CropForge

Open-source, code-first virtual farm runtime for agricultural researchers.

CropForge is a Python 3.12+ simulation engine that lets crop science researchers define, run, and visualise virtual farm experiments as pure Python code.

Key Features

Code-first: Define crops, fields, and management events in Python — no configuration files, no GUIs required.
Opt-In Physics (v0.2.0): Built-in, mathematically verified FAO-56 ET0 and root impedance physics that seamlessly integrate with custom scripts.
Multi-Field Dashboard (v0.2.0): Compare divergent physical environments and GxE scenarios directly in the UI with the new Field Selector.
Decoupled architecture: The time-stepping engine runs headless to completion; the visual dashboard reads the Parquet log afterward.
Binary streaming: Three.js 3D viewport receives Float32Array buffers directly — no JSON overhead for 3D data.
Phase-ordered steps: Attach model steps to any phase (environment, plant, soil, output) using the @step decorator; the runtime enforces execution order.
Interactive dashboard: FastAPI + Plotly Dash dashboard with 3D instanced plant renderer, time-series metrics, event log, and per-plant inspector.

Quick Start

pip install cropforge

from cropforge import Farm, Field, Crop
from cropforge.loaders import Weather, Soil

wheat = Crop(species="wheat", variety="HD-2967", sowing_doy=300)
field = Field("Plot A", rows=20, cols=30, crop=wheat,
              spacing_m=0.2, weather=Weather.from_csv("weather.csv"),
              soil=Soil.from_csv("soil.csv", apply="uniform"))

farm = Farm("WheatBasic")
farm.add_field(field)
farm.run(days=90)
farm.visualize()      # opens http://localhost:7860

CropForge

Key Features

Quick Start

Documentation

Licence