Phytoplankton variables and atmospheric drivers for machine learning modeling of phytoplankton blooms in the Salish Sea

The dataset consists of five files. Four of them correspond to the four phytoplankton variables of interest (i.e., diatom biomass, nanoflagellate biomass, diatom production rate, nanoflagellate production rate). For each one of them, the respective atmospheric drivers and spatiotemporal features that were used as inputs are included. Each of these four files contains data from different periods to match the spring bloom of each phytoplankton variable. The fifth file provides the seasonal variabilities of all four phytoplankton variables, calculated every 1st and 15th of each month. The different period of interest for each phytoplankton variable was selected based on the fifth file. The phytoplankton variables were produced by SalishSeaCast V202111, whilst the atmospheric drivers were produced by HRDPS. The phytoplankton variables were integrated from 0-100m depth, and the same flagging was applied to the atmospheric drivers. These files were used in order to train and test the machine learning algorithms used in the respective study. The respective GitHub repository can be found here: https://doi.org/10.5281/zenodo.17180298.