PYRAMID: Platform for dYnamic, hyper-resolution, near-real time flood Risk AssessMent Integrating repurposed and novel Data sources

洪水已被政府确定为英国的头号优先事项和风险。每年，洪水已经给人们的家园、基础设施和经济造成了价值数百万英镑的损失，预计在气候变化下还会变得更加严重。因此，能够有效地计划、应对和管理洪水是至关重要的。我们很幸运地拥有由环境局领导的英国洪水管理的传统。业务洪水模型使用气象数据和海拔数据相结合，向我们显示洪水将发生在哪里。这些模型制作的水浸风险图可供规划和预测之用，并协助我们设计多个地区的防洪设施。然而，水浸并不单止取决于地区的地形。随着时间的推移，还有许多其他因素在起作用：涵洞可能被堵塞，防洪闸可能会打开，防洪墙可能会年久失修。这会极大地改变洪水的范围和深度。不仅如此，我们对洪水风险的敞口也发生了变化。与高峰时段的交通相比，夜间洪水造成的干扰要小得多。一个最好的例子是2004年博斯卡斯尔的洪水。在活动期间，停在停车场的116辆汽车被冲到下游，堵塞了一座桥，导致水倒退，淹没了意想不到的区域。如果晚上下雨，车就不会停在停车场里，洪水的影响也会小一些。我们能预测到这一点吗？我们能减少洪水对未来类似事件的影响吗？我们认为，有了正确的数据和工具，我们可以做到。我们将建立一个工具，随着洪水风险的演变，改变我们应对洪水风险的方式。该工具将允许洪水风险管理人员部署及时的维护和缓解措施，如清理严重堵塞的涵洞或设置流动防洪堤。为了实现这一目标，该工具将把全新类型的数据和尖端洪水模型整合到一个易于使用的在线平台中，使用户能够可视化不断演变的洪水风险。该平台(称为金字塔)将与环境署、地方当局和社区团体共同开发，以确保其近乎实时地为关键决策提供相关信息。该平台将拥有工具包，使社区可以轻松地整合他们的数据，提供必要的当地信息。推动这种建模的新数据将是关键。我们需要的数据是可用的，但分散在一系列组织中，格式很难使用。我们将使用人工智能从隐藏的数据集中提取这些有用的信息，例如旧报告、洪水资产登记和各种类型的卫星图像。此外，我们希望纳入来自新型传感器的全新信息，这些传感器正被部署为纽卡斯尔大学城市天文台的一部分。这些传感器以非常高的分辨率监测土壤湿度和降雨量，以及交通和拥堵等其他因素。我们亦可监察影响水浸风险的特定因素的情况，例如个别暗渠是否堵塞，或某些防洪墙是否状况欠佳。通过观察卫星遥感和部署在卡车和其他车辆上的传感器的组合，可以监测这些因素。我们还将利用收集到的社区和公民的数据。所有这些信息都将输入我们的洪水模型。我们有一个超分辨率的水动力洪水模型，可以在厘米尺度上准确地模拟洪水中泥石流的运动。该模型将与一个更广泛的集水区模型一起工作，该模型将提供更广泛地区的水文条件信息。该平台将在纽卡斯尔进行试验，以利用政府对城市天文台的现有投资和在这里进行的洪水研究的遗产。

Flooding has been identified by the government as the number one priority and risk to the UK. Flooding already causes millions of pounds worth of damage to people's homes, infrastructure and the economy every year, and is projected to become even more severe under climate change. Being able to plan for, respond to and manage flooding effectively is therefore essential.We are lucky to have a tradition of flood management in the UK led by the Environment Agency. Operational flood models use meteorological data combined with elevation data to show us where flooding will occur. These models produce flood risk maps for planning and forecasting purposes and have helped us design flood defences for many areas.However, flooding is not only dependent on the topography of an area. There are many other factors at play that evolve over time: culverts can get blocked, flood gates are left open and flood walls can fall into disrepair. This can dramatically alter the extent and depth of a flood. Not only that, but our exposure to flood risk changes too. Far less disruption occurs from a flood overnight than during rush hour traffic. A prime example of this is the flooding of Boscastle in 2004. During the event, 116 cars parked in a carpark were washed downstream, blocking a bridge, causing water to back up and flood unexpected areas. If the rain had fallen in the evening, the cars would not have been in the carpark and the impact of the flood would have been smaller. Could we have predicted this? Can we reduce the impact of flooding for similar future events? We think that with the right data and tools, we can.We will build a tool that will change how we respond to flood risks as they evolve. The tool will allow flood risk managers to deploy just-in-time maintenance and alleviation measures, such as clearing critical blocked culverts or setting up mobile flood defences. To achieve this, the tool will incorporate brand new types of data and cutting edge flood models into an easy-to-use online platform that allows users to visualise evolving flood risks. The platform (called PYRAMID) will be developed in conjunction with the Environment Agency, local authorities and community groups to ensure that it delivers relevant information for critical decision-making in near-real time. The platform will have toolkits to make it easy for communities to incorporate their data, providing essential local information.The new data driving this modelling will be key. The data that we need are available but sit fragmented across a range of organisations in difficult-to-use formats. We will use artificial intelligence to extract this useful information from hidden datasets, such as old reports, flood asset registers and various types of satellite imagery. In addition, we want to incorporate brand new information from novel sensors that are being deployed as part of Newcastle University's Urban Observatory. These sensors monitor things like soil moisture and rainfall at very high resolutions, as well as other factors like traffic and congestion. We can also monitor the condition of specific factors affecting flood risk, such as whether particular culverts are blocked or whether certain flood walls are in poor condition. These factors can be monitored by looking at a combination of satellite remote sensing and sensors deployed on lorries and other vehicles. We will also harness data collected communities and citizens.All of this information will be put into our flood models. We have a hyper-resolution hydrodynamic flood model that can accurately simulate the movement of debris in flood flows at a centimetre scale. This model will work in conjunction with a broader catchment model, which will provide information on the hydrological conditions in the wider area. The platform will be trialled in Newcastle to take advantage of existing government investments in the Urban Observatory and a legacy of flood research conducted here.