phenologyFitter function

phenologyFitter

phenologyFitter(
  par.guess = NULL,
  modelfn = PhenoFlex_GDHwrapper,
  bloomJDays,
  SeasonList,
  control = list(smooth = FALSE, verbose = TRUE, maxit = 1000, nb.stop.improvement = 250),
  lower,
  upper,
  seed = 1235433,
  ...
)

Arguments

• par.guess: numeric vector. Initial guesses for fit parameters. This can be set to NULL, in which case GenSA choses initial parameters.
• modelfn: function. Model function which computes the index in temperatures at which blooming occures. It must have as first argument a data frame with at least the two columns Temp and JDays for one season, see SeasonList. It can have further arguments which can be passed via .... The modelfn must return a single numeric value for the predicted bloom JDay for that season. NA is an allowed return value if no blooming occures in that season. The default is the PhenoFlex with GDH as heat accumulation. Alternative is PhenoFlex_GAUSSwrapper with GAUSSian heat accumulation. But this function can also be user defined.
• bloomJDays: integer vector. vector of observed bloom JDays per year
• SeasonList: list. Must be a list of data frames, each data frame for one season. Each data.frame must at least have a column Temp with the temperature vector and JDays with the corresponding JDay vector. Can be generated by e.g. genSeasonList . length(SeasonList) must be equal to length(bloomJDays).
• control: control parameters to GenSA, see GenSA::GenSA
• lower: Vector with length of ‘par.guess’. Lower bounds for components.
• upper: Vector with length of ‘par.guess’. Upper bounds for components.
• seed: integer seed for the random number generator used by GenSA.
• ...: further parameters to be passed on to modelfn.

Returns

an object of class phenologyFit. It contains the named elements model_fit with the returned object from GenSA, par the best fit parameters, pbloomJDays the predicted bloom JDays and the inputs par.guess, modelfn, bloomJDays, lower, upper, control, SeasonList and ....

Examples

## this example does not make sense as a fit, but demonstrates
## how to use `phenologyFitter`
data(KA_weather)
data(KA_bloom)
hourtemps <- stack_hourly_temps(KA_weather, latitude=50.4)
SeasonList <- genSeasonList(hourtemps$hourtemps, years=c(2007,2008))
par <- c(40, 190, 0.5, 25, 3372.8, 9900.3, 6319.5, 5.939917e13, 4, 36, 4, 1.6)
upper <- c(41, 200, 1, 30, 4000, 10000, 7000, 6.e13, 10, 40, 10, 50)
lower <- c(38, 180, 0.1, 0, 3000, 9000, 6000, 5.e13, 0, 0, 0, 0.05)
X <- phenologyFitter(par.guess=par, bloomJDays=KA_bloom$pheno[c(24,25)], 
  SeasonList=SeasonList, lower=lower, upper=upper,
  control=list(smooth=FALSE, verbose=TRUE, maxit=10, nb.stop.improvement=5))
summary(X)
plot(X)

Author(s)

Carsten Urbach urbach@hiskp.uni-bonn.de