01. Getting Started with ToolsRTM

library(ToolsRTM)

The goal of this page is the smallest possible example: produce one vegetation reflectance spectrum with the minimum configuration required. Everything else in this tutorial series (LUT inversion, parallel simulation, machine learning) builds on the four-step chain below.

Vegetation parameters
        |
        v
     PROSPECT
        |
        v
Leaf reflectance / transmittance
        |
        v
     fourSAIL
        |
        v
  Canopy reflectance

1. Leaf traits and canopy structure: one row of parameters

A simulation always starts from one row of trait values – a data.frame with one column per parameter the leaf and canopy model need. Here, PROSPECT-D needs the leaf biochemistry (N, Cab, Car, Anth, Cbrown, EWT, LMA); fourSAIL needs the canopy structure (LAI, LIDFa/LIDFb/TypeLidf, hspot) and the sun-view geometry (tts/tto/psi):

row <- data.frame(
  # Canopy structure and geometry (fourSAIL)
  LAI = 3, hspot = 0.01, LIDFa = -0.35, LIDFb = -0.15, TypeLidf = 1,
  tts = 30, tto = 0, psi = 0,
  # Leaf biochemistry (PROSPECT-D)
  N = 1.5, Cab = 40, Car = 8, Anth = 2, Cbrown = 0, EWT = 0.009, LMA = 0.009,
  alpha = 40
)

A soil spectrum is also required – foursail() mixes leaf optics with whatever shows through the canopy gaps. A 50/50 blend of the package’s own bundled dry/wet reference soil spectra is a reasonable, physically-motivated default (Tutorial 05 covers building LUTs properly; Tutorial 10’s sibling article on soil/MARMIT integration covers a moisture-realistic alternative):

rsoil <- 0.5 * dataSpec_PDB[, 11] + 0.5 * dataSpec_PDB[, 12]  # dry/wet soil blend

2. PROSPECT: leaf reflectance and transmittance

foursail() runs PROSPECT internally (via its LeafModel argument) and returns the canopy-level outputs directly – there’s no need to call the leaf model on its own for this minimal example. LeafModel = "PROSPECT-D" selects the leaf model; Tutorial 02 shows every leaf model available and how to call PROSPECT on its own when only leaf-level output is needed.

3. fourSAIL: canopy reflectance

sail <- foursail(inputLUT = row, rsoil = rsoil, LeafModel = "PROSPECT-D")
names(sail)
#> [1] "rdot" "rsot" "rddt" "rsdt"

sail$rdot (hemispherical-directional reflectance factor) and sail$rsot (bi-directional reflectance factor) are fourSAIL’s two raw outputs. Compute_BRF() blends them using the sun/view geometry into the single top-of-canopy (TOC) reflectance spectrum a sensor actually observes:

reflectance <- Compute_BRF(rdot = sail$rdot, rsot = sail$rsot,
                            tts = row$tts, data.light = dataSpec_PDB)
length(reflectance)  # one value per nm, 400-2500nm
#> [1] 2101

4. Plot the result

plot(dataSpec_PDB[, 1], reflectance, type = "l", col = "#2E8B57", lwd = 2,
     xlab = "Wavelength (nm)", ylab = "Reflectance",
     main = "One simulation: fourSAIL + PROSPECT-D")

The familiar vegetation signature is already visible: low reflectance in the visible (400-700nm, chlorophyll absorption), a sharp rise at the red edge (~700-750nm), a high NIR plateau (leaf/canopy structural scattering), and two water-absorption dips in the SWIR (~1450nm, ~1950nm).

What’s next

This page deliberately stopped at one simulation. The rest of this series builds outward from here:

• Tutorial 02 – every leaf model (PROSPECT-D/-PRO, Fluspect, Liberty) and every canopy model (fourSAIL, fourSAIL2, INFORM), and how trait changes propagate from leaf to canopy.
• Tutorial 03 – SPART, when a single TOC simulation like this one isn’t enough and you need the full soil-plant-atmosphere chain to top-of-atmosphere.
• Tutorial 05 – building a LUT (many rows of trait values at once) instead of one hand-written row.