Dynamic Statistical Models to Improve Long-term Volcanic Hazard Assessments

长期火山危险评估旨在预测未来潜在危险的火山活动的时间和性质。这种评估需要考虑到火山系统的动态性质，如喷发喷口的迁移、喷发频率的变化以及各种构造环境中的岩浆集中。该项目汇集了一个多学科团队，以提高我们对火山系统如何在空间和时间上演化的理解，因为地球地幔和地壳内的复杂岩浆过程无法直接观察到。在这个项目中，研究人员将集中在加利福尼亚州拉森峰周围的分布的火山系统上，作为对空间强度(单位面积的喷口)、体积强度(单位面积的喷发量)、体积流量(单位时间和面积的喷发量)和复发率模型(单位时间的喷发事件数量)的统计模型的测试。他们计划通过整合提供地下岩浆作用线索的数据，如地球物理、地球化学和构造数据，来增强这些模型。他们将对过去的火山喷发进行新的年龄测定，并评估拉森火山系统的地球化学趋势。总而言之，这些数据和模型应该为评估未来火山活动的潜在影响提供更准确的工具。然后，研究人员计划推广该模型，以考虑美国西部不同的火山系统及其潜在的危险。具体地说，关于喷口位置和喷发量的现有数据将被用来建立拉森和美国西部其他五个研究充分的火山系统的空间强度和体积强度的非参数核密度统计模型。这些统计模型将堤坝注入、岩床发育和喷发的离散过程视为连续的密度函数。这些统计模型中的不确定性(例如，喷口埋藏的不确定性；地质年代学的不确定性)将通过在拉森以东的Cariou火山田收集更多数据来进行检验。在那里，将收集新的辐射年龄测定和额外的体积数据，以使用我们以前开发的随机重现率和熔岩流淹没模型来测试田地增长的统计模型。利用喷口和喷发单元的放射性年龄测定，将使用蒙特卡罗方法计算火山活动的复发率和相关的不确定性。将采用控制岩浆生产和运移的微分方程组的随机解来模拟岩浆上升的地下过程。利用这一连续公式，通过改变源和导电性参数的选择，可以简单地实现影响岩浆迁移的额外复杂性，例如复杂的来源、岩浆生成、岩浆流变学、构造应力和/或多孔介质的各向异性/非均质行为。通过这种方式，可以测试可能导致数值模式中的非均匀通量的物理过程，并将其与观测到的喷口分布和地表的体积通量联系起来，从而在火山活动的统计模型和观测到的地球物理、构造和地球化学数据之间建立更紧密的联系。

Long-term volcanic hazard assessments aim to forecast the timing and nature of future, potentially dangerous, volcanic activity. Such assessments need to account for the dynamic nature of volcanic systems, such as migration of eruptive vents, variations of eruption frequency, and magma focusing in various tectonic settings. This project brings together a multidisciplinary team to improve our understanding of how volcanic systems evolve in space and time due to complex magmatic processes within the Earth's mantle and crust that cannot be directly observed. In this project, researchers will concentrate on the distributed volcanic system around Lassen Peak, California, as a test for statistical models of spatial intensity (vents per unit area), volume intensity (erupted volume per unit area), volume-flux (erupted volume per unit of time and area) and recurrence rate models (number of eruptive events per unit of time). They plan to enhance these models by integrating data that provide clues to subsurface magmatic processes, such as geophysical, geochemical and tectonic data. They will make new age determinations of past volcanic eruptions and assess geochemical trends in the Lassen volcanic system. Together, these data and models should provide more accurate tools for assessing the potential impacts of future volcanic activity. Researchers then plan to generalize this model to consider diverse volcanic systems in the western United States and their potential hazards.Specifically, existing data on vent location and erupted volumes will be used to develop nonparametric kernel density statistical models of the spatial intensity and volume intensity at Lassen and for five other well-studied volcanic systems in the western U.S.A. These statistical models cast the discrete processes of dike injection, sill development, and eruption as continuous density functions. Uncertainties in these statistical models (e.g., uncertainty due to vent burial; uncertainty in geochronology) will be tested by gathering additional data in the Caribou volcanic field, east of Lassen. There, new radiometric age determinations and additional volume data will be collected to test statistical models of field growth using stochastic recurrence rate and lava flow inundation models that we have previously developed. Using radiometric age determinations of vents and erupted units, recurrence rate of volcanism and associated uncertainty will be calculated using a Monte Carlo approach. Stochastic solutions to differential equations governing magma production and transport will be implemented to model subsurface processes of magma ascent. Using this continuous formulation, additional complexities that influence magma migration such as complex sources, magma generation, magma rheology, tectonic stresses, and/or anisotropic/heterogeneous behavior of the porous medium, can be simply implemented by varying the choice of source and conductivity parameters. In this way physical processes that may give rise to heterogeneous flux in numerical models can be tested and be related to observed vent distributions and volume flux at the surface, creating stronger links between statistical models of volcanism and observed geophysical, tectonic, and geochemical data.