Resumen de Variability, geophysical consistency, and calibration errors of sea surface salinity remote sensing data
In January 1969, Jack F. Paris published a comprehensive review of all the knowledge about microwave physics, engineering and its applications to meteorology and oceanography. In this 241 pages-long review, he pointed out that radiometers operating between 1.0 GHz and 5.4 GHz could be used to survey remotely both temperature and salinity along coastal and river regions in the world.
Half a century later, a number of satellite missions have proved him right, with an accuracy beyond his initial estimates. Indeed, the Soil Moisture and Ocean Salinity (SMOS), the Aquarius and the Soil Moisture Active Passive (SMAP) missions have provided global maps of sea surface salinity. These missions required solving a number of technical issues and the development of novel data analysis methods to reduce the noise in the measurements in order to retrieve salinity values in most regions of the world ocean. However, assessment of the quality of the retrieved SSS continues to be a topic under development.
Since the launch of the SMOS mission (November 2, 2009), the validation of remotely sensed salinity has been commonly based on the direct comparison of the satellite estimate against sparse in situ measurements. However it is known that the spatial and temporal representation of satellite-derived SSS is very different from that of in situ SSS.
In the case of remote sensing instruments, the measurements concern the first centimeter of the sea surface with a spatial resolution of several tens of kilometers and a temporal resolution of a week or longer. However, the in situ instruments are providing measurements of the SSS within the first meters below the sea surface at one point in space, with daily or even hourly temporal resolution. Therefore, any kind of direct comparison needs to take into account those spatio-temporal differences (i.e., representativeness error) to improve the characterization and estimation of the remote sensing products. Moreover, as different satellite products resolve different spatio- temporal scales, it is important to assess their ability to detect and characterize geophysical processes, such as mesoscale features (e.g., fronts, eddies and filaments).
In this PhD thesis, two methods, based on singularity analysis and triple collocation, have been adapted to better assess, respectively, the geophysical consistency and the errors of different remote sensing SSS products. These methods have been applied to different gridded SSS maps, such as SMOS and Aquarius high-level products, the output of a numerical simulation, in situ reanalysis and climatology, as well as to other sea surface temperature products for reference. The singularity analysis has demonstrated that, beyond the remaining sources of uncertainty in remote sensing SSS products, valuable dynamical information of the ocean state can be extracted from these SSS products. The triple collocation has demonstrated the importance of an accurate estimation of the representativeness errors to correctly estimate the errors of remote sensing products.
