Generate a synthetic zooplankton database — generate_mock_zoop

Writes a CSV matching the schema original/create_database.R produces, so the pipeline can be exercised end-to-end without the real survey data, which stays on its owner's machine.

Usage

generate_mock_zoop_data(config, n_stations = 400, seed = 42)

Arguments

config: a config list, as returned by load_config()
n_stations: number of station visits to generate per year
seed: random seed

Value

the path written, invisibly

Details

Two details are copied from the real database's structure rather than invented, because the pipeline's behavior depends on them:

Stage columns that a survey does not resolve are filled with zeros, not NA (create_database.R lines 32-55 for ECOMON), so <species>_total collapses to the single stage ECOMON measures.
dataset takes the values ECOMON, ECOMON_STAGED, MBON, CPR, so the config's dataset_filter is actually exercised rather than matching every row.

Abundance is given real spatial and seasonal structure — a temperature-like gradient plus a summer peak — so a model fit to it should score meaningfully better than chance. A smoke test that passed on pure noise would not be testing anything.