growth_model: Function that models a d18O curve through SST and GR...

In nielsjdewinter/ShellChron: Builds Chronologies from Oxygen Isotope Profiles in Shells

Function that models a d18O curve through SST and GR sinusoids

Description

The core function of the ShellChron growth model. Uses growth rate and SST (Sea Surface Temperature) sinusoids to model d18O data to be matched with the input. In the ShellChron modeling routine, this function is optimized using the SCEUA algorithm and applied on sliding windows through the dataset to estimate the age of each datapoint

Usage

growth_model(
  pars,
  T_per = 365,
  G_per = 365,
  years = 1,
  t_int = 1,
  transfer_function = "KimONeil97",
  d18Ow = "default",
  Dsam,
  Osam,
  t_maxtemp = 182.5,
  plot = FALSE,
  MC = 1000,
  D_err = NULL,
  O_err = NULL,
  return = "SSR"
)

Arguments

pars	List of parameters for temperature and growth rate sinusoids pars <- c(T_amp, T_pha, T_av, G_amp, G_pha, G_av, G_skw)
T_per	Period of SST sinusoid (in days; default = 365)
G_per	Period of growth rate sinusoid (in days; default = 365)
years	Number of years to be modeled (default = 1)
t_int	Time interval (in days; default = 1)
transfer_function	Transfer function used to convert d18Oc to temperature data.
d18Ow	Either a single value (constant d18Ow) or a vector of length equal to the period in SST data (365 days by default) containing information about seasonality in d18Ow. Defaults to constant d18Ow of 0 permille VSMOW (the modern mean ocean value)
Dsam	Vector of D values serving as input (keep unit consistent throughout model)
Osam	Vector of d18Oc values serving as input (in permille VPDB)
t_maxtemp	Timing of the warmest day of the year (in julian day; default = 182.5, or May 26th halfway through the year)
plot	Should results of modeling be plotted? TRUE/FALSE
MC	Number of Monte Carlo simulations to apply for error propagation Default = 1000
D_err	OPTIONAL: Vector containing errors on Dsam
O_err	OPTIONAL: Vector containing errors on Osam
return	String indicating whether to return just the Sum of Squared Residuals ("SSR") or a matrix containing the results of the model and the propagated uncertainties (if applicable)

Value

Depending on the value of the "return" parameter either a single value representing the Sum of Squared Residuals ("SSR") as a measure for the closeness of the match between modeled d18O and input values, or a matrix containing the full result of the modeling including propagated uncertainties if applicable.

References

package dependencies: ggplot2 3.2.1 function dependencies: temperature_curve, d18O_model, growth_rate_curve, mc_err_orth

doi: 10.1016/j.palaeo.2017.09.034

Examples

# Set parameters
G_amp <- 20
G_per <- 365
G_pha <- 100
G_av <- 15
G_skw <- 70
T_amp <- 20
T_per <- 365
T_pha <- 150
T_av <- 15
pars <- c(T_amp, T_pha, T_av, G_amp, G_pha, G_av, G_skw)
d18Ow <- 0
# Create dummy data
Dsam <- seq(1, 40, 1)
Osam <- sin((2 * pi * (seq(1, 40, 1) - 8 + 30 / 4)) / 30)
# Test returning residual sum of squares for optimization
SSR <- growth_model(pars, T_per, G_per, Dsam = Dsam, Osam = Osam,
    return = "SSR")
# Test returning full model result
resmat <- growth_model(pars, T_per, G_per, Dsam = Dsam, Osam = Osam,
    return = "result")

nielsjdewinter/ShellChron documentation built on Aug. 21, 2022, 4:31 p.m.