AI-Driven Farming: Next-Generation Crop and Yield Forecasts with Advanced Deep Learning
This project develops AI-driven crop simulation and yield forecasting models using Transformer-based time series architectures and pre-trained LLMs. It integrates high-resolution exogenous inputs and enables multimodal, few-/zero-shot inference for robust long-range agricultural predictions.
Project Outline:
This project aims to leverage cutting-edge AI to revolutionise crop simulations and yield forecasts, addressing a critical gap in traditional agricultural modelling. Recent breakthroughs in deep learning—particularly with Transformer, generative AI, and Large Language Models (LLMs)—have demonstrated that state-of-the-art time series models can now outperform physics-based climate and process-based environmental models, with high accuracy even for long-range forecasts up to 720 steps ahead.
We will translate these advanced techniques to agriculture by developing robust, AI-driven models for crop simulation and yield forecasting. Our work will focus on two key areas: first, the effective incorporation of high-resolution, low-cost exogenous variables, such as 5km-resolution weekly rainfall or soil moisture ensemble forecasts; and second, the integration of a pre-trained LLM for high-accuracy, multimodal, few- or zero-shot forecasts, which are critical for scenarios with little or no training data. Ultimately, this research will provide the sophisticated forecasting tools necessary to enhance agricultural resilience and productivity in a changing climate, offering valuable support for strategic farming decisions well before a planting season begins.
From engaging in this project the student will learn how to:
In this project, the student will master advanced deep learning methods (Transformers, generative AI, and/or LLMs) and apply them to challenging long-range time series forecasting problems in climate and agriculture. They will gain experience integrating multimodal data sources, from climate and soil models to ensemble forecasts, to build robust predictive systems. Working within a multi-disciplinary ANU–CSIRO team, the student will also develop strong research communication skills while applying AI to real-world challenges, ultimately contributing to agricultural resilience and global food security.
Questions about this project can be directed to warren.jin@data61.csiro.au
To register an expression of interest, click here. You will need to outline why you have selected the research project and how your skills, experience and/or knowledge meet the project requirements.