Deep-SDMs in the open oceans - INPUT DATA

This repository contains input files to train the Deep-SDM model described in the preprint Predicting species distributions in the open oceans with convolutional neural networks.

This deposit contains:

1. Training data: CSV dataset + 38 subfolders with data for each species (named after its GBIF id)

2. Prediction data:

2.1. Global use case (solstices & equinoxes of 2021): CSV dataset + data folder

2.2. Western Indian Ocean use case: CSV dataset + data folder

3. species.csv contains the taxonomic name of each taxon, as well as its GBIF id.

4. stats.npy contains normalization factors for the data files

meds, perc1, perc99 = np.load("stats.npy")
item = np.load(file)[:,:,:25]
real_values = (perc99 - perc1) * item + perc1

Each of these elements can be downloaded separately by scrolling to the Files section.