A global coupled ensemble data assimilation system using the Community Earth System Model and the Data Assimilation Research Testbed. (14th December 2018)

• Record Type: Journal Article
• Authors: Karspeck, Alicia R.; Danabasoglu, Gokhan; Anderson, Jeffrey; Karol, Svetlana; Collins, Nancy; Vertenstein, Mariana; Raeder, Kevin; Hoar, Tim; Neale, Richard; Edwards, Jim; Craig, Anthony
• Is Part Of: Quarterly journal of the Royal Meteorological Society. Volume 144:Number 717(2018)
• Page Start: 2404
• Page End: 2430
• Publication Date: 2018-12-14
• Subjects: Community Earth System Model -- coupled data assimilation -- coupled reanalysis -- Data Assimilation Research Testbed -- ensemble assimilation -- ensemble Kalman filter -- modular data assimilation; Meteorology -- Periodicals; 551.5
• DOI: 10.1002/qj.3308
• Languages: English
• ISSNs: 0035-9009

Abstract

Abstract : This paper presents a description of the CESM/DART ensemble coupled data assimilation (DA) system based on the Community Earth System Model (CESM) and the Data Assimilation Research Testbed (DART) assimilation software. The CESM/DART should be viewed as a flexible system to support the DA needs of the CESM research community and not as a static reanalysis product. In this implementation of the CESM/DART, conventional insitu observations of the ocean and atmosphere are assimilated into the respective component models of the CESM using a 30‐member ensemble adjustment Kalman filter (EAKF). CESM/DART is run in a "weakly coupled" configuration wherein observations native to each climate system component only directly impact the state vector for that component. Information is passed between components indirectly through the short‐term coupled model forecasts that provide the EAKF background ensemble. This system leverages previous ensemble DA development for the Community Atmosphere Model and Parallel Ocean Program models using the DART EAKF. The CESM/DART project is a step towards providing increasingly useful DA capabilities for the CESM research community. Results are presented for our prototype 12‐year reanalysis, run from 1970 to mid 1982. Multiple lines of evidence demonstrate that the system is capable of constraining the CESM coupled model to simulate the historical variability of the climate system in the well‐observed Northern Hemisphere. A collection of monthly average variables, climate mode indices, observation diagnostics and snapshots of synoptic weather in the ocean and atmosphere are compared to established datasets, showing especially good agreement in the Northern Hemisphere. A discussion of the CESM/DART as a modular, community facility and the benefits and challenges associated with this vision is also included. Abstract : This work describes the newly developed ensemble coupled data assimilation system for the Community Earth System Model using the Data Assimilation Research Testbed (CESM/DART) and evaluates the performance of a 12‐year experiment assimilating conventional oceanic and atmospheric observations. Multiple lines of evidence demonstrate that the system is capable of constraining the CESM coupled model to simulate the historical variability of the climate system over much of the globe and especially in the well‐observed Northern Hemisphere. The Niño3 index (average SST from 5°N to 5°S and 90 to 150°W) as simulated in the 12‐year experimental ensemble coupled assimilation using the CESM model. For comparison, NoAssim is a free‐running coupled integration of the CESM model (initialized from the same ocean/atmosphere state). The HADISST dataset is not ingested in the data assimilation system.