References of "Parard, Gaëlle"
     in
Bookmark and Share    
Full Text
Peer Reviewed
See detailAnalysis of SMOS sea surface salinity data using DINEOF
Alvera Azcarate, Aïda ULiege; Barth, Alexander ULiege; Parard, Gaëlle ULiege et al

in Remote Sensing of Environment (2016), 180

n analysis of daily Sea Surface Salinity (SSS) at 0.15 ° × 0.15° spatial resolution from the Soil Moisture and Ocean Salinity (SMOS) satellite mission using DINEOF (Data Interpolating Empirical Orthogonal ... [more ▼]

n analysis of daily Sea Surface Salinity (SSS) at 0.15 ° × 0.15° spatial resolution from the Soil Moisture and Ocean Salinity (SMOS) satellite mission using DINEOF (Data Interpolating Empirical Orthogonal Functions) is presented. DINEOF allows reconstructing missing data using a truncated EOF basis, while reducing the amount of noise and errors in geophysical datasets. This work represents a first application of DINEOF to SMOS SSS. Results show that a reduction of the error and the amount of noise is obtained in the DINEOF SSS data compared to the initial SMOS SSS data. Errors associated to the edge of the swath are detected in 2 EOFs and effectively removed from the final data, avoiding removing the data at the edges of the swath in the initial dataset. The final dataset presents a centered root mean square error of 0.2 in open waters when comparing with thermosalinograph data at their original spatial and temporal resolution. Constant biases present near land masses, large scale biases and latitudinal biases cannot be corrected with DINEOF because persistent signals are retained in high order EOFs, and therefore these need to be corrected separately. The signature of the Douro and Gironde rivers is detected in the DINEOF SSS. The minimum SSS observed in the Gironde plume corresponds to a flood event in June 2013, and the shape and size of the Douro river shows a good agreement with chlorophyll-a satellite data. These examples show the capacity of DINEOF to remove noise and provide a full SSS dataset at a high temporal and spatial resolution with reduced error, and the possibility to retrieve physical signals in zones with high initial errors. [less ▲]

Detailed reference viewed: 108 (18 ULiège)