StFit: weighted least-squares fit of a vgmST model to an
empirical space-time variogram, as gstat::fit.StVariogram
(parameters in extractPar order, gstat's default bounds, weights
fit_method 1, 2, 6 or 7), by L-BFGS-B, with least-squares AIC and
BIC. StUniversalKriging: kriging with drift \(x'\beta\), as
krigeST(z ~ x). StLocalKriging: the ceiling(buffer
nmax) nearest observations in the metric
\(\sqrt{dx^2 + dy^2 + (a\,dt)^2}\), of which the nmax most
covarying are kept, as krigeST(nmax, stAni).
StBlockKriging: kriging of a space-time block mean (square block,
time window, discretised). StLeaveHOut: buffered cross-validation.
StKrigingDiagnostics: conditioning, weights, Lagrange multiplier,
relative variance. StSmoothness: mean-square differentiability
order in space and time. StPredictGrid: grid prediction per time.
StSimulate: Cholesky simulation, conditioned on data by simple
kriging of residuals. Identical to the Python arm morie.fn.stgeo;
neighbour indices are 1-based here.
Usage
StFit(
dist,
timelag,
gamma,
np_,
model,
start,
fit_method = 6,
stani = NULL,
space_model = "Exp",
time_model = "Exp",
joint_model = "Exp",
lower = NULL,
upper = NULL
)
StUniversalKriging(z, X, coords, times, X0, new_coords, new_times, model)
StLocalKriging(
z,
coords,
times,
new_coords,
new_times,
model,
nmax,
stani,
buffer = 2,
beta = NULL
)
StBlockKriging(
z,
coords,
times,
new_coords,
new_times,
model,
block,
duration,
n_space = 4L,
n_time = 3L,
beta = NULL
)
StLeaveHOut(z, coords, times, model, h, tau, beta = NULL)
StKrigingDiagnostics(coords, times, new_coords, new_times, model)
StSmoothness(model)
StPredictGrid(z, coords, times, model, xs, ys, new_times, beta = NULL)
StSimulate(
coords,
times,
model,
nsim = 1L,
seed = 1,
mean = 0,
z = NULL,
data_coords = NULL,
data_times = NULL
)Arguments
- dist, timelag, gamma, np_
Empirical variogram columns.
- model
Model name (
StFit) orSTCovariancemodel list.- start
Starting parameters.
- fit_method
Weighting (1, 2, 6 or 7).
- stani
Space-time anisotropy.
- space_model, time_model, joint_model
Marginal model families.
- lower, upper
Optional bounds.
- z
Observed values.
- X, X0
Drift design matrices for data and targets.
- coords, times
Observation locations and times.
- new_coords, new_times
Target locations and times.
- nmax
Neighbourhood size.
- buffer
Candidate multiplier.
- beta
Known mean (simple kriging).
- block, duration
Spatial block side and time window.
- n_space, n_time
Block discretisation.
- h, tau
Spatial and temporal exclusion radii.
- xs, ys
Grid coordinates.
- nsim
Number of realisations.
- seed
Philox seed.
- mean
Field mean.
- data_coords, data_times
Conditioning data locations and times.
References
Graeler, B., Pebesma, E. and Heuvelink, G. (2016). Spatio-temporal interpolation using gstat. The R Journal 8(1), 204-218.
Journel, A. G. and Huijbregts, C. J. (1978). Mining Geostatistics. Academic Press.
Stein, M. L. (1999). Interpolation of Spatial Data: Some Theory for Kriging. Springer.