The project

A shared library for the ground we depend on.

Building an open soil spectral reference library with smartphone observations, instrument spectra and laboratory evidence.

The thesis

Make soil evidence easier to collect, connect and build on.

Smartphones are the accessible capture tool: images, field context and physical sample links. Instrument spectra and verified laboratory measurements supply reference evidence. GSIP aims to connect them in an open library for researchers, communities and AI builders, with soil maps as a downstream use. Understanding the soil biome also requires biological measurements; a phone photo cannot provide them.

In development

Open source now. Public capture remains closed.

Library uploads remain closed too. Opening them requires reviewed legal terms, privacy screening, scientific checks and a separate operator decision after the launch gates pass. No opening date is promised.

Roadmap

Build the evidence plane before the prediction layer.

Now

Local library tools

Typed measurement records, a private local store, staged optical extraction and a synthetic paired-model demo. Public source and build instructions are available now.

Next

Verified paired evidence

Field samples paired with calibrated instruments and laboratory evidence, rights verification, privacy review and independent evaluation before scientific or publication claims.

Future

Open models and soil maps

Train and evaluate models on eligible evidence, publish their limits and uncertainty, and use validated outputs to inform maps and further research.

Open-source stewardship

Built by Viridis LLC with the work visible in public.

Justin Hart is the GSIP product owner. Architecture, implementation work packages, acceptance criteria, review findings, and source code are maintained in the public repository.