Downloading and subsetting

Bio-ORACLE layers are global and can be several gigabytes. Subsetting before downloading keeps requests fast and small.

Inspect first

Always check a layer's real ranges before subsetting:

import pyo_oracle as pyo

info = pyo.info_layer("thetao_baseline_2000_2019_depthsurf")
info["dimensions"]   # {'time': (...), 'latitude': (...), 'longitude': (...)}
info["variables"]    # {'thetao_mean': {'units': 'degree_C', 'long_name': ...}, ...}

Build constraints

build_constraints turns friendly (min, max) bounds and strides into the griddap dictionary ERDDAP expects:

constraints = pyo.build_constraints(
    "thetao_baseline_2000_2019_depthsurf",
    time=("2000-01-01T00:00:00Z", "2010-01-01T00:00:00Z"),
    latitude=(-40, -10),
    longitude=(110, 155),
    latitude_step=1,
    longitude_step=1,
)

The equivalent hand-written dictionary (still accepted) looks like:

constraints = {
    "time>=": "2000-01-01T00:00:00Z",
    "time<=": "2010-01-01T00:00:00Z",
    "time_step": 1,
    "latitude>=": -40,
    "latitude<=": -10,
    "latitude_step": 1,
    "longitude>=": 110,
    "longitude<=": 155,
    "longitude_step": 1,
}

Strides

*_step arguments take every n-th grid cell along a dimension — a quick way to downsample a large region.

Restrict variables

Most layers ship several statistics (_mean, _min, _max, _range, ...). Download only what you need:

pyo.download_layers(
    "thetao_baseline_2000_2019_depthsurf",
    constraints=constraints,
    variables=["thetao_mean"],
    response="nc",          # or "csv"
)

Where files go

Downloads land in the configured data directory (see print_config_values) unless you pass output_directory=. A .log file recording the constraints is written next to each download.

pyo.list_local_data()

Warning

Calling download_layers without constraints downloads the entire global layer. pyo_oracle will ask for confirmation first; pass skip_confirmation=True to bypass the prompt in scripts.