This project is based on the study and application of Empirical Mode Decomposition (EMD) and its two-dimensional extension, Bidimensional Empirical Mode Decomposition (BEMD), to minimize Radio Frequency Interference (RFI) in satellite observations. Satellite remote sensing missions, such as the SMOS (Soil Moisture and Ocean Salinity) satellite managed by the European Space Agency (ESA), monitor soil moisture and ocean surface salinity. Even so, the SMOS radiometers (MIRAS), which operate in the L-band, are vulnerable to RFI. This interference appears in the brightness temperature (BT) images as intense peaks, compromising data quality. Traditional RFI mitigation methods often struggle with the non-linear and non-stationary behavior of interference. BEMD is introduced as an adaptive, empirical alternative that decomposes an image into Intrinsic Mode Functions (IMFs), enabling the extraction and removal of high-frequency components associated with RFI while preserving low-frequency information and overall trends. Initial work validated the effectiveness of the EMD methodology by applying it to noisy Global Navigation Satellite System (GNSS) signals. Quantitative results demonstrated that EMD successfully restored critical signal information: for instance, in "File_02," the number of visible satellites increased from 3 to 4, allowing for position determination. Similarly, "File_03," which initially yielded information from only one satellite, was enhanced to provide data from four satellites after EMD application. For the core objective, a personalized BEMD algorithm was developed in MATLAB to process SMOS BT images. The decomposition successfully isolated the RFI, which manifests as high-frequency components (typically in the initial IMFs), demonstrating efficacy against both single and multiple interference sources within the same BT image. Furthermore, to optimize results given the data's geometry, the project investigated modifying the BEMD interpolation process from a standard square mesh to a hexagonal mesh. This adaptation proved highly successful, enabling the interference to largely disappear, with remaining interference areas suppressed below the critical 350K threshold, a value significantly higher than the natural maximum BT of Earth's surface (typically below 320K) and indicative of RFI contamination. The study concludes that EMD and BEMD shows an adaptable strategy for interference mitigation in dynamic remote sensing environments. Implementing this adaptive technique in SMOS data processing can improve the accuracy of salinity and humidity maps, thereby benefiting climate modeling, water resource management, and oceanography

9 - Indústria, Innovació i Infraestructura