CAFE60 reanalysis

The CSIRO Climate retrospective Analysis and Forecast Ensemble system: version 1 (CAFE60v1) provides a large ensemble retrospective analysis of the global climate system from 1960 to present with sufficiently many realizations and at spatio-temporal resolutions suitable to enable probabilistic climate studies. Using a variant of the ensemble Kalman filter, 96 climate state estimates are generated over the most recent six decades. These state estimates are constrained by monthly mean ocean, atmosphere and sea ice observations such that their trajectories track the observed state while enabling estimation of the uncertainties in the approximations to the retrospective mean climate over recent decades. Strongly coupled data assimilation (SCDA) is implemented via an ensemble transform Kalman filter in order to constrain a general circulation climate model to observations. Satellite (altimetry, sea surface temperature, sea ice concentration) and in situ ocean temperature and salinity profiles are directly assimilated each month, whereas atmospheric observations are sub-sampled from the JRA55 atmospheric reanalysis. Strong coupling is implemented via explicit cross domain covariances between ocean, atmosphere, sea ice and ocean biogeochemistry. Atmospheric and surface ocean fields are available at daily resolution and monthly resolution for the land, subsurface ocean and sea ice. The system also produces a complete data archive of initial conditions potentially enabling individual forecasts for all members each month over the 60 year period. The size of the ensemble and application of strongly coupled data assimilation lead to new insights for future reanalyses.CAFE60v1 has been validated in comparison to empirical indices of the major climate teleconnections and blocking from various reanalysis products (ERA5, JRA55, NCEP NR1). Estimates of the large scale ocean structure and transports have been compared to those derived from gridded observational products (WOA18, HadISST, ERSSTv5) and climate model projections (CMIP). Sea ice (extent, concentration and variability) and land surface (precipitation and surface air temperatures) are also compared to a variety of model (ERA5, CMIP) and observational (GPCP, AWAP, HadCRU4, GIOMAS, NSIDC, HadISST) products. This analysis shows that CAFE60v1 is a useful, comprehensive and unique data resource for studying internal climate variability and predictability, including the recent climate response to anthropogenic forcing on multi-year to decadal time scales.The data and its spatio-temporal characteristics are described in the following publications in the American Meteorological Societies Journal of Climate.CAFE60v1: A 60-year large ensemble climate reanalysis. Part I: System design, model configuration and data assimilation.Terence J. O’Kane, Paul A. Sandery, Vassili Kitsios, Pavel Sakov, Matthew A. Chamberlain, Mark A. Collier, Russell Fiedler, Thomas S. Moore, Christopher