Simulating 10,000 Years of Erosion to Assess Nuclear Waste Repository Performance
Simulating 10,000 Years of Erosion to Assess Nuclear Waste Repository Performance
复制标题
模拟一万年的侵蚀来评估核废料储存库的性能
DOI：
--
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
P. Stauffer
中科院分区：
文献类型：
--
作者：
A. Atchley;K. Birdsell;K. Crowell;R. Middleton;P. Stauffer
自然过程（包括侵蚀）的长期环境绩效评估对于废物储存库场地评估至关重要。然而，由于参数不确定性和复杂的非线性过程，评估站点连续运行的能力具有挑战性。我们提出了一个包含多个替代数据源和降阶模型的工作流程，以代替无法用于模型校准的现场数据，以验证低放射性核废料储存库长期侵蚀评估的参数。我们将这种新的工作流程应用于新墨西哥州洛斯阿拉莫斯国家实验室台面上的低放废物储存库。为了考虑参数的不确定性，我们模拟了高、中和低侵蚀情况。该评估延伸至 10,000 年，这会导致很大的侵蚀不确定性，但考虑到埋藏废物的性质，这是必要的。我们的长期侵蚀分析表明，高侵蚀情况会产生圆形台面和部分填充的峡谷，这与中度侵蚀情况不同，中度侵蚀情况会导致沟壑和尖锐的台面边缘。我们新颖的模型参数化工作流程和建模练习展示了长期评估的实用性，确定了侵蚀预测不确定性的来源，并展示了景观演化模型开发的实用性。最后，我们讨论了减少评估不确定性和增加模型置信度的方法。
Long-term environmental performance assessments of natural processes, including erosion, are critically important for waste repository site evaluation. However, assessing a site’s ability to continuously function is challenging due to parameter uncertainty and compounding nonlinear processes. In lieu of unavailable site data for model calibration, we present a workflow to include multiple sources of surrogate data and reduced-order models to validate parameters for a long-term erosion assessment of a low-level radioactive nuclear waste repository. We apply this new workflow to a low-level waste repository on mesas in Los Alamos National Laboratory in New Mexico. To account for parameter uncertainty, we simulate high-, moderate-, and low-erosion cases. The assessment extends to 10,000 years, which results in large erosion uncertainties, but is necessary given the nature of the interred waste. Our long-term erosion analysis shows that high-erosion scenarios produce rounded mesa tops and partially filled canyons, diverging from the moderate-erosion case that results in gullies and sharp mesa rims. Our novel model parameterization workflow and modeling exercise demonstrates the utility of long-term assessments, identifies sources of erosion forecast uncertainty, and demonstrates the utility of landscape evolution model development. We conclude with a discussion on methods to reduce assessment uncertainty and increase model confidence.