-AT_NNGv2_climatology

This file contains the climatology of AT computed with NNGv2 in netcdf4 format and the climatologies of oxygen (median filtered from WOA13), phosphate, nitrate and silicate (these three derived from CANYON-B). Time resolution: monthly 0-1500m; annual 1550-5500. 

-NNGv2.mat: This file is the neural network object used in the study. To use it you need MATLAB software with the Deep Learning Toolbox.

-Example.rar

This file includes a simple example to apply the NNGv2 (NNGv2.mat) on a subset of GLODAPv2 (n=500) (data_inputs.mat). All files contained in it must be extracted in the same folder and the script (Example_AT_NNGv2.m) must be executed obtaining typical AT values.

NNGv2.mat: This file is the neural network object used in the study. To use it you need MATLAB software with the Deep Learning Toolbox.

data_inputs: 2D matrix. Size must be 10 rows by n columns. Where n is the number of samples where AT will be computed. Each row is a variable and each column is a sample. The order of the variables has to be the following: 1º) latitude; 2º) clongitude; 3º) slongitude; 4º) depth; 5º) temperature; 6º) salinity; 7º) phosphate; 8º) nitrate; 9º) silicate; 10º) oxygen. The units must be: latitude and longitude in [-90º:90º] and [-180º:180º] format (the longitude must be converted to clongitude and slongitude (see Broullón et al. (under review)); depth in m; temperature ºC; and nutrients and oxygen in umol/kg.

-ATNNWOA13.mp4

This file is a video of the surface climatology, 3 vertical sections in the Pacific Ocean, Atlantic Ocean and Indean Ocean and, the variation in depth of one month (April).