Understanding dust emission in the Bodélé region by extracting locally mobilized dust aerosols from satellite Aerosol Optical Depth data using principal component analysis

Abstract

Abstract Despite the increasing availability of satellite and ground-based Aerosol Optical Depth (AOD) data, their application in dust modeling is limited because these data do not differentiate locally mobilized dust from remotely advected dust and other aerosols. In this work, we extract the locally mobilized Dust Optical Depth (DOD) in the Bodele region from historical AOD data through a principal component analysis of wind speed and AOD time series (2003–2012). Principal component analysis effectively identifies the correlated signature between wind speed and AOD making it possible to separate the dust component from AOD data. Using the reconstructed DOD, we then study the effect of key environmental variables, namely wind speed, soil moisture, soil temperature, vegetation, and boundary layer height on dust emission. Results show that all of these environmental variables are significantly correlated with the reconstructed DOD indicating their association with the dust emission process. The extraction technique described in this study can be extended to regional and global scales to identify the dust sources which are not adequately represented in regional and global dust models.