Representativeness Error Estimated From SSS Products Based on Quadruple Collocation Analysis

Satellite SSS products are usually validated by comparing them with collocated in situ observations from Argo, mooring buoys, and ship measurements. However, the remote sensed data are the measurements averaged spatially within the satellite footprint and in situ observations are pointwise measurements. The different spatial resolution of satellite and in situ data causes the problem of representativeness error (RE) that affects the validation results of satellite SSS. Many efforts have been devoted to properly estimating the REs. REs can be resolved naturally by a quadruple collocated dataset and linear algebra method if the source of REs is known beforehand. Therefore, identifying the source of REs is more important than estimating them. In this study, we find that, for a given quadruple dataset, the results in the error estimation from triple collocation subsets show different characteristics. These differences can be used to identify the source of RE. Then, we present a novel methodology of RE identification based on the characteristics of the error estimation. Using the match-up SSS dataset provided by Pi-MEP and numerical experiments, the feasibility of this method is verified. The results show that for the quadruple collocation (QC) of SMOS, SMAP, Argo, and WOA, the RE occurs in SMOS and SMAP which have similar resolutions with a value of 0.10 psu.