ToolsRTM
  • Home
  • R Tutorials
  • Python Tutorials
  • ← Back to RTM-Suite

Python Tutorials

TOOLSRTM · PYTHON

Python Tutorials

toolsrtm and scopeinpython share one continuous 20-chapter learning path rather than two separate ones — almost every chapter here is ToolsRTM material (only Chapter 06 is SCOPE-specific, covered on the SCOPEinR hub). Below: the ToolsRTM-relevant chapters, grouped the same way as the full path, plus the package-tagged runnable code examples.

I. Fundamentals & RTM simulation · ch01–05
A first simulation, what every trait means, then leaf, canopy and soil/atmosphere model families, each run and interpreted on real output.
01

Getting Started

A first simulation in minutes.

02

Parameters & Traits

What every trait means, physically.

03

Leaf Radiative Transfer Models

PROSPECT-D/PRO agree; LIBERTY (needle) diverges.

04

Canopy Radiative Transfer Models

fourSAIL vs. fourSAIL2 vs. INFORM, same leaf and LAI.

05

Soil & Atmosphere

BSM, MARMIT, and the SPART chain to top-of-atmosphere.

II. Simulation to observations · ch07–10
Chaining models into pipelines, resampling onto real sensor bands, spectral indices, and sensitivity.
07

Building RTM Workflows

How the models chain together into real pipelines.

08

Sensor Simulation

Real, measured Sentinel-2A response curves, not idealized shapes.

09

Spectral Indices

NDVI's real saturation above LAI~4, vs. MSAVI's gentler response.

10

Sensitivity Analysis

Which trait dominates at each wavelength.

III. Spectra to traits · ch11–15
A realistic training LUT, then three genuinely different inversion strategies.
11

LUT Generation

Realistic trait sampling, including real trait co-variation.

12

LUT Inversion

Observed spectrum against its best-matching neighbours.

13

Machine-Learning Inversion

Random Forest and friends, trained on the LUT.

14

Deep-Learning Inversion

A dense network that genuinely learns: real loss curve, R²=0.95.

15

Choosing an Inversion Strategy

LUT matching vs. ML vs. deep learning: when to use which.

IV. Real Earth observation · ch16–20
A live Sentinel-2 scene via STAC, prepared, inverted spatially, and a flagship end-to-end script.
16

Retrieving Real EO Data

Pulling a real Sentinel-2 scene via STAC.

17

Preparing EO Observations

A real case where naive cloud masking fails, and a working fix.

18

Applying an Inversion Model Spatially

From a trained model to a pixel-by-pixel trait map.

19

Trait Maps & Uncertainty

Not just a map — how confident is it?

20

End-to-End Workflow (flagship)

Raw RTM parameters to a real, cross-checked trait map.

Runnable code examples, by topic
Short, package-tagged examples from the Python API reference — each one runnable on its own.
Ex

Leaf → Canopy

PROSPECT + fourSAIL, chained.

Ex

Alternative Leaf Models

LIBERTY and Fluspect-B.

Ex

INFORM

Explicit forest canopy model.

Ex

SPART

Full soil-plant-atmosphere chain.

Ex

MARMIT

Soil moisture model.

Ex

Sensor Convolution & Indices

Native spectrum to real bands to NDVI.

Ex

Global Sensitivity Analysis

Sobol indices, in code.

Ex

Machine-Learning Trait Inversion

Training and applying an inversion model.

Ex

Deep-Learning Trait Inversion

Dense network, minimal example.

Ex

1D-CNN on Hyperspectral Bands

PRISMA-resolution spectra, direct.

Ex

Real Sentinel-2 Capstone

Data-driven spatial index & Cab mapping.

The full 20-chapter path in context lives on the RTM-Suite Python tutorials overview; the complete API reference is the toolsrtm Sphinx site.

RTM-Suite · ToolsRTM tutorials & docs

 

R and Python kept numerically in sync