Collaborative Research: Novel Integration of Direct Measurements with Numerical Models for Real-time Estimation and Forecasting of Streamflow Response to Cyclical Processes

对天然河流的持续监测支持与水资源管理和预测有关的社会经济需求，并为科学调查提供基准数据。目前用于连续监测河流流量的协议是基于无法准确捕捉洪水波传播的短期影响和河流岸边植被变化的季节性影响的方法。虽然几十年来仪器技术的长足进步极大地改变了我们的现场测量能力，但这种进步并没有反映在水流监测协议的进步上。拟议研究的结果将填补这一空白，为水科学和管理提供关键支持，并增加用于发布保护社区和关键基础设施所需的洪水预警的预测模型的可靠性。在洪水继续对社区构成重大威胁之际，提高实时流量测量的准确性和提高预报模型的可靠性将对公众的福祉和安全产生直接的积极影响。研究成果将包括利用人工智能技术衍生的快速工具，以补充现有的预测能力。拟议的研究结合了实验，数据驱动（即机器学习）和基于物理的数值调查的推断，以便采用河段范围的监测方法（而不是横断面），并实时跟踪由非定常流和河岸植被生长引起的所有流量变量的变化。异质路线方法辅以广泛的数据分析，将能够提取由与上述循环过程相关的滞后行为的微妙特征产生的流动变量之间的相互依赖性。推广对大范围流量和现场条件的推断，将经济有效地改进仅使用原位获取的数据来预测流量关系的协议，而无需借助建模。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。

Continuous monitoring of natural rivers supports socio-economic needs related to water resources management and forecasting and provides benchmark data for scientific investigations. The current protocols for continuously monitoring streamflows are based on approaches that fail to accurately capture the short-time effects of flood wave propagation and the seasonal impacts of changes to stream bank vegetation. While decades of substantial advancements in instrumentation technologies have dramatically transformed our in-situ measurement capabilities, this progress has not been mirrored by advancements in streamflow monitoring protocols. The outcomes of the proposed research will fill this gap by offering critical support for water science and management and by adding reliability to predictive models used to issue flood warnings required to protect communities and critical infrastructures. Increasing the accuracy of streamflow measurements in real time and improving the reliability of forecast models will have direct positive impacts on the wellbeing and safety of the public at a time when flooding continues to be a major threat to communities. The research outcomes will include rapid tools derived with artificial intelligence techniques that supplement existing forecasting capabilities. The proposed research combines inferences from experimental, data-driven (i.e., machine learning) and physically-based numerical investigations to enable adoption of a reach-scale monitoring method (rather than cross-sectional) and real time tracking of changes to all flow variables induced by unsteady flows and riparian vegetation growth. A heterogeneous routing approach complemented by extensive data analysis will allow the extraction of interdependencies among flow variables produced by subtle features of the hysteretic behavior associated with the above-mentioned cyclical processes. Generalization of the inferences for a wide range of flow and site conditions will cost-effectively improve the protocols for predictive streamflow relationships using only in-situ acquired data, without making recourse to modeling.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.