Explainable AI for UK agricultural land use decision-making

Lead Research Organisation: Lancaster University
Department Name: Lancaster Environment Centre

Abstract

Agricultural land use dynamics and their associated driving factors represent highly complex systems of flows that are subject to non-linearities, sensitivities, and uncertainties across spatial and temporal scales. They are therefore challenging to represent using traditional statistical modelling approaches. Existing process-based modelling has enabled advances in understanding of individual biophysical processes underpinning agricultural land use systems (e.g. crop, livestock and biogeochemical models). However, these tend to focus on individual processes in detail or link a limited number of processes at large scales, thereby mostly ignoring the complex interdependencies between the multiple interacting biophysical and socio-economic components of land use systems. Artificial intelligence (AI) techniques offer great potential to complement such modelling approaches by mining the deep knowledge (e.g. farming patterns and behaviours) encapsulated in 'big' data from ground-based sensors (such as frequently used for precision farming) and Earth Observation satellites. This will deliver enhanced insight on the past and current state and spatio-temporal dynamics of agricultural land use system flows and how they can be influenced by decisions on agricultural policies and related farm management practices.

Our proposal aims to develop a novel explainable AI framework that is transparent, data-driven and spatially-explicit by using probabilistic inference and explicit "if-then" rules. We will demonstrate proof-of-concept for two pilot regions of the UK (Oxfordshire and Lincolnshire), and the framework will be set up in a way that can be readily expanded to the whole UK. Specifically, we will draw on time-series of agricultural land use and production datasets (in-kind support from industry project partner SOYL) to identify the key socio-economic and environmental driving factors that have led to historic agricultural land use changes in the pilot regions. We will then establish explainable AI-rules for the characterisation of these agricultural land use changes and refine them within the framework through machine learning and parameter optimisation.

We will demonstrate and test the potential of the explainable AI framework for providing a new and robust method for predicting changing patterns of agricultural land use in the two pilot regions. This will include testing the ability of the AI framework for improving understanding of past and present agricultural land use dynamics across multiple temporal and spatial scales from 'big' data. It will also assess the potential for continually updating the predictions of land use dynamics in real-time using data from sensors. This could provide early warning when certain driving conditions are triggered or used to repeatedly refine short-term projections of land use change and their estimates of uncertainty.

Planned Impact

We will engage with four major types of beneficiaries to maximise the societal impact of this project:

1. UK Government, Devolved Administrations and Policy-makers:
Our project will provide decision-makers with an innovative and integrated knowledge base supporting agricultural land use decision-making. We will engage with policy-makers including Defra, Natural England (a project partner), the Environment Agency, and the devolved administrations through consultations on the development of the explainable AI system to ensure it meets a broad set of decision-making needs. We will also demonstrate the final AI framework at a 1-day policy workshop that explores the potential of AI approaches for supporting landscape decision-making alongside other modelling and information products, including defining future needs beyond the remit of this proposal. This will include exploring how the framework can help inform the objective in the 25 Year Environment Plan to deliver a clear evidence base to promote precision agriculture and land management.

2. Farming Industries and Agri-business:
Through our strategic partnership with SOYL, the leading precision crop production service provider in the UK (see letter of support), we will have access to a unique 'big' dataset of in-situ and EO-based agricultural land use statistics to establish the explainable AI framework. The methodological innovation in the project will be co-created with SOYL (and our other project partner, Natural England) through four meetings throughout the one-year project lifetime. This will ensure that our outcomes, disseminated through a one-week training event at SOYL, are fit-for-purpose in informing the decision-making needs of agri-businesses in order to catalyse business change and innovation.

3. AI Industries and the Economy:
The AI innovations developed in this project will support the socio-economic development of the UK by promoting automation as part of the Industry 4.0 revolution. The AI framework is fully transformative and scalable, such that it can be elaborated or enhanced as required to support decision-making within a growing UK economy that aims to balance economic, environmental and societal aspects of the UK agricultural sector. Continued technological innovations in the agricultural sector are expected through precision farming and Earth Observation, creating an increasing demand for AI and big data-driven models.

