Calling Data Assimilation Methods Independently

Ensemble Square Root Filter

DAPyr.DA.EnSRF_update(xf: ndarray, hx: ndarray, y: ndarray, HC: ndarray, HCH: ndarray, var_y: float, inf_flag: int, e_flag: int, qc: ndarray, **kwargs)

Performs an Ensemble Square Root Filter update based on Whitaker and Hamill (2002).

Parameters:

  • xf (np.ndarray) – Array of size Nx x Ne containing the ensemble members

  • hx (np.ndarray) – Array of size Ny x Ne containing the ensemble members projected into obs-space

  • y (np.ndarray) – Array of size Ny x 1 containing the observations at time T

  • HC (np.ndarray) – Localization matrix of size Ny x Nx in state-space

  • HCH (np.ndarray) – Localization matrix of size Ny x Ny in obs-space

  • var_y (float) – Observation variance

  • inf_flag (int) – Flag to choose what inflation scheme to choose

  • e_flag (int) – Error flag

  • qc (np.ndarray) – Array of length Ny repesenting quality control of observations

  • **kwargs – Extra arguments to EnSRF_update to pass to inflation parameters. See documentation for EnSRF_update for a list of all possible arguments.

Returns:

  • xa (np.ndarray) – Array of size Nx x Ne representing analysis ensemble states

  • e_flag (int) – Error flag after data assimilation step

**kwargs Parameters

Property

Description

gamma

Inflation parameter for RTPP and RTPS inflation methods

infs

Array of size (Nx) containing initial estimates of inflation means.

infs_y

Array of size (Ny) containing initial estimates of inflation means in obs space.

var_infs

Array of size (Nx) containing initial estimates of inflation variances.

var_infs_y

Array of size (Nx) containing initial estimates of inflation variances in obs space.

Local Particle Filter

DAPyr.DA.lpf_update(x: ndarray, hx: ndarray, Y: ndarray, H: ndarray, C_pf: ndarray, N_eff: float, gamma: float, min_res: int, maxiter: int, kddm_flag: int, e_flag: int, qcpass: ndarray, L: partial)

Performs a Local Particle Filter update based on Poterjoy et al. (2022).

Parameters:

  • xf (np.ndarray) – Array of size Nx x Ne containing the ensemble members

  • hx (np.ndarray) – Array of size Ny x Ne containing the ensemble members projected into obs-space

  • Y (np.ndarray) – Array of size Ny x 1 containing the observations at time T

  • var_y (float) – Observation variance

  • H (np.ndarray) – Array of size Ny x Nx representing the measurement operator

  • C_pf (np.ndarray) – Localization matrix of size Ny x Nx in state-space

  • N_eff (float) – Effective Ensemble Size

  • gamma (float) – Mixing coefficient parameter

  • min_res (float) – Minimum residual for computing betas

  • maxiter (int) – Maximum number of iterations of incremental LPF updates to perform

  • kddm_flag (int) – Flag to turn on kernal density estimation. 0 for off, 1 for on.

  • e_flag (int) – Error flag

  • qcpass (np.ndarray) – Array of length Ny repesenting quality control of observations

  • L (functools.partial) – Callable likelihood function L(y, hx). The first argument is a vector of scalar observations, and the second argument is the model state projected into observation space.

Returns:

  • xa (np.ndarray) – Array of size Nx x Ne representing analysis ensemble states

  • e_flag (int) – Error flag after data assimilation step