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，使之成为生态系统研究的标准组成部分。例如，为整个森林流域构建预算，极大地提高了我们对水、碳和其他物质如何在更普遍的生态系统中循环的一般理解。 然而，研究报告中很少包括由于观测结果的自然变化或测量误差造成的不确定性。这使得很难确定随时间推移的实际变化率，也很难与不同地点或研究的置信度结果进行比较。 不解决不确定性可能导致错误的结论，例如在确定森林中氮的缺失源或汇时。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所获得的信息，而不是收集和分析数据所需的时间和金钱。该网络将组织工作组讨论五个专题。统计专家将担任这些工作组的咨询委员会成员，并审议跨领域问题，例如评价在观测和建立计算机模型时量化不确定性的可能方法，展示如何利用不确定性分析来评价监测设计的效率，并确定如何最好地检测生态系统测量随时间的变化。 建立这一网络将需要使用几种不同的方法来协调和沟通，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。 将通过季度网络研讨会和项目网站接触更广泛的受众，除了工作组维基之外，该网站还将有常见问题页面、教育材料、图解分步示例、数字图书馆以及用于共享软件和模型计算机代码、工作流程和其他产品的信息交流中心。总体目标是促进文化变革，使不确定性分析成为生态系统研究中预期和容易实现的实践。

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.