چکیده (انگلیسی):

The multifractal fusion is a non-parametric technique that merges remotely-sensed maps of different ocean variables to produce higher quality remote sensing products. This method was first introduced to reduce the noise level in a map of a given ocean variable provided that a higher-quality map of a second ocean variable is available. In this work, we go one step further in both the theoretical justification and the application of the multifractal fusion. In the first part of the paper we explore if the quality of the derived maps can be improved by using a higher-dimensional tensor relating the gradient of one variable with that of the other. Higher-dimension tensors require the introduction of appropriate integration schemes to avoid inducing long-range artifacts, something that is not completely solved yet. The results show that although higher-dimension tensors lead to an improvement in the spatial consistency of the derived maps (i.e., a higher quality of the associated singularity exponents), the comparison with in situ data still show significant reconstruction errors as compared to the simplified scalar approach. In the second part of the paper we have used the scalar approach to increase the spatial and temporal resolution of the current Soil Moisture and Ocean Salinity (SMOS) Level 3 sea surface salinity (SSS) maps produced at the Barcelona Expert Center (BEC). As such, new daily SMOS SSS global maps are generated at a 0.05° × 0.05° grid resolution. The quality of the resulting products has been assessed using in-situ comparison, singularity analysis and power density spectra. It is concluded that the multifractal fusion can be used to significantly increase the spatial and temporal resolution of the SSS maps.