C. Chapman, Bernadette M. Sloyan, and Richard J. MatearDOI: https://doi.org/10.1175/JCLI-D-20-0974.1Published Online: 22 Mar 2021 CAFE60v1: A 60-year large ensemble climate reanalysis. Part II: EvaluationTerence J. O’Kane, Paul A. Sandery, Vassili Kitsios, Pavel Sakov, Matthew A. Chamberlain, Dougal T. Squire, Mark A. Collier, Christopher C. Chapman, Russell Fiedler, Dylan Harries, Thomas S. Moore, Doug Richardson, James S. Risbey, Benjamin J. E. Schroeter, Serena Schroeter, Bernadette M. Sloyan, Carly Tozer, Ian G. Watterson, Amanda Black, Courtney Quinn, and Richard J. MatearDOI: https://doi.org/10.1175/JCLI-D-20-0518.1Published Online: 22 Mar 2021Lineage: CAFE60v1 has been designed with the intention of simultaneously generating both initial conditions for multi-year climate forecasts and a large ensemble retrospective analysis of the global climate system from 1960 to present. The data was generated over a 12 month period at Australia's NCI facility by CSIRO scientists and in collaboration with Pavel Sakov at the Australian Bureau of Meteorology. The data produced is archived at CSIRO IM&T facilities and AWS.澳大利亚联邦科学与工业研究组织（CSIRO）的气候回顾分析与预报集合系统版本1（CAFE60v1）提供了1960年至今全球气候系统的大集合回顾分析，包含足够多的集合成员，时空分辨率适配概率气候研究需求。本系统采用集合卡尔曼滤波（ensemble Kalman filter）的变体，生成了近60年的96份气候状态估计结果。这些状态估计结果受月均海洋、大气与海冰观测数据约束，其轨迹能够匹配观测状态，同时可估算近几十年回顾性平均气候近似值的不确定性。 强耦合数据同化（Strongly coupled data assimilation, SCDA）通过集合变换卡尔曼滤波实现，用于将全球环流气候模型与观测数据进行约束匹配。卫星观测（测高数据、海表温度、海冰密集度）与原位海洋温度盐度廓线每月直接被同化，而大气观测数据则从JRA55大气再分析资料中进行子采样获取。强耦合通过海洋、大气、海冰与海洋生物地球化学之间的显式跨域协方差实现。大气与表层海洋场的时空分辨率为日尺度，陆面、次表层海洋及海冰数据则采用月尺度分辨率。该系统还生成了完整的初始条件数据档案，可支持1960年至今60年间每月所有集合成员的独立预报试验。该集合的规模以及强耦合数据同化技术的应用，为未来的气候再分析研究提供了全新的视角。 CAFE60v1已通过与主要气候遥相关经验指数及各类再分析产品（ERA5、JRA55、NCEP NR1）的阻塞场进行对比验证。大尺度海洋结构与输运的估计结果已与格点观测产品（WOA18、HadISST、ERSSTv5）及气候模式投影结果（CMIP）进行了对比验证。海冰（范围、密集度与变率）及陆面要素（降水、地表气温）也与多种模式产品（ERA5、CMIP）和观测产品（GPCP、AWAP、HadCRU4、GIOMAS、NSIDC、HadISST）进行了对比验证。上述分析表明，CAFE60v1是一项实用、全面且独特的数据资源，可用于研究气候内部变率与可预报性，包括多年至年代际尺度上近期气候对人为强迫的响应。 该数据集及其时空特征已发表于美国气象学会旗下《气候杂志》（Journal of Climate）的以下两篇论文中： 1. 《CAFE60v1：一项60年大集合气候再分析 第一部分：系统设计、模式配置与数据同化》 作者：Terence J. O’Kane、Paul A. Sandery、Vassili Kitsios、Pavel Sakov、Matthew A. Chamberlain、Mark A. Collier、Russell Fiedler、Thomas S. Moore、Christopher C. Chapman、Bernadette M. Sloyan及Richard J. Matear DOI：https://doi.org/10.1175/JCLI-D-20-0974.1 在线发表时间：2021年3月22日 2. 《CAFE60v1：一项60年大集合气候再分析 第二部分：验证评估》 作者：Terence J. O’Kane、Paul A. Sandery、Vassili Kitsios、Pavel Sakov、Matthew A. Chamberlain、Dougal T. Squire、Mark A. Collier、Christopher C. Chapman、Russell Fiedler、Dylan Harries、Thomas S. Moore、Doug Richardson、James S. Risbey、Benjamin J. E. Schroeter、Serena Schroeter、Bernadette M. Sloyan、Carly Tozer、Ian G. Watterson、Amanda Black、Courtney Quinn及Richard J. Matear DOI：https://doi.org/10.1175/JCLI-D-20-0518.1 在线发表时间：2021年3月22日 数据集溯源：CAFE60v1的设计目标是同时生成多年期气候预报的初始条件，以及1960年至今全球气候系统的大集合回顾分析结果。 该数据集由CSIRO科研人员于12个月内，在澳大利亚国家计算基础设施（NCI）设施完成，并与澳大利亚气象局的Pavel Sakov合作开发。生成的数据集已归档于CSIRO IM&T设施及亚马逊云服务（AWS）。