run_model {ShellChron}

Function that optimizes sinusoid parameters to fit d18O data

Description

The second core function of the ShellChron growth model. Loops through all data windows and uses the growth_model function to create d18O series that match the input data. This step is iterated and optimized (minimizing the Sum of Squared Residuals) through the SCEUA algorithm (by Duan et al., 1992) which finds the optimal input parameters to the growth rate and Sea Surface Temperature (SST) sinusoids to simulate d18O data.

Usage

run_model(
  dat,
  dynwindow,
  transfer_function = "KimONeil97",
  d18Ow = 0,
  T_per = 365,
  G_per = 365,
  t_int = 1,
  t_maxtemp = 182.5,
  SCEUApar = c(1, 25, 10000, 5, 0.01, 0.01),
  sinfit = TRUE,
  MC = 1000,
  plot = FALSE
)

Arguments

dat

Matrix containing the input data

dynwindow

Information on the position and length of modeling windows

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)

T_per

Period of SST sinusoid (in days; default = 365)

G_per

Period of growth rate sinusoid (in days; default = 365)

t_int

Time interval (in days; default = 1)

t_maxtemp

Timing of the warmest day of the year (in julian day; default = 182.5, or May 26th halfway through the year)

SCEUApar

Parameters for SCEUA optimization (iniflg, ngs, maxn, kstop pcento, peps). For details, refer to Duan et al. (1992) in references

sinfit

Apply sinusoidal fitting to guess initial parameters for SCEUA optimization? TRUE/FALSE

MC

Number of Monte Carlo simulations to apply for error propagation Default = 1000

plot

Should results of modeling be plotted? TRUE/FALSE

Value

A list containing the resultarray which contains the full result of all simulations on each data window and the parmat listing all optimized growth rate and SST parameters used to model d18O in each data window

References

package dependencies: ggplot2 3.2.1; rtop 0.5.14 Function dependencies: sinreg, d18O_model, growth_model

doi: 10.1029/91WR02985

See Also

Duan, Qingyun, Soroosh Sorooshian, and Vijai Gupta. "Effective and efficient global optimization for conceptual rainfall runoff models." Water resources research 28.4 (1992): 1015-1031. https://doi.org/10.1029/91WR02985

Examples

# Create dummy input data column by column
dat <- as.data.frame(seq(1000, 40000, 1000))
colnames(dat) <- "D"
dat$d18Oc <- sin((2 * pi * (seq(1, 40, 1) - 8 + 7 / 4)) / 7)
dat$YEARMARKER <- c(0, rep(c(0, 0, 0, 0, 0, 0, 1), 5), 0, 0, 0, 0)
dat$D_err <- rep(100, 40)
dat$d18Oc_err <- rep(0.1, 40)
# Create dummy dynwindow data
dynwindow <- as.data.frame(seq(1, 29, 2))
colnames(dynwindow) <- "x"
dynwindow$y <- rep(12, 15)
# Run model function
resultlist <- run_model(dat = dat,
    dynwindow = dynwindow,
    transfer_function = "KimONeil97",
    d18Ow = 0,
    T_per = 365,
    G_per = 365,
    t_int = 1,
    t_maxtemp = 182.5,
    SCEUApar = c(1, 25, 10000, 5, 0.01, 0.01),
    sinfit = TRUE,
    MC = 1000,
    plot = FALSE)