4. Land owners and Trusts:
The knowledge and explainable AI rules derived in the system will support the decision-making of land owners and national trusts, by reducing environmental hazards (e.g. pesticides), enhancing ecosystem services and natural capital, and promoting smart and profitable agriculture, as well as building an inclusive society that allows people and nature to thrive. We will engage with relevant land owners and trusts through Natural England (see letter of support) via dissemination meetings and communications (e.g. media).
The 'Pathways to Impact' document details the activities we will undertake to deliver these impacts.

Publications

10 25 50
 
Description Spatial land use models are indispensable for providing scientific evidence that can inform sustainable land use planning. However, understanding and predicting agricultural land use change and the factors that drive such change is difficult due to the highly complex systems of flows that are subject to non-linearities, sensitivities, and uncertainties across spatial and temporal scales. This complexity is challenging to represent using traditional statistical modelling approaches. In this project, we explored and tested the ability of Artificial Intelligence (AI) techniques to complement traditional land use modelling approaches by learning complex spatial and temporal relationships between large geospatial land use datasets and datasets of important driving factors of land use change.
We found three categories of driving factors to be particularly important determinants of UK land use change: climatic factors (temperature, precipitation, evapotranspiration), topographic factors (elevation, slope, distance to water courses) and human-related factors (distance to urban areas, distance to major roads, distance to railways). We developed a novel approach that combines the AI deep learning method, known as Generative Adversarial Networks, with a traditional land use modelling approach, known as Cellular Automata models. We tested the improvement in predicting land use changes from 1990 to 2019 using UKCEH Land Cover Map data to assess if the inclusion of the AI method provided better characterisation of spatial patterns and landscape dynamics.
The key findings we identified include:
1. The Cellular Automata model was able to predict future land use change. However, it only utilised limited neighbourhood information at fine scales without considering the large-scale heterogeneous environment. The transition between arable and grassland land cover classes was particularly challenging to predict using a Cellular Automata model, with relatively low accuracy (71%).
2. Capturing fine resolution spatial patterns and their context are essential for predicting land use changes. This was demonstrated in both a pilot study in Oxfordshire and wider application of our method for England. We show that in both case studies accuracy of predicting land use change was increased by 8% with the incorporation of spatial and context information.
3. Combining the AI (deep learning) method with the Cellular Automata model provided both fine-scale neighbourhood and large-scale context information to make joint landscape decisions. The spatial patterns among woodland, arable, grassland, freshwater and urban are accurately characterised, with an overall accuracy of 89%.
4. Human-related factors were found to be key drivers of land use changes. In particular, the distance to urban and distance to road were the two most important drivers that have significant impact on land use changes. These drivers are highly related to human activities and road accessibility, and influenced UK landscape dynamics over the past 30 years.
Exploitation Route The project has produced a variety of outcomes such as land use maps and their driving factors for England at a 100 m resolution. The project has provided proof of concept that innovative AI methods can add significant precision to land use prediction, but further effort is needed to make the approach (and associated model code) operational at the large-scale so that it can be implemented to inform policy-makers and agri-businesses. In particular, cloud computing and edge computing are required to establish real-time prediction, where farmers and stakeholders can use mobile device for rapid decision-making. In combination with in-situ precision agriculture, this further effort could help increase agricultural productivity and support land management.
Sectors Agriculture, Food and Drink,Education

URL https://www.ceh.ac.uk/our-science/projects/explainable-ai-uk-agricultural-land-use-decision-making
 
Description Detecting soil degradation and restoration through a novel coupled sensor and machine learning framework
Amount £811,651 (GBP)
Funding ID NE/T012307/1 
Organisation Natural Environment Research Council 
Sector Public
Country United Kingdom
Start 01/2020 
End 12/2022
 
Description Pilot study: Can we use drone technology for quantifying and monitoring flower resources in field margins?
Amount £15,969 (GBP)
Organisation Natural England 
Sector Charity/Non Profit
Country United Kingdom
Start 01/2021 
End 04/2021
 
