A new method to retrieve aerosol optical thickness from satellite images on a parallel system

A wide variety of algorithms have been developed to monitor aerosol burden from satellite images. Still, few solutions currently allow for real-time and efficient retrieval of aerosol optical thickness (AOT), mainly due to the extremely large volume of computation necessary for the numeric solution of atmospheric radiative transfer equations. Taking into account the efforts to exploit the SYNergy of Terra and Aqua Modis (SYNTAM, an AOT retrieval algorithm), we present in this paper a novel method to retrieve AOT from Moderate Resolution Imaging Spectroradiometer (MODIS) satellite images, in which the strategy of block partition and collective communication was taken, thereby maximizing load balance and reducing the overhead time during inter-processor communication. Experiments were carried out to retrieve AOT at 0.44, 0.55, and 0.67 μm of MODIS/Terra and MODIS/Aqua data, using the parallel SYNTAM algorithm in the IBM System Cluster 1600 deployed at China Meteorological Administration (CMA). Results showed that parallel implementation can greatly reduce computation time, and thus ensure high parallel efficiency. AOT derived by parallel algorithm was validated against measurements from ground-based sun-photometers; in all cases, the relative error range was within 20%, which demonstrated that the parallel algorithm was suitable for applications such as air quality monitoring and climate modeling.