R: Visualise systematic model errors against covariates

event.xyplot.hydromad {hydromad}

R Documentation

Visualise systematic model errors against covariates

Description

Visualise systematic model errors against covariates.

Usage

## S3 method for class 'hydromad.runlist':
## S3 method for class 'hydromad'
event.xyplot(x, events,
             formula =
             ~ log2(e(Q,mean)+.01) +
               log2(e(lag(Q,-2),first)+.01) +
               log2(e(U,max)+.01) +
               e(E,mean),
             extract = residuals,
             with.U = TRUE,
             ...,
             panel = panel.superpose,
             panel.groups = function(x, y, ...) {
                 panel.xyplot(x, y, ...)
                 library(mgcv)
                 panel.smoother(x, y, y ~ s(x), method = "gam", ...)
             },
             abline = list(h = 0), pch = ".", 
             ylab = "residual flow sums in event windows (mm)",
             data = NULL)

Arguments

`x`	a `hydromad` or `hydromad.runlist` object.
`events`	event sequence produced by `eventseq`, or a vector defining continguous groups on the specified variables.
`formula`	formula defining the covariates to plot, as passed to the formula method of `event.xyplot`. It may refer to any of the variables in the model data frame (`observed(x, select = TRUE)`); additionally it may refer to the value `U`, for the effective rainfall series derived from the model. Finally `julian` may be referred to (from data time index).
`extract`	a function to apply to `x` to extract the response variable; by default this is `residuals` but could be e.g. `fitted` or `function(x) residuals(x, boxcox = TRUE)`.
`with.U`	to include modelled effective rainfall `U` as a covariate.
`...`	further arguments passed to `event.xyplot` and on to `xyplot` and the panel function.
`panel, panel.groups, abline, pch, ylab`	passed to `xyplot`.
`data`	ignored.

Value

this function returns a trellis object which can be plotted.

Author(s)

Felix Andrews felix@nfrac.org

Examples

data(Cotter)
x <- Cotter[1:1000,]
mod <- hydromad(x, sma = "scalar",
         routing = "armax", rfit = list("sriv", order = c(2,1)))
ev <- eventseq(x$P, thresh = 3, inthresh = 1, indur = 5)
event.xyplot(mod, events = ev)
event.xyplot(mod, events = ev,
    extract = function(x) residuals(x, boxcox = TRUE))

foo <- event.xyplot(mod, events = ev,
    ~ sqrt(e(P,max)) + sqrt(e(rollmean(lag(P,-1),20,align="left"),first)))
dimnames(foo)[[1]] <- c("sqrt. peak rain (mm/day)", "mean 20-day ante. rain")
foo

[Package hydromad version 0.9-18 Index]