Description Sustainable soil management to unleash soil biodiversity potential and increase environmental, economic and social wellbeing (SOILGUARD)
Amount € 7,000,000 (EUR)
Organisation European Commission H2020 
Sector Public
Country Belgium
Start 06/2021 
End 05/2025
 
Description Turing AI Fellowship: Probabilistic Algorithms for Scalable and Computable Approaches to Learning (PASCAL)
Amount £1,097,294 (GBP)
Funding ID EP/V022636/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Public
Country United Kingdom
Start 01/2021 
End 12/2025
 
Description CRAFTY-UK: Development of a UK agent-based model of land use change taking account of land owner/manager decision-making 
Organisation Forest Research
Country United Kingdom 
Sector Public 
PI Contribution Information on UK datasets and statistical analysis of land use change over time for different agricultural land classes
Collaborator Contribution Set-up, calibration and coding on the CRAFTY agent-based model for the UK
Impact Outcomes still under development. Collaboration is multi-disciplinary involving environmental scientists, social scientists, computer programmers and statisticians
Start Year 2020
 
Description CRAFTY-UK: Development of a UK agent-based model of land use change taking account of land owner/manager decision-making 
Organisation Karlsruhe Institute of Technology
Country Germany 
Sector Academic/University 
PI Contribution Information on UK datasets and statistical analysis of land use change over time for different agricultural land classes
Collaborator Contribution Set-up, calibration and coding on the CRAFTY agent-based model for the UK
Impact Outcomes still under development. Collaboration is multi-disciplinary involving environmental scientists, social scientists, computer programmers and statisticians
Start Year 2020
 
Description CRAFTY-UK: Development of a UK agent-based model of land use change taking account of land owner/manager decision-making 
Organisation University of Edinburgh
Country United Kingdom 
Sector Academic/University 
PI Contribution Information on UK datasets and statistical analysis of land use change over time for different agricultural land classes
Collaborator Contribution Set-up, calibration and coding on the CRAFTY agent-based model for the UK
Impact Outcomes still under development. Collaboration is multi-disciplinary involving environmental scientists, social scientists, computer programmers and statisticians
Start Year 2020
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Integrating quantitative social, ecological and mathematical sciences into landscape decision-making" 
Organisation Isaac Newton Institute for Mathematical Sciences
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Collaborator Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Impact Workshop took place, which enabled networking between multiple disciplines around landscape decision-making. Disciplines involved were quantitative social sciences, environmental sciences, mathematics and statistics.
Start Year 2020
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Integrating quantitative social, ecological and mathematical sciences into landscape decision-making" 
Organisation University of Exeter
Department College of Engineering, Mathematics & Physical Sciences
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Collaborator Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Impact Workshop took place, which enabled networking between multiple disciplines around landscape decision-making. Disciplines involved were quantitative social sciences, environmental sciences, mathematics and statistics.
Start Year 2020
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Integrating quantitative social, ecological and mathematical sciences into landscape decision-making" 
Organisation University of Leeds
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Collaborator Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Impact Workshop took place, which enabled networking between multiple disciplines around landscape decision-making. Disciplines involved were quantitative social sciences, environmental sciences, mathematics and statistics.
Start Year 2020
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Integrating quantitative social, ecological and mathematical sciences into landscape decision-making" 
Organisation University of Southampton
Department Faculty of Natural and Environmental Sciences
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Collaborator Contribution Contributed to the design, invitations and organisation of the 4-day online workshop
Impact Workshop took place, which enabled networking between multiple disciplines around landscape decision-making. Disciplines involved were quantitative social sciences, environmental sciences, mathematics and statistics.
Start Year 2020
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Mathematical and statistical challenges in landscape decision-making" 
Organisation Isaac Newton Institute for Mathematical Sciences
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the organisation of the main 1-month programme and the opening and closing workshops
Collaborator Contribution Contributed to the organisation of the main 1-month programme and the opening and closing workshops
Impact Final Scientific Report created from the programme. Collaboration is multi-disciplinary involving a range of different environmental scientists, mathematicians and statisticians
Start Year 2019
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Mathematical and statistical challenges in landscape decision-making" 
Organisation University of Exeter
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the organisation of the main 1-month programme and the opening and closing workshops
Collaborator Contribution Contributed to the organisation of the main 1-month programme and the opening and closing workshops
Impact Final Scientific Report created from the programme. Collaboration is multi-disciplinary involving a range of different environmental scientists, mathematicians and statisticians
Start Year 2019
 
