time the optical depth dominates all the other factors.

Because clouds are dynamic as well as non-uniform, both the spectral radiance and the equivalent width are always changing. For optically thin clouds these quantities are positively correlated; for optically thick clouds they are negatively correlated. By examining the sign of the correlation over a short time window (about 5 seconds) the brightness ambiguity can be resolved. For those times when the equivalent width variation is not dominated by the optical depth, a suitable nonlinear filter is able still to resolve the ambiguity. This filter, while based on the well-known statistical method of maximum likelihood (ML), is not strictly a ML estimate of the optical depth. Rather, it assigns a likelihood of the observed portion of sky being in one of five mutually exclusive states based on the spectral measurements, their past values and physics-based heuristics. This technique appears capable of identifying the thick and thin states with a notably low error rate and thus is an effective means of retrieving the correct optical depth.

We have applied this algorithm to more than 30 days of the recent DOE Two-Column Aerosol Project at the ARM Mobile Facility on Cape Cod using the Aerodyne Three-Waveband Spectrally-agile Technique (TWST) sensor. Samples and a summary of this data will be presented. Other applications of the A-band equivalent width to monitoring cloud properties will also be discussed.