geoprior.params.FixedHRef#

class geoprior.params.FixedHRef(value=0.0, mode='auto', name=None, non_negative=False, **kws)[source]#

Bases: BaseFixed

Reference head configuration \(h_{\mathrm{ref}}\) for drawdown.

Drawdown convention used in GeoPrior: \(\Delta h = \max(h_{\mathrm{ref}} - h, 0)\).

This is a modelling datum (not a material parameter). In regional hydrogeology it may represent a pre-development head, a long-term mean head, or (recommended here) a rolling baseline derived from the last observed historical head at forecast start.

Parameters:

  • value (float or None, default 0.0) – Fallback reference head [m] used when mode=”auto” cannot be resolved. If None, defaults to 0.0.

  • mode ({"auto", "fixed"}, default "auto") –

    • “auto”: derive \(h_{\mathrm{ref}}\) per batch from the last historical groundwater observation (preferred).

    • ”fixed”: always use value as a global datum.

  • name (str, optional) – Variable name.

  • non_negative (bool, default False) – Allow negative values since heads can be negative depending on datum.

__init__(value=0.0, mode='auto', name=None, non_negative=False, **kws)[source]#
Parameters:

Methods

__init__([value, mode, name, non_negative])

from_config(config)

Re-instantiate from configuration dict.

get_config()

Return a JSON-serialisable dict for tf.keras serialization.

get_value()

Retrieve the fixed parameter value.
__init__(value=0.0, mode='auto', name=None, non_negative=False, **kws)[source]#
Parameters:
get_config()[source]#

Return a JSON-serialisable dict for tf.keras serialization.