kalmanFilter.uncertainties

Set or return the proxy uncertainties for a Kalman filter

Syntax



obj = obj.uncertainties(Rvar)


obj = obj.uncertainties(Rcov)


obj = obj.uncertainties(R, whichR)


obj = obj.uncertainties(R, whichR, type)


obj = obj.uncertainties(R, whichR, “c”|”cov”|”covariance”|true)


obj = obj.uncertainties(R, whichR, “v”|”var”|”variance”|false)


[Rvar, whichR] = obj.uncertainties


[Rcov, whichR] = obj.uncertainties


obj = obj.uncertainties(‘delete’)

Description



obj = obj.uncertainties(Rvar)

Provides error variances to the Kalman filter object. Overwrites any previous set of uncertainties. Each row holds the uncertainties for a particular site. Each column holds a unique set of error variances. If Rvar has a single column, uses the same set of error variances in all assimilation time steps. If the number of sets of variances matches the number of time steps, uses the indicated set of variances in each time step. For a given set of variances, a NaN value is permitted for a site if the site has no observations in that time step.



obj = obj.uncertainties(Rcov)

Provides error covariances to the Kalman filter object. Overwrites any previous set of uncertainties. Rcov should have one row and one column for each observation site. Each element along the third dimension holds a unique set of covariances. Each set of covariances must be a symmetric matrix. If there is a single set of covariances, uses the same covariances in all assimilation time steps. If the number of sets of covariances matches the number of time steps, uses the indicated set of covariances in each time step. For a given set of covariances, NaN values are permitted between two sites if at least one of the two sites is missing an observation in that time step.



obj = obj.uncertainties(R, whichR)

Specify which set of R variances or covariances to use in each assimilation time steps. This syntax allows the number of sets of R values to differ from the number of time steps.



obj = obj.uncertainties(R, whichR, type)


obj = obj.uncertainties(R, whichR, “c”|”cov”|”covariance”|true)


obj = obj.uncertainties(R, whichR, “v”|”var”|”variance”|false)

Indicate the type of uncertainties being provided to the particle filter. If “c”|”cov”|”covariance”|true, treats the input uncertainties as covariances. If “v”|”var”|”variance”|false, treats the uncertainties as variances.



[Rvar, whichR] = obj.uncertainties


[Rcov, whichR] = obj.uncertainties

Returns the current R uncertainties for the Kalman filter object, and indicates which set of R variances or covariances is used in each assimilation time step.



obj = obj.uncertainties(‘delete’)

Deletes any current uncertainties from the Kalman filter object.

Input Arguments

Rvar

numeric matrix [nSite x 1|*nTime*|*nR*]
Error variances for the proxies. A numeric matrix with one row per site. Each column is a unique set of variances. If there is a single column, uses the same set of error variances in all assimilation time steps. If the number of sets of variances matches the number of time steps, uses the indicated set of variances in each time step. If the number of sets of variances is neither 1 nor the number of time steps, use the whichR input to indicate which set of variances to use in each assimilation time step. For a given set of variances, a NaN value is permitted for a site if the site has no observations in all time steps associated with that set of error variances.

Rcov

numeric array [nSite x nSite x 1|*nTime*|*nR*]
Error covariances for the proxies. A numeric array with one row and one column per site. Each element along the third dimension is a unique set of covariances. Each set of covariances must be a symmetric matrix. If there is a single set of covariances (i.e. Rcov is a matrix), uses the same set of covariances for all assimilation time step. If the number of sets of covariances matches the number of time steps, uses the indicated set of covariances in each time step. If the number of sets of covariances is neither 1 nor the number of time steps, use the whichR input to indicate which set of covariances to use in each assimilated time step. For a given set of covariances, NaN values are permitted between two sites if at least one of the sites is missing an observation in all time steps associated with the that set of covariances.

whichR

vector, positive integers [nTime]
Indicates which set of R variances or covariances to use in each time step. Must have one element per assimilation time step. Each element of whichR is the index of a set of R variances or covariances. If using R variances, these indices are for the columns of Rvar. If using R covariances, these indices are for the thrid dimension of Rcov.

type

string scalar | scalar logical
Indicates whether the uncertainties are variances or covariances

[“v”|”var”|”variance”|false]: Error variances

[“c”|”cov”|”covariance”|true]: Error covariances

Output Arguments

obj

scalar kalmanFilter object
The kalmanFilter object with updated R uncertainties

Rvar

numeric matrix [nSite x 1|*nTime*|*nR*]
The current error variances for a kalmanFilter object. If you have not provided R uncertainties, an empty array.

Rcov

numeric array [nSite x nSite x 1|*nTime*|*nR*]
The current error covariances for the kalmanFilter object. If you have not provided R uncertainties, an empty array.

whichR

vector, positive integers [nTime] | []
Indicates which set of R variances or covariances is used in each assimilation time step. If there is a single set of R variances or covariances, returns an empty array.