agriscope 0.1.1

Agricultural geospatial analysis toolkit

AgriScope-V0.1.1

Agricultural geospatial analysis toolkit extending GeoPandas.

---

Installation

pip install AgriScope

# With performance extras (Dask chunking support)
pip install "AgriScope[performance]"

Quickstart

import geopandas as gpd
import AgriScope          # registers the .agri accessor

gdf = gpd.read_file("fields.geojson")

# NDVI
gdf = gdf.agri.compute_ndvi("red", "nir")

# Soil management zones
gdf = gdf.agri.soil_zones(n_clusters=4)

# Irrigation needs
gdf = gdf.agri.irrigation_needs(evapotranspiration=5.2, rainfall=2.1)

# Summary report
print(gdf.agri.summary())

CLI

# Compute NDVI
AgriScope compute-ndvi fields.geojson --red red --nir nir

# Soil zones
AgriScope soil-zones fields_ndvi.geojson --n-clusters 4

# Irrigation needs
AgriScope irrigation-needs fields_zones.geojson --et0 5.2 --rain 2.1

# Summary
AgriScope summary fields_final.geojson

# JSON output for pipeline use
AgriScope summary fields_final.geojson --json-output | jq .ndvi

Features

Method	Description
gdf.agri.compute_ndvi(red, nir)	NDVI from column names or arrays
gdf.agri.sample_raster(path, band)	Sample raster at geometry locations
gdf.agri.soil_zones(method, n_clusters)	KMeans / DBSCAN / Agglomerative zoning
gdf.agri.irrigation_needs(et0, rain)	Water-balance irrigation recommendations
gdf.agri.compute_ndvi_chunked(...)	Chunked NDVI for very large datasets
gdf.agri.summary()	NDVI stats, zone distribution, irrigation counts

Project Structure

geopandas_agri/
├── __init__.py     # Package entry; registers accessor
├── accessor.py     # Thin routing layer (@pd.api.extensions accessor)
├── ndvi.py         # NDVI formula + raster sampling
├── soil.py         # Clustering-based soil zonation
├── irrigation.py   # Water-balance irrigation model
├── utils.py        # Shared validation + numeric helpers
└── cli.py          # Click CLI commands

Development

git clone https://github.com/your-org/AgriScope-V1
cd AgriScope
pip install -e ".[dev]"
pytest

Roadmap / Future Extensions

1. Crop classification — integrate satellite time-series (Sentinel-2 / Landsat) with a scikit-learn or LightGBM classifier to map crop types per field.

2. Yield prediction — regression models (XGBoost / neural nets) trained on historical NDVI trajectories, soil attributes, and weather data to predict per-field yield before harvest.

3. Climate data integration — gdf.agri.attach_era5(date_range) to automatically download ERA5 reanalysis data (via cdsapi) and join precipitation, temperature, and wind speed to field geometries.

License

MIT © 2024 Agricultural GeoData Team