TECHNOLOGY MARKET ASSESSMENT AND FOLLOW ON FUND APPLICATION SUPPORT FOR THE SAFE SOFTWARE TOOL

复杂的数值模型是许多行业使用的核心工具。它们的用途从设计复杂的机械结构，如风车，到为保险业分析整个欧洲的风暴风险，到用于制定泰晤士河地区可持续使用政策的水管理模型。对于这些通常非常复杂和非线性的模型来说，校准和评估是困难的，对于这些模型，非正式的方法，如试错法、目视分析或专家判断很快就变得不可行。对于用于设计和政策制定的数字模型，越来越多地要求甚至要求更严格的质量标准。这样的分析通常还必须包括通过模型传播的不确定性。除非这些模型所产生的预测中的不确定性能够被量化和表征，否则它们将不适合作为决策的基础。需要专门的工具来分析、理解和传达这些不确定性、其来源及其对模型产出的影响。这些工具需要易于使用，并与专家和非专家进行交流。在NERC资助的可信项目中，我们开发了一个名为SAFE(每个人的敏感度分析)的软件工具箱。外管局是一种执行全球敏感性分析(GSA)的数值工具，GSA是一套统计技术，为开发和分析数值模型提供了一种量化和结构化的方法。它实现了一套全面的GSA方法和可视化工具，并结合在一个经过广泛测试和用户友好的工具箱中。由于可信项目专门针对广泛的自然灾害领域开发通用工具，该工具箱非常适合进一步开发成为一个完全通用的软件工具。这种GSA软件产品的商业潜力将扩展到几乎任何使用数值模型的工业领域，从制造公司到环境咨询或保险。尽管在外管局实施的GSA方法已得到很好的确立、验证和发布，但非专业受众并不容易接触到这些方法，也不能迅速为新模型实施。现有的软件工具要么是“赤裸裸的”代码，需要大量的编程专业知识才能实现，要么是非常专业的工具，专注于特定的方法，并没有更广泛的适用性。SAFE工具箱克服了这两个缺点，我们希望将其发展成一个用户友好、易于适应的工具，可在多个领域使用。在我们的探路者项目中，我们打算(I)调查将能够从该工具箱中获得最大好处的市场部门，因此可能是最有吸引力的市场部门，以便尽早采用；(Ii)收集最终用户对该工具所需的技术和支持功能的意见，以便为SAFE商业版本的技术开发提供信息；关键词：数值模型、灵敏度分析、不确定性分析、校准、验证、评估持有者：行业、政府、政策制定者、学术界

Complex numerical models are core tools that are used in many industries. Their uses range from the design of complex mechanical structures such as windmills, over the analysis of windstorm risk across Europe for the insurance industry, to water management models used for developing sustainable use policies for the Thames region. Calibration and evaluation is difficult for these typically very complex and non-linear models, for which informal approaches such as trial-and-error, visual analysis or expert judgement quickly become unfeasible. More stringent quality criteria are increasingly demanded or even required for numerical models, which are used for design and policy development. Such analyses often also have to include uncertainty propagation through the model. Unless the uncertainty in the predictions produced by these models can be quantified and characterised they will not be suitable as a basis for decision-making. Specific tools are required to analyse, understand and convey these uncertainties, their sources and their effects on the model outputs. Such tools need to be easy to use and communicate to specialists and non-specialists alike.Within the NERC-funded CREDIBLE Project, we have developed a software toolbox named SAFE (Sensitivity Analysis For Everybody). SAFE is a numerical tool to perform Global Sensitivity Analysis (GSA), a set of statistical techniques providing a quantitative and structured approach to the development and analysis of numerical models. It implements a comprehensive suite of GSA methods and visualisation tools, combined in an extensively tested and user-friendly Toolbox. Since the CREDIBLE Project was specifically aimed at developing generic tools across a wide range of natural hazard domains, the Toolbox is well suited for further development into a completely generic software tool. The commercial potential of such GSA software product would extend to virtually any industrial domain where numerical models are used, from manufacturing companies to environmental consultancy or insurance. Although the GSA approaches implemented in SAFE are well established, validated and published, they are not readily accessible to a non-specialist audience and cannot quickly be implemented for new models. Existing software tools are either "bare" code, which requires significant programing expertise to implement, or very specialised tools that focus on a specific methodology and are not more widely applicable. The SAFE toolbox overcomes both these drawbacks, and we hope to develop it into a user-friendly, easily adaptable tool that can be used in a range of sectors.In our Pathfinder project, we intend to (i) investigate the market sectors that will be able to benefit most from the toolbox and therefore could be the most attractive market sectors to target for early adoption; (ii) to gain input from end-users on the technical and support features needed for the tool so to inform the technical development of the commercial version of SAFE; and (iii) to identify the best route to market for the product.Keywords: Numerical Modelling, Sensitivity Analysis, Uncertainty Analysis, Calibration, Validation, EvaluationStakeholders: Industry, Government, Policymakers, Academia