Description Organising committee for UKRI Landscape Decisions Programme series of events on "Mathematical and statistical challenges in landscape decision-making" 
Organisation University of Southampton
Department Faculty of Natural and Environmental Sciences
Country United Kingdom 
Sector Academic/University 
PI Contribution Contributed to the organisation of the main 1-month programme and the opening and closing workshops
Collaborator Contribution Contributed to the organisation of the main 1-month programme and the opening and closing workshops
Impact Final Scientific Report created from the programme. Collaboration is multi-disciplinary involving a range of different environmental scientists, mathematicians and statisticians
Start Year 2019
 
Description Blog on "Overcoming barriers to model coupling" 
Form Of Engagement Activity Engagement focused website, blog or social media channel
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact Blog on the topic of Model coupling (the communication and interchange of information between models) as a key component in meeting the challenges of many of today's complex real-world environmental problems, such as landscape decisions.
Year(s) Of Engagement Activity 2020
URL https://ceeds.ac.uk/blogs/overcoming-barriers-model-coupling
 
Description CEEDS Seminar: Machine Learning for the Natural Environment 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact The presentation was designed as an introductory tutorial into machine learning techniques. The focus of the talk was to educate environmental scientists on the available machine learning tools that are applicable for analysing environmental data.
Year(s) Of Engagement Activity 2020
URL https://ceeds.ac.uk/blogs/ceeds-seminar-machine-learning-natural-environment
 
Description CEEDS Seminar: Machine Learning for the Natural Environment 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Professional Practitioners
Results and Impact This seminar was attended by over 170 people - a fantastic and amazing attendance across Lancaster University and UK Centre for Ecology and Hydrology (UKCEH). It helps build the strong links between LU and UKCEH, but also the links across UKCEH science areas and different sites.
Year(s) Of Engagement Activity 2021
URL https://ceeds.ac.uk/blogs/ceeds-seminar-machine-learning-natural-environment
 
Description Intergovernmental science-policy Platform on Biodiversity and Ecosystem Services Scoping Workshop on the Nexus Assessment 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Policymakers/politicians
Results and Impact Government nomination to participate in the scoping workshop for the IPBES Nexus Assessment on the interlinkages among biodiversity, water, food and health in the context of climate change. A scoping report was produced following the scoping workshop, which will be considered and hopefully approved at the IPBES-8 plenary in 2021.
Year(s) Of Engagement Activity 2020
URL https://ipbes.net/nexus/scoping-experts
 
Description Invited speaker (International Workshop) 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Presentation given on methods for modelling complex systems, such as landscape decisions, including how uncertainty effects decision-making, which sparked debate and capacity building with groups undertaking similar modelling initiatives in South Korea and other parts of Asia
Year(s) Of Engagement Activity 2019
 
Description Presentation on "Coupling models to represent interactions within landscape systems" at Workshop on "Integrating quantitative social, ecological and mathematical sciences into landscape decision-making" 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact Isaac Newton Institute (INI) workshop on "Integrating quantitative social, ecological and mathematical sciences into landscape decision-making" organised as part of the UKRI Landscape Decisions Programme. Presentation summarised discussion from INI Programme in 2019. Research priorities were divided into four themes: (i) transparency, reproducibility and communication in coupled models; (ii) model coupling toolbox; (iii) model coupling technicalities; and (iv) taking advantage of the benefits of model coupling. The key insights that emerged in these four themes were captured within short, medium and longer term research roadmaps.
Year(s) Of Engagement Activity 2020
URL https://www.newton.ac.uk/seminar/20200907142014401
 
