Quickstart: Generate a Gobabeb Scene

This example walks through generating a 3D scene of the Gobabeb site in the Namib Desert, Namibia. You will create a scene configuration, run the generation pipeline, and produce the meshes, textures, and scene description YAML that feed into the simulator.

By the end you will have a gobabeb.yml scene description file ready for radiative transfer simulation.

---

Prerequisites

• Pixi installed and the s2gos-generator environment activated.
• A s2gos_settings.yaml file in your working directory with DEM and landcover data paths configured. See the Configuration page for details.

---

Step 1 — Create a scene configuration

Use create_scene_config() to define the area of interest around the Gobabeb site. Then attach a buffer zone, a background zone, and a molecular atmosphere.

from s2gos_generator import create_scene_config
from s2gos_generator.core.config import (
    BufferConfig,
    BackgroundConfig,
    MolecularAtmosphereConfig,
    ThermophysicalConfig,
    AbsorptionDatabase,
)

# Target area: 10 km around Gobabeb at 10 m resolution
config = create_scene_config(
    scene_name="gobabeb",
    center_lat=-23.6015,
    center_lon=15.1259,
    aoi_size_km=10,
    output_dir="./gen_output",
    target_resolution_m=10.0,
    description="Gobabeb PICS site in the Namib Desert.",
)

# Buffer zone — surrounds the target for realistic boundary conditions
config.buffer = BufferConfig(size_km=60.0, resolution_m=60.0)

# Background zone — coarse terrain extending to the horizon
config.background = BackgroundConfig(
    size_km=150.0, resolution_m=200.0, elevation=0.0
)

# Molecular atmosphere (US Standard, GECKO absorption database)
config.set_atmosphere_molecular(
    MolecularAtmosphereConfig(
        thermoprops=ThermophysicalConfig(identifier="afgl_1986-us_standard"),
        absorption_database=AbsorptionDatabase.GECKO,
        has_absorption=True,
        has_scattering=True,
    )
)

Parameter	Value	Meaning
aoi_size_km	10	10 km × 10 km target area
target_resolution_m	10.0	One mesh cell per 10 m
buffer.size_km	60.0	60 km buffer at 60 m resolution
background.size_km	150.0	150 km background at 200 m resolution

---

Step 2 — Save the configuration

Persist the configuration to JSON so you can reproduce or share the run later.

from pathlib import Path

config_dir = Path("./gen_config")
config_dir.mkdir(exist_ok=True)

config.to_json(config_dir / "gobabeb_gen_config.json")

---

Step 3 — Run the generation pipeline

Create a SceneGenerationPipeline from the configuration and execute it. The pipeline fetches DEM and landcover data, builds terrain meshes and textures, and writes the final scene description.

from s2gos_generator import SceneGenerationPipeline

pipeline = SceneGenerationPipeline(config)
pipeline.run()

Depending on the size of the area and your data source, this step may take several minutes.

---

Step 4 — Inspect the output

After the pipeline completes, the output directory contains:

gen_output/gobabeb/
├── gobabeb.yml            # Scene description (input for the simulator)
├── meshes/                # PLY terrain meshes (target, buffer, background)
├── textures/              # PNG landcover textures
└── data/                  # Intermediate data like crops of the DEM

The key output is gobabeb.yml — a YAML file describing every mesh, material, and texture in the scene. This is the file you pass to the simulator.

---

Next step

To simulate an RGB observation over this scene, continue with Quickstart: Simulate an RGB Observation.