14TSB_ATC_IR_Optimising Big Data to Drive Sustainable Agricultural Intensification

该项目将解决目前阻碍可持续集约化（SI）的主要农艺/土壤相关挑战。目前英国园艺部门的浪费和田间损失水平是不可持续的，主要归因于与土壤退化相关的限制和不适当的作物管理方法。该项目将侧重于以可持续土壤管理为基础，最大限度地减少浪费和田间损失。土壤是粮食生产的基础，因为它们支持作物生产，对种植者来说，土壤是他们的主要商业资产。我们将通过创建一个连贯的农业信息学数据集来推动园艺和耕地生产系统的SI，该数据集涵盖了马铃薯、谷物、油籽、甜菜、葱属植物、根茎和芸苔的全面轮作，具有前所未有的分辨率和规模（5-10年的数据覆盖了20万公顷）。我们将使用特定于农业信息学的方法，这些方法包括数据挖掘和计算技术，这些技术可以识别和描述作物生产和作物利用的环境和农艺驱动因素，如贝叶斯信念网络、神经网络和随机森林，以确定一个领域的历史和主要背景如何影响其当前状态、性能和未来效用的潜力（SI）。并阐明如何将土壤健康指标用于系统监测、管理和改进。实现该项目的数据框架已经通过“生命土壤”（SfL）系统建立起来，这是一个现有的公司规模的土壤信息管理系统，由上述“农业信息学”方法支撑。SfL系统允许对土壤理化和生物参数、“土壤健康”的具体指标、作物、品种、农艺、气象、水和经济数据进行整理，所有这些数据都用于支持相关的分析和数据挖掘工具套件。该项目将开发移动应用程序和相关的网络数据服务，为现场种植者提供同步的、地理参考的SfL数据库访问。种植者将能够在一个包裹的基础上，逐个领域地咨询统一的数据持有，包括历史和当代的操作。还将提供农业信息学分析工具的总结结果。因此，从被判断为环境相似的地理位置得出的结果将被提供，突出所采取的行动和结果——允许种植者之间的知识转移。该项目将允许对这些“大数据”进行深入分析和探索，为支持SI和保持农田、农场和企业规模的土壤健康提供有力的科学证据。该项目将成为催化剂，推动英国农业企业整合和充分利用目前在当地产生和存储的“大数据”的方式发生转变。

This project will address key agronomic/soil related challenges currently impeding sustainable intensification (SI). Current levels of wastage and field losses across the UK horticultural sector are not sustainable and are largely attributed to soil degradation-related constraints and inappropriate crop management approaches. This project will focus on minimising wastage and field losses as underpinned by sustainable soil management. Soils are the foundation of food production as they support crop production, and for growers, soil is their primary business asset.We will drive SI in horticultural and arable production systems by creating a coherent agri-informatics data set covering a comprehensive range of rotations derived from potato, cereals, oil seeds, sugarbeet, alliums, roots and brassica's of unprecedented resolution and scale (5-10 years' worth of data covering >20,000 ha). We will use methods particular to agri-informatics, which consist of data-mining and computational techniques that identify and describe the environmental and agronomic drivers of crop production and crop utilization such as Bayesian belief networks, neural networks, and random forests to determine how the history and prevailing context of a field influences its current state, performance and potential for future utility (SI), and elucidate how metrics of soil health can be used for system monitoring, management and improvement. The data framework to realise this project is already established via the 'Soil-for-Life' (SfL) system an existing company scale Soil Information Management System underpinned by the aforementioned 'agri-informatics' approaches. The SfL system permits the collation of soil physico-chemical and biological parameters, specific indicators of 'soil health', crop, varietal, agronomic, meteorological, water and economic data, all deployed to support the suite of related analytical and data mining tools. This project will develop mobile applications and related web data services which will be deployed to provide growers in the field with synchronous, geo-referenced access to the SfL databank. Growers will be able to consult the harmonised data holdings on a parcel basis, field by field, both for historical and contemporary operations. Summary results from the agri-informatics analytical tools will also be provided. Thus outcomes from geographic locations judged to be similar environmentally will be made accessible, highlighting actions taken and outcomes - allowing for knowledge transfer between growers. This project will allow an in-depth analysis and exploration of this 'big data', providing robust scientific evidence to support SI and to maintain soil health at a field, farm and enterprise scale. This project will act as a catalyst for a step-change in the way that 'big data' currently generated and stored locally by UK Agri-businesses is combined and critically, fully utilised.