Numerical models of pyroclastic currents are widely used for fundamental research and for hazard and risk modeling that supports decision-making and crisis management. Because of their potential high impact, the credibility and adequacy of models and simulations needs to be assessed by means of an established, consensual validation process. To define a general validation framework for pyroclastic current models, we propose to follow a similar terminology and the same methodology that was put forward by Oberkampf and Trucano (Prog Aerosp Sci, 38, 2002) for the validation of computational fluid dynamics (CFD) codes designed to simulate complex engineering systems. In this framework, the term validation is distinguished from verification (i.e., the assessment of numerical solution quality), and it is used to indicate a continuous process, in which the credibility of a model with respect to its intended use(s) is progressively improved by comparisons with a suite of ad hoc experiments. The methodology is based on a hierarchical process of comparing computational solutions with experimental datasets at different levels of complexity, from unit problems (well-known, simple CFD problems), through benchmark cases (complex setups having well constrained initial and boundary conditions) and subsystems (decoupled processes at the full scale), up to the fully coupled natural system. Among validation tests, we also further distinguish between confirmation (comparison of model results with a single, well-constrained dataset) and benchmarking (inter-comparison among different models of complex experimental cases). The latter is of particular interest in volcanology, where different modeling approaches and approximations can be adopted to deal with the large epistemic uncertainty of the natural system.

中文翻译：

---

火山碎屑流模型的验证和基准测试框架

火山碎屑流的数值模型广泛用于基础研究以及支持决策和危机管理的灾害和风险建模。由于其潜在的重大影响，模型和模拟的可信度和充分性需要通过既定的、双方同意的验证过程进行评估。为了定义火山碎屑流模型的通用验证框架，我们建议遵循 Oberkampf 和 Trucano (Prog Aerosp Sci, 38, 2002) 提出的用于验证计算流体动力学 (CFD) 代码的类似术语和相同方法旨在模拟复杂的工程系统。在这个框架中，术语验证与验证（即对数值解质量的评估）不同，它用于表示一个连续的过程，通过与一组临时实验进行比较，模型在其预期用途方面的可信度逐渐提高。该方法基于将计算解决方案与不同复杂程度的实验数据集进行比较的分层过程，从单元问题（众所周知的简单 CFD 问题）到基准案例（具有良好约束的初始和边界条件的复杂设置）和子系统（全面解耦过程），直至完全耦合的自然系统。在验证测试中，我们还进一步区分了确认（模型结果与单个、约束良好的数据集的比较）和基准测试（复杂实验案例的不同模型之间的相互比较）。后者对火山学特别感兴趣，