Decluster#

bruces.decluster.decluster(catalog, algorithm='nearest-neighbor', return_indices=False, **kwargs)[source]#

Decluster earthquake catalog.

Parameters:

  • catalog (bruces.Catalog) – Earthquake catalog.

  • algorithm (str, optional, default 'nearest-neighbor') –

    Declustering algorithm:

    • ’gardner-knopoff’: Gardner-Knopoff’s method (after Gardner and Knopoff, 1974)

    • ’nearest-neighbor’: nearest-neighbor algorithm (after Zaliapin and Ben-Zion, 2008)

    • ’reasenberg’: Reasenberg’s method (after Reasenberg, 1985)

  • return_indices (bool, optional, default False) – If True, return indices of background events instead of declustered catalog.

  • window (str {'default', 'gruenthal', 'uhrhammer'}, optional, default 'default') –

    Only if algorithm = "gardner-knopoff". Distance and time windows:

    • ’default’: Gardner and Knopoff (1974)

    • ’gruenthal’: personnal communication (see van Stiphout et al., 2012)

    • ’uhrhammer’: Uhrhammer (1986)

  • method (str, optional, default 'gaussian-mixture') –

    Only if algorithm = "nearest-neighbor". Declustering method:

    • ’gaussian-mixture’: use a Gaussian Mixture classifier

    • ’thinning’: random thinning (after Zaliapin and Ben-Zion, 2020)

  • d (scalar, optional, default 1.6) – Only if algorithm = "nearest-neighbor". Fractal dimension of epicenter/hypocenter.

  • w (scalar, optional, default 1.0) – Only if algorithm = "nearest-neighbor". Magnitude weighting factor (usually b-value).

  • use_depth (bool, optional, default False) – Only if algorithm = "nearest-neighbor". If True, consider depth in interevent distance calculation.

  • eta_0 (scalar or None, optional, default None) – Only if algorithm = "nearest-neighbor" and method = "thinning". Initial cutoff threshold. If None, invoke bruces.Catalog.fit_cutoff_threshold().

  • alpha_0 (scalar, optional, default 1.5) – Only if algorithm = "nearest-neighbor" and method = "thinning". Cluster threshold.

  • M (int, optional, default 16) – Only if algorithm = "nearest-neighbor" and method = "thinning". Number of reshufflings.

  • seed (int or None, optional, default None) – Only if algorithm = "nearest-neighbor". Seed for random number generator.

  • rfact (int, optional, default 10) – Only if algorithm = "reasenberg". Number of crack radii surrounding each earthquake within which to consider linking a new event into a cluster.

  • xmeff (scalar or None, optional, default None) – Only if algorithm = "reasenberg". “Effective” lower magnitude cutoff for catalog. If None, use minimum magnitude in catalog.

  • xk (scalar, optional, default 0.5) – Only if algorithm = "reasenberg". Factor by which xmeff is raised during clusters.

  • tau_min (scalar, timedelta_like or None, optional, default None) – Only if algorithm = "reasenberg". Look ahead time for non-clustered events (in days if scalar). Default is 1 day.

  • tau_max (tau_min : scalar, timedelta_like or None, optional, default None) – Only if algorithm = "reasenberg". Maximum look ahead time for clustered events (in days if scalar). Default is 10 days.

  • p (scalar, optional, default 0.95) – Only if algorithm = "reasenberg". Confidence of observing the next event in the sequence.

Returns:

Declustered earthquake catalog or indices of background events.

Return type:

bruces.Catalog or array_like

Available declustering algorithms#

Gardner and Knopoff (1974)#

bruces.decluster.gardner_knopoff(catalog, return_indices=False, window='default')#

Decluster earthquake catalog using Gardner-Knopoff’s method.

Parameters:

  • catalog (bruces.Catalog) – Earthquake catalog.

  • return_indices (bool, optional, default False) – If True, return indices of background events instead of declustered catalog.

  • window (str {'default', 'gruenthal', 'uhrhammer'}, optional, default 'default') –

    Distance and time windows:

    • ’default’: Gardner and Knopoff (1974)

    • ’gruenthal’: personnal communication (see van Stiphout et al., 2012)

    • ’uhrhammer’: Uhrhammer (1986)

Returns:

Declustered earthquake catalog or indices of background events.

Return type:

bruces.Catalog or array_like

Nearest-neighbor algorithm#

bruces.decluster.nearest_neighbor(catalog, return_indices=False, method='gaussian-mixture', d=1.6, w=1.0, eta_0=None, alpha_0=1.5, use_depth=False, M=16, seed=None)#

Decluster earthquake catalog (after Zaliapin and Ben-Zion, 2008).

Parameters:

  • catalog (bruces.Catalog) – Earthquake catalog.

  • return_indices (bool, optional, default False) – If True, return indices of background events instead of declustered catalog.

  • method (str, optional, default 'gaussian-mixture') –

    Declustering method:

    • ’gaussian-mixture’: use a Gaussian Mixture classifier

    • ’thinning’: random thinning (after Zaliapin and Ben-Zion, 2020)

  • d (scalar, optional, default 1.6) – Fractal dimension of epicenter/hypocenter.

  • w (scalar, optional, default 1.0) – Magnitude weighting factor (usually b-value).

  • use_depth (bool, optional, default False) – If True, consider depth in interevent distance calculation.

  • eta_0 (scalar or None, optional, default None) – Only if method = "thinning". Initial cutoff threshold. If None, invoke bruces.Catalog.fit_cutoff_threshold().

  • alpha_0 (scalar, optional, default 1.5) – Only if method = "thinning". Cluster threshold.

  • M (int, optional, default 16) – Only if method = "thinning". Number of reshufflings.

  • seed (int or None, optional, default None) – Seed for random number generator.

Returns:

Declustered earthquake catalog or indices of background events.

Return type:

bruces.Catalog or array_like

Reasenberg (1985)#

bruces.decluster.reasenberg(catalog, return_indices=False, rfact=10, xmeff=None, xk=0.5, tau_min=1.0, tau_max=10.0, p=0.95)#

Decluster earthquake catalog using Reasenberg’s method.

Parameters:

  • catalog (bruces.Catalog) – Earthquake catalog.

  • return_indices (bool, optional, default False) – If True, return indices of background events instead of declustered catalog.

  • rfact (int, optional, default 10) – Number of crack radii surrounding each earthquake within which to consider linking a new event into a cluster.

  • xmeff (scalar or None, optional, default None) – “Effective” lower magnitude cutoff for catalog. If None, use minimum magnitude in catalog.

  • xk (scalar, optional, default 0.5) – Factor by which xmeff is raised during clusters.

  • tau_min (scalar, timedelta_like or None, optional, default None) – Look ahead time for non-clustered events (in days if scalar). Default is 1 day.

  • tau_max (scalar, timedelta_like or None, optional, default None) – Maximum look ahead time for clustered events (in days if scalar). Default is 10 days.

  • p (scalar, optional, default 0.95) – Confidence of observing the next event in the sequence.

Returns:

Declustered earthquake catalog or indices of background events.

Return type:

bruces.Catalog or array_like