Machine learning for managing soil constraints

This project was completed in 2025.

WHY THIS PROJECT IS NEEDED

This project aimed to find the best ways to manage multiple soil constraints such as sodicity, acidity, and salinity helping farmers make informed soil management decisions to maximise productivity and profitability.

While there were different ways to manage constraints in isolation, it was difficult to know which method to use and when, due to high variability in the responsiveness of soils to ameliorants where multiple soil constraints exist.

In short: The project used data and a computer-based approach to predict which management and amelioration practices will work best for a particular soil, enhancing soil productivity and profitability for farmers.

Project focus

To address this, the project is developed and tested a computer-based approach (using a knowledge-guided machine learning framework) that incorporated scientific understanding and learnings from existing data. This was then used to predict which combinations of soil management will work best for a particular soil affected by multiple constraints, under specific weather and farming conditions.

This project also standardised and used data from published studies and past or current experiments conducted by Soil CRC participants to ensure that the data was findable, accessible, interoperable, and reusable.

Full project title: Knowledge-guided machine learning optimisation of soil constraint management