EnKPF: Global EnKPF

In robertsy/assimilr: Local Ensemble Kalman Particle Filters

Description

Special cases: setting gam.fix=1: EnKF setting gam.fix=0: PF setting gam.fix=1 and kloc=TRUE recovers the LEnKF

Otherwise gamma is chosen adaptively (then pass e.0 and e.1 as arguments) see also utils_enkpf for the exact procedure.

Usage

EnKPF(xb, y, H, R, taper = 1, P = cov(t(xb)) * taper, gam.fix = NA,
  d = length(y), K = ncol(xb), q = nrow(xb), R2 = msqrt(R), eps1 = R2
  %*% matrix(rnorm(d * K), d, K), eps2 = R2 %*% matrix(rnorm(d * K), d,
  K), unif = runif(1), ...)

Arguments

xb	the background ensemble
y	the observations
H	the observation linear operator
R	the observations error covariance
taper	the tapering correlation matrix applied to P
P	the background covariance
gam.fix	fixed gamma value
d, K, q	dimension of observation, ensemble and state space
R2	matrix square-root of R
eps1, eps2	random perturbations N(0,R)
unif	used for balanced sampling
...	additional parameters passed to adaptive_gamma, typically e.0 and e.1

Value

a list with xa, gam and ess

robertsy/assimilr documentation built on May 27, 2019, 10:33 a.m.