Appendix A: Cruises included in the Cape Basin hydrographic database

Appendix B: GEM-ETTA gridding algorithm

A combination of two-dimensional nearest neigbor (median filter) and linear regression is used to create the GEM-ETTA estimate of T or S on each pressure surface. The nearest neighbor algorithm has the benefit of being robust in the face of large areas with no data (see Figs. 3b and 3c). Nearest neighbor gridding is applied first to the whole domain, and then repeated on successively smaller regions. Results of successive iterations are averaged, effectively smoothing the transitions across the data gap areas. Temperature and salinity are treated independently.
Figure B-1. The equivalent filter weight and shape for the iterative procedure detailed above; the normalized distance spans the entire domain of Fig. 3c.
next up previous