MVR: Multi-variate regression

In metno/esd: Climate analysis and empirical-statistical downscaling (ESD) package for monthly and daily data

Multi-variate regression

Description

MVR solves the equation

Y = \Psi X

and estimates

\Psi

by inverting the equation. Predictions give the value of Y, given this matrix and some input. MVR is useful for data where Y contains several time series where the spatial coherence/covariance is important to reproduce. For instance, Y may be a combination of stations, the two wind components from one station, or a set of different elements from a group of stations.

Usage

MVR(Y, X, SVD = TRUE, LINPACK = FALSE, verbose = FALSE)

Arguments

Y	An object with climate data: field, eof, or pca.
X	Same as Y or any zoo object.
SVD	Use a singular value decomposition as a basis for the PCA.
LINPACK	an option for svd.
verbose	a boolean; if TRUE print information about progress

Value

A CCA object: a list containing a.m, b.m, u.k, v.k, and r, describing the Canonical Correlation variates, patterns and correlations. a.m and b.m are the patterns and u.k and v.k the vectors (time evolution).

Author(s)

R.E. Benestad

Examples

## Not run: 
# Example for using EOF and MVR
slp <- slp.NCEP(lat=c(-40,40),anomaly=TRUE)
sst <- sst.NCEP(lat=c(-40,40),anomaly=TRUE)
eof.1 <- EOF(slp,mon=1)
eof.2 <- EOF(sst,mon=1)
mvr <- MVR(eof.1,eof.2)
plot(mvr)

# Example for using PCA and MVR
oslo <- station(src="NACD",loc="Oslo")
bergen <- station.nacd("Bergen")
stockholm <- station.nacd("Stockholm")
copenhagen <- station.nacd("Koebenhavn")
helsinki <- station.nacd("Helsinki")
reykjavik <- station.nacd("Stykkisholmur")
edinburgh <- station.nacd("Edinburgh")
debilt <- station.nacd("De_Bilt")
uccle <- station.nacd("Uccle")
tromso <- station.nacd("Tromsoe")
falun <- station.nacd("Falun")
stensele <- station.nacd("Stensele")
kuopio <- station.nacd("Kuopio")
valentia <- station.nacd("Valentia")
X <- combine(oslo,bergen,stockholm,copenhagen,helsinki,reykjavik,
           edinburgh,debilt,uccle,tromso,falun,stensele,kuopio,valentia)
pca <- PCA(X)
slp <- slp.NCEP(lon=c(-20,30),lat=c(30,70))
eof <- EOF(slp)
mvr <- MVR(pca,eof)
plot(mvr)

# Find the teleconnection pattern to the NAO 
data("NAOI")
data("sunspots")
data("NINO3.4")
X <- merge(NAOI,sunspots,NINO3.4,all=FALSE)

mvr <- MVR(pca,X)

# Find the pattern for NAOI:
teleconnection <- predict(mvr,newdata= c(1,0,0))
map(teleconnection,cex=2)

## End(Not run)

metno/esd documentation built on Feb. 25, 2025, 6:44 p.m.