RCN: Quantifying Uncertainty in Ecosystem Studies

该研究网络的使命是帮助明确考虑统计不确定性，成为生态系统研究的一个标准组成部分。例如，构建整个森林流域的预算，极大地提高了我们对水、碳和其他物质如何在更广泛的生态系统中循环的总体理解。然而，由于观测值的自然变化或测量误差引起的不确定性很少包括在研究报告中。这使得很难确定随时间的实际变化率，也难以比较来自不同地点或研究的可信结果。不处理不确定性可能导致错误的结论，例如在确定森林中缺失的氮源或汇方面。不确定性分析还可以帮助提高环境监测的效率，因为它允许抽样设计优化收集和分析数据所需的时间和金钱。该网络将组织工作组处理五个主题。专家统计学家将在这些工作组的咨询委员会任职，并考虑跨领域问题，例如评估对观测和建立计算机模型中的不确定性进行量化的可能方法，展示如何使用不确定性分析来评估监测设计的效率，以及确定如何最好地检测生态系统测量随时间的变化。建立这一网络需要使用几种不同的方法进行协调和沟通，并为共享学习提供有效的结构，包括网络会议、电话会议、电子邮件和年度面对面会议。更广泛的受众将通过季度网络研讨会和项目网站接触到，除了工作组wiki之外，该网站还将提供常见问题页面、教育材料、图解的逐步示例、数字图书馆以及共享软件和模型计算机代码、工作流程和其他产品的交换中心。总体目标是促进文化变革，使不确定性分析在生态系统研究中成为一种预期的和容易实现的实践。

The mission of this research network is to help make the explicit consideration of statistical uncertainties a standard component of ecosystem studies. The construction of budgets for whole forest watersheds, for example, has greatly increased our general understanding of how water, carbon, and other substances cycle through ecosystems more generally. However, uncertainty due either to natural variation in observations or measurement error is rarely included in research reports. This makes it difficult to determine actual rates of change over time or to compare with confidence results from different sites or studies. Failure to address uncertainties can lead to wrong conclusions, for example in identifying missing sources or sinks for nitrogen in a forest. Uncertainty analysis can also help improve the efficiency of environmental monitoring by allowing sampling designs to optimize information gained relative to the time and money required to collect and analyze the data. This network will organize working groups to address five topics. Expert statisticians will serve on an advisory board for these working groups and also consider cross-cutting issues, such as evaluating the possible approaches to quantifying uncertainties in making observations and building computer models, demonstrating how uncertainty analysis can be used to evaluate the efficiency of monitoring designs, and determining how changes in ecosystem measurements over time can best be detected. Building this network will require using several different approaches to coordination and communication, and for providing an effective structure for shared learning, including web meetings, conference calls, email, and annual face-to-face meetings. A broader audience will be reached through quarterly webinars and through the project web site, which will have, in addition to working group wikis, pages for frequently asked questions, educational materials, illustrated step-by-step examples, digital libraries, and a clearinghouse for sharing software and model computer codes, workflows, and other products. The overall goal is to facilitate a cultural change that makes uncertainty analysis an expected and readily attainable practice in ecosystem studies.