The project will investigate the use of archive satellite imagery to predict spatial variability within arable fields. Many applications of precision agriculture use current satellite imagery to provide guidance on localised management operations, for example application of Nitrogen fertiliser, but assumptions have to be made about the causes of spatial variation. A 20-year archive of satellite image data will be used to develop potential productivity maps, derived from vegetation indices, to assess the degree of persistence of spatial patterns over years and their dependence on weather and cropping factors. Maps of potential yield variation and other interpretive tools will allow more intelligent, context-based assessments that are expected to lead to improved land management with both economic and environmental benefits. The project is being conducted by a consortium led by SOYL, together with ADAS, RSAC, AHDB and the University of Nottingham.

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.

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.

Severe weather can cause crops to become uprooted or their stems to break, a process called lodging. This means that the crops do not grow to their full potential which reduces the quantity of seed they produce (the yield). Lodging makes crops more susceptible to infection by fungi which can produce toxic chemicals known as mycotoxins which render the grain unusable. These impacts of lodging substantially reduce the value of a crop and there can be additional costs of drying the grain. By taking appropriate action (e.g. choice of crop variety and how it is managed) it is possible for farmers to reduce the likelihood of lodging. Over the past 20 years great advances have been made with improving understanding and control of lodging in cereal crops (e.g. wheat) grown in the UK through the development of realistic models of the lodging process which have been used to develop practical husbandry strategies enabling farmers to reduce lodging. Understanding of the lodging process in maize and rice is less advanced than in UK cereal crops and lodging in maize and rice crops commonly reduces yields by up to 40% representing a major constraint for crop productivity, particularly for low/middle income countries. This project will use world leading UK expertise in lodging science to develop understanding and mitigation strategies that enable maize and rice producing regions to minimise lodging. This strategy will increase the resilience of maize and rice production systems to climate variability and produce safer food by reducing mycotoxin development in lodged crops. Both rice and maize have very different morphologies from the crops that have already been modelled for the lodging process. New lodging models will be developed that take account of the large hanging panicles, high tiller numbers and shallow rooting system of rice; and the thick hollow stem, braced root system, much taller stature and large leaves of maize. Additionally characteristics of the climate (wind and rainfall) of different regions will be built into the models. These lodging models will be developed and tested using wind tunnel tests on pot grown plants and field tests. A framework for identifying regions and fields with the highest risk of lodging will be produced which will account for how landscape features affect wind flow over crops and how Earth Observation (EO) technologies (e.g. vegetation index maps from satellites) may be used to identify the fields with crops that are most at risk to lodging. Frequently it is found that crops with large canopies (vegetation indices) have a greater risk to lodging later in the growing season. The understanding about wind flow, EO technologies and the new lodging models will be combined into a single integrated system. This system will then be used to help farmers to mitigate lodging risk; 1) strategically by planting lodging resistant varieties in regions with a high risk of lodging and 2) tactically by reducing nitrogen fertiliser and prioritising harvest for fields (and even part fields) which are shown to have developed a high risk of lodging during the growing season. The lodging models will be used to understand how variation in crop parameters caused by changes in crop husbandry, environmental parameters (wind, rainfall and soil type) and predictions of climate change will affect lodging risk in maize and rice within different global regions. This work will identify which crop parameters have the greatest influence on lodging risk and therefore should be targeted by plant breeders.

Severe weather can cause cereal and oilseed rape crops to become uprooted or their stems to break, a process called lodging. This means that the crops do not grow to their full potential, the quantity of seed they produce (the yield) is substantially reduced and the quality of the grain decreases meaning that it cannot be used for certain purposes such as bread making. Lodging makes crops more susceptible to infection by fungi which can produce toxic chemicals which render the grain unusable. These impacts of lodging can substantially reduce the value of a crop and there can be additional costs of drying the grain harvested from lodged crops. Hence, it is estimated that lodging can cost UK farmers £170M in a severe lodging year. High winds can also cause oilseed rape pods to shatter which releases the seeds and they cannot be harvested. This costs UK farmers in excess of £7M per year. By taking appropriate action (e.g. choice of crop variety and how it is managed) it is possible for farmers to reduce the likelihood of lodging and pod shatter. However, farmers need information to guide their decisions and currently this is largely absent. This project will develop a computerised system for predicting the risks of lodging and pod shatter. It will be based on a model of how crops behave under conditions of high wind speed and soil moisture that will be developed from field experiments. The system will calculate the distribution of lodging and pod shatter across a farm that is likely to occur under severe weather conditions. This information is useful to farmers for developing plans in advance of a growing season. It will show farmers how weather damage can be reduced by selecting particular crop varieties to plant in particular fields and by adjusting the timing and density of seed planting. The system will also support farmers to make decisions within a growing season. To do this it will use satellite images to monitor the growth of crops early in the growing season and use this information together with scenarios of different weather conditions during the season to predict which fields or parts of fields are likely to be damaged by weather. This will allow farmers to take action to avoid weather damage in vulnerable fields or parts of fields by controlling the growth of crops (by altering the timing or amount of fertiliser and chemical growth regulators) and by applying chemical pod sealants. Later in the growing season the computerised system will download short-range weather forecast information and use this to predict the risks of lodging in the forthcoming weather conditions. If certain fields are predicted to be vulnerable to lodging then the farmer can arrange to harvest those fields before lodging occurs. Overall, the decision-support tool produced by this project will enable farmers to reduce the risks of weather damage to crops. This will increase farmer's capacity to produce food and reduce unnecessary use of chemicals and energy on farms which will be beneficial for the environment.