Optimal parameters for generation of gridded product of Argo temperature and salinity using DIVA

Determining an oceanographic parameter on regular grid positions, using a set of data at random locations both in space and time, is the most sought after typical problem since long in the field of oceanography. This is usually called the gridding problem, and the outcome is useful for many applications such as data analysis, graphical display, forcing or initialization of models, etc. In the present study temperature and salinity profiles data obtained from Argo profiling floats were used, and data on regular grids were generated. Data-interpolating variational analysis (DIVA) method was chosen for generating the gridded product. Extensive analysis was done to obtain correct choices of correlation length (L) and signal-to-noise ratio (λ), which results in an optimal gridded product. The gridded data obtained for different choices of L and λ were later validated with datasets deliberately set aside before performing the analyses. For each combination of L and λ, the resultant gridded data was also validated with subsurface data from OMNI buoys. Based on the statistics of comparison with OMNI, the best-fit choice for L and λ was concluded. Later, a comparative analysis was performed with the obtained gridded products from DIVA against the gridded product obtained from objective analysis (OA) to demonstrate the method's reliability. The resultant optimal combination of L and λ will be used for generating Argo gridded data, which will be subsequently used for generating value-added products like mixed layer depth, ocean heat content, D20, etc., and will be made available on INCOIS Live Access Server.