Workload and task management of Grid-enabled quantitative aerosol retrieval from remotely sensed data

As the quality and accuracy of remote sensing instruments improve, the ability to quickly process remotely sensed data is in increasing demand. Quantitative retrieval of aerosol properties from remotely sensed data is a data-intensive scientific application, where the complexities of processing, modeling and analyzing large volumes of remotely sensed data sets have significantly increased computation and data demands. While Grid computing has been a prominent technique to tackle computational issues, little work has been done on making Grid computing adapted to remote sensing applications. In this paper, we intended to demonstrate the usage of Grid computing for quantitative remote sensing retrieval applications. A workload estimation and task partition algorithm was developed, and it executes a generic remote sensing algorithm in parallel over partitioned datasets, which is embedded in a middleware framework for remote sensing retrieval named the Remote Sensing Information Service Grid Node (RSIN). A case study shows that significant improvement of system performance can be achieved with this implementation. It also gives a perspective on the potential of applying Grid computing practices to remote sensing problems.