← RTM-Suite tutorials

Python tutorials overview

A 20-chapter, Python-first learning path through toolsrtm and scopeinpython -- from a first simulation to a real Sentinel-2 trait map. Every chapter follows the same template (concept, runnable code, real output figure, scientific interpretation) and every number/figure below was produced by actually executing the documented code -- not written by hand. This page is the five-minute visual map; the Python documentation is where each chapter lives in full, and the API reference has exact function signatures.

Tip: use your browser's Print → Save as PDF to keep an offline copy of this page.

I. Fundamentals & RTM Simulation II. Simulation to Observations III. Spectra to Traits IV. Real Earth Observation

Chapters 01–06I. Fundamentals & RTM Simulation

A first simulation in minutes, what every trait means, then all four model families -- leaf, canopy, soil/atmosphere, and SCOPE -- each run, plotted, and interpreted on real output.

PROSPECT-D vs PROSPECT-PRO vs LIBERTY leaf reflectance and transmittance

Three leaf models, same traits: PROSPECT-D/PRO agree, LIBERTY (needle) diverges (Ch. 03)

fourSAIL vs fourSAIL2 vs INFORM TOC reflectance

fourSAIL vs. fourSAIL2 vs. INFORM, same leaf and LAI (Ch. 04)

Full SCOPE TOC reflectance and emitted SIF spectrum

SCOPE: reflectance and emitted SIF from one coupled simulation (Ch. 06)

#	Chapter
01	Getting Started
02	Parameters & Traits
03	Leaf Radiative Transfer Models
04	Canopy Radiative Transfer Models
05	Soil & Atmosphere
06	SCOPE

Chapters 07–10II. From Simulation to Observations

How the models chain together into real pipelines, resampling onto real sensor bands (with the actual measured Sentinel-2A response curves, not idealized shapes), spectral indices, and a quantified, wavelength-resolved sensitivity analysis.

Sentinel-2A spectral response functions and resulting convolved band reflectances

Real, measured Sentinel-2A response curves + resulting band values (Ch. 08)

NDVI, MSAVI, and red-edge position vs LAI

NDVI's real saturation above LAI~4, vs. MSAVI's gentler response (Ch. 09)

Which trait dominates at each wavelength -- Cab (VIS), EWT+LMA (SWIR) (Ch. 10)

#	Chapter
07	Building RTM Workflows
08	Sensor Simulation
09	Spectral Indices
10	Sensitivity Analysis

Chapters 11–15III. From Spectra to Traits

Building a realistic (1000-row) training LUT, then three genuinely different ways to invert it: LUT matching, machine learning, and deep learning -- the DL chapter shows a real, successful fit (R²=0.95, healthy training/validation loss curves, early stopping), not a broken demo.

Cab Gaussian and LAI Uniform distributions, and Cab vs Car constrained scatter

Realistic trait sampling: distributions plus real trait co-variation (Ch. 11)

Observed spectrum vs best-matching LUT rows, and retrieved vs true Cab

LUT matching: observed spectrum against its best-matching neighbours (Ch. 12)

Dense neural network training/validation loss curve and predicted vs true Cab

A dense network that genuinely learns -- real loss curve, R²=0.95 (Ch. 14)

#	Chapter
11	LUT Generation
12	LUT Inversion
13	Machine-Learning Inversion
14	Deep-Learning Inversion
15	Choosing an Inversion Strategy

Chapters 16–20IV. Real Earth Observation

Retrieving a live Sentinel-2 scene via STAC, preparing it properly (including a real, verified case where naive cloud masking fails and a working fix), applying a trained inversion model spatially, and a flagship end-to-end script from raw RTM parameters to a real, cross-checked trait map.

True color Sentinel-2 image and a brightness-based cloud flag

A naive mask misses the real cloud; a verified brightness heuristic catches it (Ch. 17)

REP index map and retrieved Cab map over the real Loobos Sentinel-2 scene

Retrieved Cab and an independent index (REP), same real forest gap (Ch. 18–20)

True color, NDVI, retrieved Actot map, and seasonal time series over real Speulderbos Sentinel-2 data

NDVI vs. retrieved net photosynthesis, tracking across a real season (Ch. 19)

#	Chapter
16	Retrieving Real EO Data
17	Preparing EO Observations
18	Applying an Inversion Model Spatially
19	Trait Maps & Uncertainty
20	End-to-End Workflow (flagship)

This is a curated visual subset of the full 20-chapter path -- the complete Python documentation has every chapter in full, each with runnable code, real output, and a scientific interpretation, and the toolsrtm / scopeinpython API references have exact function signatures. Using R? See the equivalent R tutorials overview.