On the Misclassification of Dust as Cloud at an AERONET Site in the Sonoran Desert

Abstract

Here we present retrievals of aerosol optical depth t from an Aerosol Robotic Network (AERONET) station in the southeastern corner of California, an area where dust storms are frequent. By combining AERONET data with collocated ceilometer measurements, camera imagery, and satellite data, we show that during significant dust outbreaks the AERONET cloud-screening algorithm oftentimes classifies dusty measurements as cloud contaminated, thus removing them from the aerosol record. During dust storms we estimate that approximately 85% of all dusty retrievals of t and more than 95% of retrievals when t > 0.1 are rejected, resulting in a factor-of-2 reduction in dust-storm averaged t. We document the specific components in the screening algorithm responsible for the misclassification. We find that a major reason for the loss of these dusty measurements is the high temporal variability in t during the passage of dust storms over the site, which itself is related to the proximity of the site to the locations of emission. We describe a method to recover these dusty measurements that is based on collocated ceilometer measurements. These results suggest that AERONET sites that are located close to dust source regions may require ancillary measurements to aid in the identification of dust.