phenopix (version 2.4.5)

KlostermanFit: Fit a double logistic equation to a vector according to Klosterman et al. (2014)

Description

This function fits a double logistic curve to observed values using the function as described in klosterman et al. (2014), eq 7.

Usage

KlostermanFit(ts, which = "light", uncert = FALSE, nrep = 100, 
ncores='all', sf=quantile(ts, probs=c(0.05, 0.95), na.rm=TRUE))

Value

A list containing the following items.

fit

A list as returned by the function FitDoubleLogGu

uncertainty

A list containing a zoo data.frame with the uncertainty predicted values, and a dataframe containing the respective uncertainty curve parameters

Arguments

ts

A ts or zoo object with gcc data. index(ts) must be numeric days of year (doys) or a POSIXct vector

which

A character to be chosen between 'light' (default) and 'heavy'. See details.

uncert

Should uncertainty be estimated?

nrep

Number of relications to estimate uncertainty, defaults to 100.

ncores

Number of processors to be used in parallel computation, defaults to 'all' which will accidentally slow down any other activity on your computer. Otherwise set the number of processors you want to use in parallelization.

sf

Scaling factors required to normalize the data prior to the fitting. If the function is called by e.g. greenProcess sf is automatically calculated.

Author

Gianluca Filippa <[e-mail]>

Details

The function estimates parameters of the double logistic equation from Klosterman et al. 2014. Two optimization procedures are available. If which='light' (the default) equation parameters are optimized using the function optim and computation is faster, whereas if which='heavy' the optimization procedure calls the function nsl and is based on a greater number of iteractions with different first guesses for parameters. This option is about ten times slower than the light one.

References

Klosterman ST, Hufkens K, Gray JM, Melaas E, Sonnentag O, Lavine I, Mitchell L, Norman R, Friedl MA, Richardson A D (2014) Evaluating remote sensing of deciduous forest phenology at multiple spatial scales using PhenoCam imagery, Biogeosciences, 11, 4305-4320, doi:10.5194/bg-11-4305-2014.

See Also

FitDoubleLogKlLight FitDoubleLogKlHeavy

Examples

if (FALSE) {
library(zoo)
data(bartlett2009.filtered)
## fit without uncertainty estimation
fitted.kl1 <- KlostermanFit(bartlett2009.filtered, which='light')
fitted.kl2 <- KlostermanFit(bartlett2009.filtered, which='heavy')
## check fitting
plot(bartlett2009.filtered)
lines(fitted.kl1$fit$predicted, col='red')
lines(fitted.kl2$fit$predicted, col='blue')
legend('topleft',col=c('red', 'blue'), lty=1, 
  legend=c('light', 'heavy'), bty='n')
}