Enhancing High-resolution Terrain Data Model for Improving the Delineation of Multi-scale Hydrological Connectivity

该项目旨在创建一种新的地理空间和水文建模方法，利用高分辨率数字高程模型（hrdem）改进对河流和溪流的识别。准确识别这些河流和溪流网络对于监测诸如营养物质和水生物种的运输等环境过程是必要的。研究结果将通过结合地理信息科学、人工智能和网络基础设施的方法，创建一个智能水文模型，从而实现河流和溪流网络的多尺度分析。该项目生成的模型、算法和数据集将免费向公众提供，并有利于广泛的自然资源管理活动，如流域监测、湿地保护和水生物种保护。这项工作的成果将纳入教育课程，并向更广泛的教育界传播。目前迫切需要数据集和建模方法，这些数据集和建模方法可用于生成水文校正的hrdem，并在精细尺度上描绘湿地、道路涵洞和桥梁等水文特征。为了提高建模精度，本项目将开发与hrdem兼容的地理空间人工智能水文建模框架。将开发一个深度学习模型和排水交叉位置的图像数据集，以识别作为虚拟流障的水力排水结构的位置。水力排水结构、HRDEM分辨率和流向算法对所描绘的水文特征及其连通性的影响将通过控制实验进行评估，以获得最佳组合。该模型将在高性能计算集群上实现，以加快地理计算速度。该项目将极大地推动地理信息科学、人工智能和网络基础设施等领域基于地形的水文建模技术的发展。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 project is to create a new geospatial and hydrological modeling approach that improves the identification of rivers and streams using high-resolution digital elevation models (HRDEMs). Accurate identification of these river and stream networks is necessary for monitoring environmental process such as the transport of nutrients and aquatic species. The results will create an intelligent hydrological model by combining methods in geographic information sciences, artificial intelligence and cyberinfrastructure that will enable multi-scale analysis of river and stream networks. The model, algorithms, and datasets generated from this project will be freely available to the public and benefit a broad scope of natural resources management activities, such as watershed monitoring, wetland conservation, and aquatic species protection. The results of this work will be incorporated into educational curriculums and disseminated to the broader educational community.There is a critical need for datasets and modeling approaches that can be used to generate hydrologically corrected HRDEMs and delineate hydrologic features like wetlands, road culverts and bridges at a fine scale. To improve the modeling accuracy, this project will develop a geospatial artificial intelligence-hydrological modeling framework that is compatible with HRDEMs. A deep learning model and an image dataset of drainage crossing locations will be developed to identify the locations of hydraulic drainage structures as virtual flow barriers. The effects of hydraulic drainage structures, HRDEM resolutions, and flow direction algorithms on the delineated hydrologic features and their connectivity will be assessed via controlled experiments for an optimal combination. The model will be implemented on high-performance computing clusters to speed up the geocomputation. This project will substantially advance the state-of-the-art of terrain-based hydrologic modeling integrating methods in geographic information science, artificial intelligence, and cyberinfrastructure.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.