Decadal Prediction Large Ensemble Project

The CESM Decadal Prediction Large Ensemble (DPLE) is a set of simulations carried out at NCAR to support research into near-term Earth System prediction. The DPLE comprises 62 distinct ensembles, one for each of 62 initialization times (November 1 of 1954, 1955, ..., 2014, 2015). For each start date, a 40-member ensemble was generated by randomly perturbing the atmospheric initial condition at the round-off level. The simulations were integrated forward for 122 months after initialization. Observation-based ocean and sea ice initial conditions for the 1954-2015 period were obtained from a reanalysis-forced simulation of the CESM ocean and sea ice models. The initial conditions for the atmosphere and land models were obtained from CESM Large Ensemble (LENS) simulations at corresponding historical times. Full field initialization was used for all component models, and so drift adjustment prior to analysis is generally recommended (e.g., see here). The DPLE was run using the same CESM code base, configuration details, component resolutions (nominally 1-degree in both atmosphere and ocean), and external forcing datasets as for the CESM LENS project. DPLE therefore represents the "initialized" complement to the LENS simulations. Steve Yeager [[e-mail]] at NCAR is the primary contact for DPLE-related inquiries. Nan Rosenbloom and Gary Strand were instrumental in the setup, running, and post-processing of the DPLE simulations, with assistance from Sheri Mickelson, Alice Bertini, Jim Edwards, and Shiquan Su. The initial 10 members were made possible by a Department of Energy award of computer time on machines at the National Energy Research Scientific Computing Center (NERSC). The additional 30 members were completed in early 2017 thanks to a CISL Accelerated Scientific Discovery award on Cheyenne. The DPLE is a CESM community project, and the output from the simulations is available to anyone who is interested. Monthly, daily, and 6-hourly outputs are archived on the NCAR HPSS system and accessible from the Earth System Grid (http://www.earthsystemgrid.org) as single variable timeseries. A list of available output fields can be found here. (NOTE: Not all fields are available for all 40 members. See Known Issues link for further details.) If you are interested in in analyzing the CESM-DPLE, we kindly ask that you provide a short description of your proposed research focus to Steve Yeager [[e-mail]] to be included in the list of ongoing projects (see sidebar). An overview article describing the DPLE has been published in the Bulletin of the American Meteorological Society: Yeager, S. G., G. Danabasoglu, N. Rosenbloom, W. Strand, S. Bates, G. Meehl, A. Karspeck, K. Lindsay, M. C. Long, H. Teng, and N. S. Lovenduski, 2018: Predicting near-term changes in the Earth System: A large ensemble of initialized decadal prediction simulations using the Community Earth System Model, Bulletin of the American Meteorological Society, , doi: . Please reference this paper in oral presentations and papers that make use of CESM-DPLE data and acknowledge the supercomputing resources provided by DOE/NERSC as well as NSF/CISL/Cheyenne. A list of publications using the DPLE data (along with closely related publications that made use of earlier versions of the CESM decadal prediction system) can be found at the Publications link on the sidebar. If you would like to add a publication to this list, please contact [[e-mail]].CESM年代际预测大集合（DPLE，Decadal Prediction Large Ensemble）是美国国家大气研究中心（NCAR，National Center for Atmospheric Research）为支撑近地地球系统预测研究而开展的一系列模拟实验。该数据集包含62组独立集合模拟，分别对应1954年至2015年每年11月1日的62个初始时刻。针对每个起算日期，研究团队通过在舍入误差级别随机扰动大气初始条件，生成了包含40个成员的集合模拟。所有模拟均从初始时刻起向前积分122个月。1954-2015年期间基于观测的海洋与海冰初始条件，源自CESM海洋和海冰模式的再分析强迫模拟结果；大气与陆面模式的初始条件，则取自对应历史时刻的CESM大集合（LENS，Large Ensemble）模拟结果。所有分量模式均采用全场初始化方案，因此通常建议在分析前进行漂移校正（详见此处）。 DPLE所使用的CESM代码库、配置参数、分量模式分辨率（大气与海洋名义分辨率均为1°）以及外部强迫数据集，均与CESM LENS项目保持一致，因此DPLE可视为LENS模拟的“初始化补充”版本。NCAR的Steve Yeager（邮箱：[e-mail]）是DPLE相关咨询的主要联系人。Nan Rosenbloom与Gary Strand主导了DPLE模拟的搭建、运行与后处理工作，Sheri Mickelson、Alice Bertini、Jim Edwards及Shiquan Su提供了协助。初始的10个集合成员的计算资源，由美国能源部（DOE）为其在国家能源研究科学计算中心（NERSC，National Energy Research Scientific Computing Center）的机器分配机时支持。剩余30个集合成员则于2017年初通过Cheyenne超算平台的CISL加速科学发现项目资助完成。 DPLE是CESM社区项目，其模拟结果面向所有感兴趣的研究者开放。模拟产出的逐月、逐日及每6小时数据，均归档于NCAR的HPSS系统，并可通过地球系统网格（http://www.earthsystemgrid.org）以单变量时间序列形式获取。可用输出变量列表详见此处。（注：并非所有变量均对全部40个集合成员开放，详细信息请参阅已知问题链接。）若计划开展CESM-DPLE相关分析研究，恳请您向Steve Yeager（[e-mail]）提交简短的研究计划说明，以便将您的项目纳入在研项目列表（详见侧边栏）。 一篇介绍DPLE的综述文章已发表于《美国气象学会公报》（Bulletin of the American Meteorological Society）： Yeager, S. G., G. Danabasoglu, N. Rosenbloom, W. Strand, S. Bates, G. Meehl, A. Karspeck, K. Lindsay, M. C. Long, H. Teng, and N. S. Lovenduski, 2018: Predicting near-term changes in the Earth System: A large ensemble of initialized decadal prediction simulations using the Community Earth System Model, Bulletin of the American Meteorological Society, , doi: . 使用CESM-DPLE数据进行口头汇报或发表论文时，请务必引用该文章，并对美国能源部/国家能源研究科学计算中心（DOE/NERSC）及美国国家科学基金会/CISL/Cheyenne提供的超算资源致谢。使用DPLE数据发表的论文列表（以及使用早期CESM年代际预测系统相关版本的相近研究）可通过侧边栏的“出版物”链接查阅。若您希望将自己的研究成果添加至该列表，请联系[[e-mail]]。