Using machine learning and process-based crop modelling for regional scale prediction

Aug 2, 2026

This thesis evaluates machine learning techniques against process-based crop growth models for predicting the impacts of climate variability on crops at the regional scale. It benchmarks leading machine learning methods for crop yield prediction against the process-based model GLAM, examining data requirements, conditions for advantage, and performance for forecasting crop failures. It also quantifies how climate-data errors affect both approaches and contrasts model sensitivity to climatic drivers to guide improved simulations. Results show complementary strengths: machine learning improves crop-model sensitivity, while errors in temperature and rainfall propagate differently across models and regions, supporting combined use.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages329

Reading Time13 h 49 min

AccessFree to download

1 Introduction

1.1 Motivation and Overview

1.2 Climate variability and crop yields

1.3 Climate Change and its impacts

1.4 Crop modelling

1.5 Machine Learning definition and uses

1.6 Literature review

2 Methods

2.1 The GLAM crop model

2.2 Standard GLAM calibration procedures

2.3 Choice of crop model

2.4 Challenges and issues when calibrating and evaluating crop models

2.5 Challenges of spatial scale in crop modelling

2.6 Challenges and issues when comparing different crop models

2.7 Machine Learning architectures

3 A dual approach using a mechanistic crop model and machine learning enhances predictions across a range of conditions

3.1 Introduction

3.2 Methods

Using machine learning and process-based crop modelling for regional scale prediction Joseph William Gallear Submitted in accordance with the requirements for the degree of Doctor of Philosophy The University of Leeds School of Earth and Environment July 2023 Declaration The candidate confirms that the work submitted is his own and that appropriate credit has been given where reference has been made to the work of others. This copy has been supplied on the understanding that it is copyright material and that no quotation from the thesis may be published without proper acknowledgement. The right of Joseph William Gallear to be identified as Author of this work has been asserted by Joseph William Gallear in accordance with the Copyright Designs and Patents Act 1998 . ©The University of Leeds and Joseph William Gallear Acknowledgements Firstly, I would like to thank my supervisors Andrew Challinor, Anthony Cohn, Netta Cohen and Julia Chatterton for support and guidance throughout my PhD. I have very much enjoyed learning from and working with you over the last few years. I would also like to thank the climate impacts group at Leeds including Stewart Jennings, Chetan Deva and Ioannis Droutsas for fascinating discussions, valued feedback, and useful advice. Thanks to Jim Watson formerly of the climate impacts group (now working for the government of New South Wales) for providing some of the GLAM model parameter values and observed data used in this thesis. This work was funded by the Natural Environment Research Council (NERC), part of UK research and Innovation (UKRI) with an additional CASE award from Unilever PLC. I would like to thank them for making this project possible. Personal thanks to my Friends and Family for their support throughout this project, in particular my parents for their unwavering support and the valued friendships I have made during my time at the university of Leeds. Abstract The aims of this thesis are to assess the effectiveness of machine learning techniques in comparison to process-based crop growth models for the purpose of prediction of the impact of climate variability on crops at the regional scale. Comparisons are made between popular and most effective machine learning methods to predict crop yields and the process-based crop growth model GLAM. Firstly, it is asked how much data is required for machine learning to outperform process-based crop modelling, and under which conditions? Secondly, the prediction performance of both methods for prediction of crop failures is compared as well as the effect of potential errors in climate data. Thirdly, machine learning and crop modelling are compared to bench-mark crop model sensitivity to climatic drivers of crop yield, hence providing a data driven approach to learn how to further improve crop model simulations. Results show that machine learning and process-based crop modelling have contrasting strengths and weaknesses. However, machine learning can be leveraged to improve process-based crop modelling through increased sensitivity to climatic drivers of crop yields. Furthermore, the effects of potential errors in data upon machine learning simulations is determined. In doing so it is shown that sensitivity of machine learning toclimatological errors varies depending on model, and region, with different time-scales of effects depending on if errors are in temperature or rainfall. Overall, this thesis shows that machine learning can provide great benefit to regional scale crop yield prediction. However, due to disadvantages of reduced model interpretability and difficulty in predicting effects of extreme events, machine learning is not a perfect solution for regional scale crop yield prediction. Therefore, it is argued that machine learning should be used in cooporation with process-based crop modelling to improve understanding rather than replace existing methods or knowledge. Contents 1 Introduction 1 1.1 Motiva

FAQs

What comparison does the thesis make between machine learning and crop modelling?

The thesis compares machine learning methods with process-based crop growth modelling to predict crop impacts of climate variability at the regional scale, benchmarking both against GLAM.

How does the thesis evaluate prediction of crop failures and climate-data errors?

It compares the prediction performance of both approaches for crop failures and assesses how potential errors in climate inputs affect simulation outcomes.

What key conclusion is drawn about using machine learning for regional crop yield prediction?

Machine learning can improve regional-scale crop yield prediction and enhance process-based crop models via increased sensitivity, but it is limited by reduced interpretability and difficulty in handling extreme-event effects.

Aug 2, 2026

Upload by Ava Thompson