Skip to contents

Build a LUT with a per-trait distribution choice (Uniform or Gaussian) and an optional Car~Cab correlation

get_distributionLUT.Rd

Alternative to getLUT for when different traits need different sampling distributions in the same LUT (e.g. LAI sampled Uniform while Cab is sampled Gaussian), rather than one distribution for every trait. Gaussian traits are drawn by truncated-normal rejection sampling (gauss_byMin_Max), so they still respect minval/maxval bounds.

Usage

get_distributionLUT(
  minval = NULL,
  maxval = NULL,
  nSamples = NULL,
  TypeDistrib = NULL,
  Mean_gauss = NULL,
  Std_gauss = NULL,
  DepCab = NULL,
  setseed = NULL
)

Arguments

minval

one-row data.frame/list, column names = trait names, minimum value per trait.

maxval

one-row data.frame/list, column names = trait names, maximum value per trait.

nSamples

numeric. Number of LUT rows to generate.

TypeDistrib

named list, one entry per trait in minval, each either "Uniform" or "Gaussian".

Mean_gauss

one-row data.frame/list, mean per trait – only read for traits where TypeDistrib is "Gaussian".

Std_gauss

one-row data.frame/list, standard deviation per trait – only read for traits where TypeDistrib is "Gaussian".

DepCab

logical. If TRUE and "Car" is one of the traits, Car is not drawn independently – it's redrawn as correlatedValue(Cab/4, r = 0.8), the empirical Cab-Car co-variation seen in leaf pigment data.

setseed

integer. Random seed.

Value

LUT as a data.frame, one column per trait in minval.

Examples

minv <- data.frame(Cab = 10, Car = 2, LAI = 0.5)
maxv <- data.frame(Cab = 80, Car = 20, LAI = 7)
distrib <- list(Cab = "Gaussian", Car = "Uniform", LAI = "Uniform")
meang <- data.frame(Cab = 40, Car = NA, LAI = NA)
stdg  <- data.frame(Cab = 15, Car = NA, LAI = NA)
LUT <- get_distributionLUT(minval = minv, maxval = maxv, nSamples = 100,
                            TypeDistrib = distrib, Mean_gauss = meang, Std_gauss = stdg,
                            DepCab = TRUE, setseed = 1)