DOI: 10.1007/s00382-015-2678-y
Scopus记录号: 2-s2.0-84959158942
论文题名: Drift dynamics in a coupled model initialized for decadal forecasts
作者: Sanchez-Gomez E. ; Cassou C. ; Ruprich-Robert Y. ; Fernandez E. ; Terray L.
刊名: Climate Dynamics
ISSN: 9307575
出版年: 2016
卷: 46, 期: 2017-05-06 起始页码: 1819
结束页码: 1840
语种: 英语
英文关键词: Decadal forecast
; Initialization
; Model drift
; Physical mechanisms
英文摘要: Drifts are always present in models when initialized from observed conditions because of intrinsic model errors; those potentially affect any type of climate predictions based on numerical experiments. Model drifts are usually removed through more or less sophisticated techniques for skill assessment, but they are rarely analysed. In this study, we provide a detailed physical and dynamical description of the drifts in the CNRM-CM5 coupled model using a set of decadal retrospective forecasts produced within CMIP5. The scope of the paper is to give some physical insights and lines of approach to, on one hand, implement more appropriate techniques of initialisation that minimize the drift in forecast mode, and on the other hand, eventually reduce the systematic biases of the models. We first document a novel protocol for ocean initialization adopted by the CNRM-CERFACS group for forecasting purpose in CMIP5. Initial states for starting dates of the predictions are obtained from a preliminary integration of the coupled model where full-field ocean surface temperature and salinity are restored everywhere to observations through flux derivative terms and full-field subsurface fields (below the prognostic ocean mixed layer) are nudged towards NEMOVAR reanalyses. Nudging is applied only outside the 15°S–15°N band allowing for dynamical balance between the depth and tilt of the tropical thermocline and the model intrinsic biased wind. A sensitivity experiment to the latitudinal extension of no-nudging zone (1°S–1°N instead of 15°, hereafter referred to as NOEQ) has been carried out. In this paper, we concentrate our analyses on two specific regions: the tropical Pacific and the North Atlantic basins. In the Pacific, we show that the first year of the forecasts is characterized by a quasi-systematic excitation of El Niño-Southern Oscillation (ENSO) warm events whatever the starting dates. This, through ocean-to-atmosphere heat transfer materialized by diabatic heating, can be viewed for the coupled model as an efficient way to rapidly adjust to its own biased climate mean state. Weak cold ENSO events tend to occur the second year of the forecast due to the so-called discharge–recharge mechanism while the spurious oscillatory behavior is progressively damped. The latter mechanism is much more pronounced in retrospective forecasts initialized from the NOEQ configuration for which the ENSO flip-flop is still detectable at leadtime 4 year. Associated atmospheric teleconnections interfere worldwide with regional drifts, especially in the North Pacific and more remotely in the North Atlantic. In the latter basin, the drift can be interpreted as the model response to intrinsic atmospheric circulation biases found in the stand-alone atmosphere component of the model, which project onto the negative phase of the North Atlantic Oscillation. A fast adjustment (up to ~5-year leadtime) occurs leading to a rapid slackening of both the vertical (Atlantic meridional overturning circulation) and horizontal circulations, especially in the subpolar gyre. Slower adjustment of the entire water masses distribution in the North Atlantic then takes over involving several mechanisms. We show that a weak feedback is locally present between the atmospheric circulation and the ocean drift that controls the timescale of the setting of the coupled model biases. © 2015, The Author(s). Citation statistics:
资源类型: 期刊论文
标识符: http://119.78.100.158/handle/2HF3EXSE/53810
Appears in Collections: 过去全球变化的重建
There are no files associated with this item.
作者单位: CERFACS/CNRS, URA1875, 42 Av. G. Coriolis, Toulouse Cedex, France
Recommended Citation:
Sanchez-Gomez E.,Cassou C.,Ruprich-Robert Y.,et al. Drift dynamics in a coupled model initialized for decadal forecasts[J]. Climate Dynamics,2016-01-01,46(2017-05-06)