Abstract
Multisensor-multispectral scientific data applications require a tremendous investment regarding data preparation and analysis. A data fusion method is developed that is general enough for use with any scan-line-based datasets (satellite and ground based) and enables multisensor-multispectral datasets to be merged on a routine basis. In particular, advanced software processing techniques in the Polar Orbiter Remapping and Transformation Application Library (PORTAL) are used to show how polar (e.g., Special Sensor Microwave/Imager data from DMSP) and geostationary data (e.g., Visible-Infrared Spin Scan Radiometer data from GOES) can be spatially combined routinely in an efficient manner. The PORTAL system can also be used to combine data from sensors on the same satellite that have radically different earth scan patterns and ground resolutions. A self-describing, generalized data format is used to modularize the data processing flow and obtain significant improvements in terms of flexibility, extensibility, and generality of application. A new data fusion method is introduced that consists of a comprehensive set of data fusion and merger software tools that generate merged datasets in an original satellite projection space. Computational efficiencies are compared between this data fusion method and that of conventional remapping methods. While comparable processing times are needed to physically merge the datasets, results show significant performance gains on any subsequent analysis of the merged datasets since scientific algorithms operate within the original satellite projection space. Effective microwave surface emittance is retrieved from muitisensor-multispectral data as a demonstration of the technique.
