SCC-IRG Track 2: Scalable Modeling and Adaptive Real-time Trust-based Communication (SMARTc) System for Roadway Inundations in Flood-Prone Communities

由于海平面上升（SLR）和涨潮，美国东部和墨西哥湾大部分地区的经常性有害洪水（RNF）事件的频率正在增加和加速。这些经常发生的洪水淹没了雨水排水系统，造成道路封闭，对已建成的基础设施构成重大威胁，并扰乱了社区。在本研究中，近乎实时的RNF数据处理、街道和社区规模的RNF长期预测以及这些信息的有效共享将向当地司机灌输信任，并提高社区的安全性。该智能互联社区（SCC）项目的主要目标是开发用于道路淹没检测和监测的可扩展建模和自适应实时基于信任的通信（SMARTc）系统。SMARTc系统将在弗吉尼亚州诺福克市的一个洪水易发地区进行评估，使用来自该市摄像机、潮汐计以及现场现有和新的陆上水位传感器的数据。通过获得这些近乎实时的信息，市民将能够避免在被洪水淹没的道路上开车，应急车辆可以在被洪水淹没的道路上改道，城市将更好地了解洪水的模式，以及投资于雨水和沿海防洪系统的需求。该团队还将与诺福克的非营利组织RISE合作，该组织致力于帮助企业为沿海社区开发新的解决方案，以适应SLR和RNF。这种合作有望加快方法和技术的推广，以及未来向实践的过渡。此外，还计划提供各种教育和外展机会，以增加项目的影响。其中包括一个由代表广泛利益相关者的参与者组成的区域论坛，为正在进行的NSF REU项目设计项目，将研究成果整合到本科和研究生课程中，为访问高中生提供实践活动，跨学科顶点项目，以及向少数民族中学生展示原型系统。这项研究的结果将与诺福克当地社区分享，以提高对RNF的认识和提高对RNF的恢复能力的技术。一旦部署在诺福克的现场，该解决方案可以在RNF下为汉普顿路的数十万公民，企业和紧急服务提供准确的道路状况。这些解决方案可以被全国其他社区采用，可能会帮助数百万人。该项目的预期成果与美国国家科学基金会十大构想中的“利用数据革命”部分直接相关。本研究将包括以下任务：(i)基于在不同天气条件下收集的监控摄像头图像，近乎实时地检测街道洪水范围和深度的新型机器学习算法；水动力模型，结合水文-暴雨-海岸耦合模型预测街道至社区范围的洪水水位，并根据传感器和图像数据实时更新这些预测；（iii）局部淹没条件下道路通行能力的实时预测以及洪水深度和程度与驾驶员行为的相关性；（iv）利用洪水信息的粒度和不确定性，向公众有效传播洪水风险和道路淹没情况。设想中的系统将利用传感器数据和相机图像进行近实时的道路淹没检测，并将提取的动态信息与水动力学模型相结合，用于街道到社区规模的道路淹没预测。第一个任务的结果将产生用于街道尺度RNF范围和深度识别的近实时学习模型。第二项任务将使用城市摄像机和其他传感器数据，生成改进的社区尺度道路网络洪水预测模型，用于RNF范围、深度和洪水持续时间。第三项任务将为部分被水淹没的路段生成改进的微观车辆跟随模型，以便可以近乎实时地准确估计和表征通行能力和瓶颈。最后，第四项任务预计将为使用RNF范围、深度的驾驶员提供有效的“风险”沟通策略。通过获得这些目前无法获得的近乎实时的信息，预计公民将避免在被淹没的道路上开车，紧急车辆可以在被淹没的道路周围重新安排路线，城市将更好地了解洪水模式以及投资于雨水和沿海防洪系统的需求。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

The frequency of recurrent nuisance flooding (RNF) events is increasing and accelerating along much of the US East and Gulf Coasts due to Sea Level Rise (SLR) and high tides. These recurrent floods overwhelm the stormwater drainage systems, cause road closures, pose a major threat to the built infrastructure, and disrupt communities. In this research, the near real-time processing of RNF data, longer term prediction of RNF at both street and community scale and effective sharing of this information will instill trust in local drivers and improve safety in the community. The main objective of this Smart & Connected Communities (SCC) project is to develop a Scalable Modeling and Adaptive Real-time Trust-based communication (SMARTc) system for roadway inundation detection and monitoring. The SMARTc system will be evaluated for a flood-prone region in the City of Norfolk, Virginia, using data from the City’s cameras, tide gauges, and existing and new overland water level sensors in the field. By having access to such information in near real-time, citizens will be able to avoid driving through flooded roads, emergency vehicles can be rerouted around inundated roads, and cities will have a better understanding of flooding patterns and the needs to invest in storm-water and coastal flood protection systems. The team will also engage with RISE – a non-profit organization in Norfolk focused on helping businesses develop new solutions for coastal communities to adapt to SLR and RNF. This collaboration is expected to expedite scaling up methods and technologies, and future transition to practice. In addition, various educational and outreach opportunities are planned to increase the project impact. These include a regional forum with participants representing a broad range of stakeholders, design projects for ongoing NSF REU programs, integration of research outcomes into undergraduate and graduate classes, hands on activities for visiting high school students, interdisciplinary capstone projects, and presentation of a prototype system to minority middle school students. The results of this research will be shared to the local community in Norfolk to increase awareness of RNF and technologies for increasing resilience to RNF. Once deployed in the field in Norfolk, the solutions could provide hundred-thousands of citizens, businesses, and emergency services in Hampton Roads with accurate roadway conditions under RNF. These solutions could then be adopted by other communities across the nation, potentially helping millions. The expected outcomes of this project are directly relevant to Harnessing the Data Revolution component of the NSF’s Ten Big Ideas.This research will include the following tasks: (i) Novel machine learning algorithms for detecting floodwater extent and depth at street level in near real-time based on surveillance camera images collected under varying weather conditions; (ii) Hydrodynamic modeling integrating a coupled hydrologic-stormwater-coastal model to predict flood levels at street to community scales and real-time update of these predictions based on sensor and image data; (iii) Prediction of roadway capacities in real-time under partial inundations and correlation of floodwater depth and extent with driver behavior; and (iv) Effective communication of flood risk and road inundation to the public, leveraging granularity and uncertainty of flood information. The envisioned system will leverage sensor data and camera images for near real-time road inundation detection and will integrate the extracted dynamic information with hydrodynamic models for street to community-scale road inundation prediction. The outcome of the first task will yield near real-time learning model for street-scale RNF extent and depth recognition. The second task will yield an improved community-scale road network flood prediction model for RNF extent, depth, and flood duration using City camera and other sensor data. The third task will yield improved microscopic car-following models for partially flooded roadway segments so that the capacities and bottlenecks may be estimated and characterized accurately in near real-time. Finally, the fourth task is expected to offer effective ‘risk’ communication strategies for drivers using the RNF extent, depth. By having access to such information in near real-time, which is currently not available, citizens are expected to avoid driving through flooded roads, emergency vehicles can reroute around inundated roads, and cities will have a better understanding of flooding patterns and the needs to invest in storm-water and coastal flood protection systems.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.