dathomog {seas}

Homogenize daily data sets

Description

Homogenizes daily data from two data sets into one data set; optionally show cross-plots to examine how well correlated that data sets are.

Usage

dathomog(x1, x2, by = "date", plot = FALSE)

Arguments

x1

a data.frame of seasonal data; 1st selection

x2

a data.frame or seasonal data; 2nd selection

by

name of common column, usually ‘date’, which is of class Date

plot

logical; produce cross-plots and correlation statistics of the variables between the two data sets

Details

Data from x1 has priority over x2. Where data from x1 is either NA or missing (outside of time range), data from x2 will be used (if available). Otherwise data form x1 will be used directly. Variables will be homogenized where their names are identical, found using names.

The cross-plots of the data are shown only for interest. They show useful correlation statistics, and a best-fit line using perpendicular offsets (which are preferred in this case over traditional linear regression). At some point, the equations for this line may be used to adjust the values from x2, however this can always be done externally to this function by pre-processing x2.

Value

Returns a data.frame of seasonal data required by most functions in seas. Variable names of the structure are found by a union of the names of x1 and x2.

Warning

Weather stations should be sufficiently close enough to approximate the same weather. This distance depends on the spatial distance and local climatology.

Author(s)

Mike Toews

References

http://mathworld.wolfram.com/LeastSquaresFittingPerpendicularOffsets.html

Examples

data(mscdata)
dat1 <- mksub(mscdata, id=2100630)
dat2 <- mksub(mscdata, id=1108447)
year.plot(dat1)
year.plot(dat2)
newdata <- dathomog(dat1, dat2)
year.plot(newdata)
message(paste(c("This is a rather poor example, since the",
                "two stations are nowhere near each other"),
                collapse="\n"))