Description Presentation on "Pathways to sustainable land-use and food systems in the UK" at EAT@Home event side session 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach International
Primary Audience Other audiences
Results and Impact Side event at the EAT Foundation online conference on sustainable land-use and food systems. The presentation on the UK landscape decision-making sparked questions and discussion on the method and the importance of assessing the contribution that sustainable land-use change can make to multiple policy targets, e.g. climate change, biodiversity, health diets.
Year(s) Of Engagement Activity 2020
 
Description Presentation on the "Explainable AI for UK agricultural land-use decision-making" project at the Landscape Decisions Programme virtual networking event 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact Landscape Decisions Programme series of online workshops over 4 days covering (i) Capturing people's interaction with landscapes and landscape decision making; (ii) Multi-functional landscapes - focusing on ecosystem services; (iii) What pressures on Landscapes are currently not being considered - i.e. where are the gaps in our understanding of the pressures?; (iv) Can we systematically capture different levels/scales of landscape decision making and which levels/scales does your project interact with? Sparked questions and discussions around these issues and networking opportunities in relation to the project.
Year(s) Of Engagement Activity 2020
URL https://www.youtube.com/watch?v=sIbc8-_AqUo&list=PLrlZ6FipN5mlS4_lN9PwZAR6HCyppGTDP&index=4&t=149s
 
Description Public consultation workshop on "Living Landscapes" 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Public/other audiences
Results and Impact Royal Society Workshop on Living Landscapes (provided expert input to a public consultation exercise on landscapes in the UK).
Year(s) Of Engagement Activity 2020
 
Description Seminar on "Model coupling for evaluating complex systems" 
Form Of Engagement Activity A talk or presentation
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Other audiences
Results and Impact Centre for Excellence in Environmental Data Science seminar that looked at the important topic of model coupling as a means to address interdependencies between different aspects of the environment. The seminar built understanding around the extent to which complicated environmental questions can be better assessed by integrating models, and discussed the advantages, the challenges, as well as the limitations of joined-up modelling for complex systems.
Year(s) Of Engagement Activity 2020
 
Description Session organised on AI for the environment (Natural Capital Initiative Summit) 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact Session organised on "AI for the environment" at the Natural Capital Initiative Summit on "Valuing our Life Support Systems" in London, which consisted of talks and debate around the potential of AI approaches to deliver natural capital solutions to practitioners and policy-makers
Year(s) Of Engagement Activity 2019
 
Description Workshop (Environmental Data Science) 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact Workshop at Alan Turing Institute on exploring collaborative opportunities between the computer and environmental sciences, which sparked discussions and ideas around new collaborations and innovations in environmental data science
Year(s) Of Engagement Activity 2019
 
Description Workshop on "Overcoming barriers to model coupling" 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach Regional
Primary Audience Other audiences
Results and Impact Centre for Excellence in Environmental Data Science workshop on model coupling. Workshop report produced to summarise discussions and research needs
Year(s) Of Engagement Activity 2020
 
Description Workshop on "Seamless integration between natural and built environment modelling" 
Form Of Engagement Activity Participation in an activity, workshop or similar
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Other audiences
Results and Impact Human activity - the buildings and transport systems we create, the energy we generate and consume, the services we use - has an impact on global environmental systems and vice versa. But what role can digital models play in helping decision-makers to navigate these complex interrelationships and to manage the built and natural environment sustainably? And what if we link models of each of the built and natural environments?
This workshop seeks to prompt a discussion of the question: What outcomes and opportunities could seamless integration between built and natural environment modelling deliver?
After looking at it from the perspective of different stakeholders and exploring opportunities to get better insights from data and models, the desired outcomes are as follows:
• The seeds of a discussion community for identifying opportunities for model integration and interdisciplinary work.
• Contacts between people pursuing other strands of work happening in the integration of models for decision support;
• A report on insights from the workshop (end of March 2021), which will serve as a resource for future interdisciplinary research projects as well as the Information Management Framework (IMF) community of the National Digital Twin programme.
Year(s) Of Engagement Activity 2020