SatDataKit

Unified satellite data analysis toolkit — one API for all Earth Observation formats.

What is SatDataKit?

SatDataKit solves a real problem in the Earth Observation community: every satellite data format has its own API, forcing scientists to learn GDAL, NetCDF4, h5py, rasterio, and Sentinel-specific tools just to read a single image.

SatDataKit unifies all of that into one clean API:

Built on the same stack NASA uses (xarray, rioxarray, rasterio, netCDF4, h5py) but with a unified abstraction layer that eliminates boilerplate.

Quick Start

from satdatakit import read, compute_index, Pipeline

# Read any format
ds = read("sentinel2.tif")  # GeoTIFF, NetCDF, HDF, SAFE

# Compute indices
ds = compute_index(ds, "NDVI")

# Pipeline
result = (
    Pipeline()
    .read("data.tif")
    .reproject("EPSG:4326")
    .resample(30)
    .compute_index("NDVI")
    .to_geotiff("output.tif")
)

---
## Installation

# Docker (recommended)
docker-compose up --build satdatakit

# Or Conda
conda env create -f environment.yml
conda activate satdatakit
pip install -e ".[dev]"

## Optional Extensions

SatDataKit core supports GeoTIFF, NetCDF, HDF5, and SAFE out of the box.

For large-scale processing, install optional extensions:

```bash
# Parallel processing with Dask
pip install satdatakit[dask]

# Cloud catalogs (STAC) + Zarr format
pip install satdatakit[cloud]

# Everything (production servers)
pip install satdatakit[full]

| Extension | Command                         | Use Case                                 |
| --------- | ------------------------------- | ---------------------------------------- |
| **Dask**  | `pip install satdatakit[dask]`  | 10+ files, lazy chunks, parallel compute |
| **STAC**  | `pip install satdatakit[stac]`  | Search cloud catalogs (AWS, Copernicus)  |
| **Zarr**  | `pip install satdatakit[zarr]`  | Cloud-native format, chunked storage     |
| **Cloud** | `pip install satdatakit[cloud]` | STAC + Zarr + S3 access                  |
| **Full**  | `pip install satdatakit[full]`  | All extensions (servers, production)     |


### Dask Example

from satdatakit.extensions.dask_ext import enable_dask, read_dask

enable_dask()

# Lazy load with chunks
ds = read_dask(["file1.tif", "file2.tif"], chunks={"x": 1024})

# Compute when ready
ds = ds.compute()


## Features

Unified API: One read() for GeoTIFF, NetCDF, HDF, SAFE
Spectral Indices: NDVI, NDWI, EVI, SAVI, and more
Pipeline API: Fluent, chainable operations
Time Series: Stack multiple scenes automatically


## License

MIT License — see LICENSE for details.
Author: Rafael Cañete Vazquez