Radiative transfer modelling
for vegetation & Earth observation

# Generate a small Look-Up Table (LUT) of vegetation traits
LUT <- as.data.frame(getLUT(inputs = ToolsRTM::inputsPROSAIL,
                            nLUT = 10, setseed = 123))

# Simulate canopy reflectance using PROSPECT-PRO + fourSAIL
sim <- simulate_RTM(
    inputLUT    = LUT[1, ],
    rsoil       = rep(0.15, 2101),
    leaf.model  = "PROSPECT-PRO",
    canopy.model = "fourSAIL")

plot(sim$rsot, type = "l", xlab = "Wavelength", ylab = "Reflectance")

Real output of the exact code above.

LUT <- as.data.frame(getLUT(inputs = ToolsRTM::inputsPROSAIL, nLUT = 50, setseed = 42))
rsoil <- rep(0.15, 2101)

plot(NULL, xlim = c(400, 2500), ylim = c(0, 0.5),
     xlab = "Wavelength (nm)", ylab = "Reflectance")
for (i in seq_len(50)) {
  sim <- simulate_RTM(inputLUT = LUT[i, ], rsoil = rsoil,
                       leaf.model = "PROSPECT-PRO", canopy.model = "fourSAIL")
  lines(400:2500, sim$rsot)
}

Real output: 50 simulations from a random LUT, colored by LAI (light = low, dark = high).

cab_values <- c(10, 25, 40, 55, 70)
rsoil <- rep(0.15, 2101)

plot(NULL, xlim = c(400, 2500), ylim = c(0, 0.5),
     xlab = "Wavelength (nm)", ylab = "Reflectance")
for (cab in cab_values) {
  row <- LUT[1, ]
  row$Cab <- cab
  sim <- simulate_RTM(inputLUT = row, rsoil = rsoil,
                       leaf.model = "PROSPECT-PRO", canopy.model = "fourSAIL")
  lines(400:2500, sim$rsot)
}

Real output: higher Cab lowers the visible-green reflectance; NIR/SWIR are unaffected, as expected.

opts <- read.table(system.file("input", "setoptions.csv", package = "SCOPEinR"),
                    header = TRUE, sep = ",")
lut <- read.table(system.file("input", "LUT_input.csv", package = "SCOPEinR"),
                   header = TRUE, sep = ",")

scope <- get.SCOPE(
  LUT = lut[1, ], options.SCOPE = opts, optipar = optipar2021.Pro.CX,
  leaf.model = "fluspect-CX", canopy.model = "fourSAIL",
  get.outputs = "ALL", get.plots = FALSE)

res <- scope[[1]]
plot(res$data.spectral$wlS, res$data.rad$refl, type = "l",
     xlab = "Wavelength (nm)", ylab = "Reflectance", xlim = c(400, 2400))

Real output of the exact code above.

import numpy as np
from toolsrtm import inform

# Same leaf/canopy traits as the R example, plus INFORM's explicit
# crown geometry (understorey LAI, stem density, crown diameter, tree height)
row = dict(
    N=1.5, Cab=40, Car=8, Anth=1, Cbrown=0.1, EWT=0.01, LMA=0.009, alpha=40,
    Prot=0, CBC=0,
    LAI=3, hspot=0.01, LIDFa=-0.35, LIDFb=-0.15, TypeLidf=1,
    tts=30, tto=0, psi=0,
    LAIu=2, sd=500, cd=6, h=12, skyl=0.1,
)
rsoil = np.full(2101, 0.15)

reflectance = inform(row, rsoil, leaf_model="PROSPECT-PRO")

Real output of the exact code above.

import numpy as np
from toolsrtm import inform

rng = np.random.default_rng(42)
rsoil = np.full(2101, 0.15)

for _ in range(50):
    row_i = dict(row)
    row_i.update(
        N=rng.uniform(1.2, 2.5), Cab=rng.uniform(10, 70), LAI=rng.uniform(0.5, 7.0),
        LIDFa=rng.uniform(-0.5, 0.5), sd=rng.uniform(200, 800), h=rng.uniform(8, 20),
    )
    refl = inform(row_i, rsoil, leaf_model="PROSPECT-PRO")

Real output: 50 simulations from a random LUT, colored by LAI (dark = low, light = high).

cab_values = [10, 25, 40, 55, 70]

for cab in cab_values:
    row_i = dict(row)
    row_i["Cab"] = cab
    refl_i = inform(row_i, rsoil, leaf_model="PROSPECT-PRO")
    plt.plot(wl, refl_i, label=f"Cab={cab}")

plt.xlabel("Wavelength (nm)")
plt.ylabel("Reflectance")
plt.legend()
plt.show()

Real output: same pattern as the R version — visible-green sensitivity, NIR/SWIR unaffected.

from toolsrtm.inversion import get_inversion

# dataset: a DataFrame with spectral_bands as columns and a "Cab" column
fit = get_inversion(dataset, dep_var="Cab", inputs=spectral_bands, algorithm="RF")
pred = fit.predictions["test"]

import csv
import numpy as np
from scopeinpython import ScopeOptions, get_scope
from scopeinpython.spectral import get_spectra_scope

with open("SCOPEinR/inst/input/LUT_input.csv", newline="") as f:
    row = next(csv.DictReader(f))

wl = get_spectra_scope().wlS
options = ScopeOptions(calc_fluor=True, calc_xanthophyllabs=True)

refl_curves, sif_values = [], []
for cab in np.linspace(10, 80, 20):
    row_i = dict(row)
    row_i["Cab"] = cab
    result = get_scope(row_i, options=options)
    refl_curves.append(result.rtmo.refl)      # top-of-canopy reflectance
    sif_values.append(result.rtmf.EoutF)      # sun-induced fluorescence (SIF)

Real output of the exact code above: reflectance for all 20 runs (left), SIF response to Cab (right).