Improved Projection Algorithms for Long-Term and High-Resolution Satellite Datasets.

Satellite mission datasets increase in size as their life span grows and the resolutions of the instruments increase. Accurately projecting antenna-based satellite measurements on a geographical grid while maintaining reasonable computational time and costs can be challenging. It is thus necessary to include big data algorithms and dedicated management techniques in the processing of such datasets.In this regard, the optimization of the number of interpolations and projections is one of the key aspects, as the native errors of the measurements propagate at each processing step. Besides each interpolation and projection implies a non-negligible increase in total computational time.This work is based on the Sea Surface Salinity (SSS) processor of the Soil Moisture and Ocean Salinity (SMOS) mission, but it could be easily extended to any other satellite mission where individual values for each measurement are retrieved. We propose a redefinition of the complete processor chain so it can work with the measurements within the instrument coordinate system. This allows us to avoid projection-related errors during the generation of the final product.Additionally, we introduce a novel algorithm to project those measurements taking into account the actual spatial extent of the acquisitions instead of taking them as points, so measures are averaged weighted by the area they cover on the Earth-based grid. This method is optimized to transform 2D areas into discrete measurements, increasing its computational efficiency and favoring parallelization. Our algorithm has demonstrated its potential when incorporated into the SMOS SSS processor at the Barcelona Expert Center (BEC), allowing us to keep a final resolution very close to the one attained at the antenna coordinate system.