kalmanFilter.estimateCovariance
Estimates Kalman filter covariance for a queried time step
Syntax
Description
Returns the user-specified covariance estimate for a queried time step a requested observation sites.
Estimates covariance from ensemble deviations and applies inflation, localization, and blending (in that order) as appropriate.
Input Arguments
t
scalar linear index
The index of an assimilation time step for which to estimate and return covariance
s
vector, linear indices [nQuery]
The indices of observation sites for which to estimate and return covariance.
Xdev
numeric matrix [nState x nMembers]
The ensemble deviations for the prior for the time step.
Ydev
numeric matrix [nSite x nMembers]
The ensemble deviations of the estimates for the time step. Must have one row per observation site in the filter, regardless of the number of queried sites.
Output Arguments
Knum
numeric matrix [nState x nQuery]
The covariance of the state vector elements with the queried observation sites. This covariance is the numerator of the Kalman Gain.
Ycov
numeric matrix [nQuery x nQuery]
The covariance between the queried observation sites and each other. This covariance is the Y covariance term that appears in the denominator of the Kalman gain.