PREEVENTS Track 2: Collaborative Research: Defining precursors of ground failure: a multiscale framework for early landslide prediction through geomechanics and remote sensing

人口增长、城市扩张和极端天气比以往任何时候都更容易造成灾害。在由天气模式驱动的灾害中，由滑坡引起的地面变形具有巨大的全球影响。地球表面的大部分地区都面临着地面故障的风险，影响着世界上相当一部分人口。山体滑坡每年造成全球数千人死亡，仅在美国每年就造成超过10亿美元的经济损失。预测地面故障最可怕的挑战是，尽管没有观测到前兆，但地面故障仍能突然加速。事实上，天然斜坡可以以多种方式变形，有时通过显示缓慢的运动，而在其他时候在流化状态下快速移动。这种变形模式在同一地点共存，影响彼此相邻的地形部分，并且可能在不同时间由同一山坡经历。目前预测大尺度地面变形模型的缺乏主要是由于地面监测数据的空间覆盖较差。为了克服这些障碍，该项目将依靠遥感技术的进步，使人们能够以时空分辨率探测降雨模式和地面运动，这在十年前是不可想象的。这些观测进展有可能通过丰富的、公开可用的、空间分布的信息，为水文和地面变形模型的制定、校准和验证提供新的可能性。具体来说，该项目的动机是这样一种想法，即破坏前的变形与地面破坏发生的时间、方式和原因有很大关系，目的是证明分析地面的变形特征是解释为什么山坡在不同的天气模式下以不同的方式失败的关键。如果成功，该项目将带来新的方法来解码地面不稳定的物理根源，并定义灾难性滑坡触发的可测量前兆，从而可能启发设计创新的实时预警系统，能够更好地保护人类生命和基础设施。在这个项目的过程中，地球表面和大气之间的相互作用将从一个新的多学科的角度来研究。具体而言，该项目将制定：(1)能够解释动态变化的环境条件导致的滑坡速度变化的岩土材料流变规律；(2)可量化山地尺度上空间异质性降水输入和土壤湿度的多尺度气象水文模拟平台；(3)景观尺度的地质力学模型，能够通过地形近端部分之间的力传递规律再现遥感变形的演变；(4)基于复杂系统物理的表面过程网络理论，该理论可以识别模式并定义失控不稳定性的前兆。这种方法的组合将提供滑坡动力学的全面代表，从而提高我们在景观尺度上预测滑坡和减轻灾害的能力。最重要的是，它将提供创新的工具来解决灾害预测领域的开放性问题，例如：(i)我们能否利用景观尺度的观测来推断山坡的流变？（ii）哪种景观尺度的测量对预测初期滑坡的命运最有用？同时收集降雨模式和位移率的空间分布数据是否足以确定滑坡前兆？这些问题将通过结合来自最先进的遥感工具（例如，卫星和机载干涉合成孔径雷达，高分辨率数字高程模型和气象雷达）的数据，以及基于物理的土壤和岩石变形本构定律，大气水文模型和复杂网络理论来回答。可用于美国国内外各种地质环境和土质的丰富数据集将用于测试所提出方法的预测能力。这一策略将为验证项目核心概念提供独特的机会，并测试其在广泛的地貌和气候背景下的适用性。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。

Population growth, urban expansion, and extreme weather are contributing more than ever to hazard vulnerability. Among the hazards driven by weather patterns, ground deformation due to landslides has immense global impacts. Large portions of the Earth's surface are at risk to ground failures, affecting a considerable fraction of the world's population. Landslides cause a global annual death toll of several thousand and financial losses of more than $1B per year in the United States alone. The most formidable challenge in predicting ground failures derives from their ability to suddenly accelerate despite the lack of observed precursors. In fact, natural slopes can deform in multiple ways, sometimes by displaying slow movements, while at other times moving rapidly in a fluidized state. Such modes of deformation coexist at the same site, affect portions of terrain proximal to one another, and may be experienced by the same hillslopes at different times. The current scarcity of predictive large-scale ground deformation models is largely a consequence of the poor spatial coverage of ground-based monitoring data. To overcome these obstacles, this project will rely on technological advances in remote sensing that allow the detection of rainfall patterns and ground movements at spatiotemporal resolutions that were unthinkable just a decade ago. These observational advances have the potential to unleash new formulation, calibration and validation possibilities for hydrologic and ground deformation models by means of abundant, openly-available, spatially-distributed information. Specifically, the project is motivated by the idea that pre-failure deformations have much to say about when, how, and why ground failure occurs, and aims to demonstrate that analyzing the deformation signature of the ground is the key to explain why hillslopes fail in different ways when subjected to variable weather patterns. If successful, this project will lead to new ways to decode the physical origin of ground instability and define measurable precursors of catastrophic landslide triggering, thus potentially inspiring the design of innovative real-time early warning systems able to better protect human life and infrastructure. During the course of this project, the interaction between the Earth's surface and the atmosphere will be studied from a new multi-disciplinary perspective. Specifically, the project will formulate: (1) rheological laws for geomaterials able to explain variations in landslide velocity resulting from dynamically changing environmental conditions; (2) a multiscale weather-hydrology simulation platform able to quantify spatially heterogeneous rainfall inputs and soil moisture at the scale of mountain ranges; (3) landscape-scale geomechanical models able to reproduce the evolution of remotely sensed deformations via force-transfer laws between proximal portions of terrain; (4) a network theory for surface processes based on the physics of complex systems, by which patterns can be identified and precursors of runaway instability defined. Such a combination of methods will provide a comprehensive representation of landslide dynamics, thus improving our ability to forecast landslides and mitigate hazards at the landscape scale. Most importantly, it will provide innovative tools to address open questions in the domain of hazard forecasting, such as: (i) Can we use landscape-scale observations to infer the rheology of hillslopes? (ii) Which landscape-scale measurements are most useful for predicting the fate of incipient landslides? (iii) Is the concurrent collection of spatially-distributed data of rainfall patterns and displacement rates sufficient to identify landslide precursors? These questions will be answered by combining data from state-of-the-art remote sensing tools (e.g., satellite and airborne interferometric synthetic aperture radar, high-resolution digital elevation models, and weather radar) with physics-based constitutive laws for soil and rock deformation, atmospheric-hydrologic models, and complex network theories. Rich datasets available for a variety of geological settings and earthen materials within and outside the United States will be used to test the predictive capabilities of the proposed approaches. This strategy will offer unique opportunities to validate the concepts at the core of the project and test their applicability to a wide range of geomorphic and climatic contexts.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.