The role of atmospheric internal variability on the prediction skill of interannual North Pacific sea-surface temperatures
The sensitivity of the sea-surface temperature (SST) prediction skill to the atmospheric internal variability (weather noise) in the North Pacific (20∘–60∘N;120∘E–80∘W) on decadal timescales is examined using state-of-the-art Climate Forecasting System model version 2 (CFS) and a variation of CFS in an Interactive Ensemble approach (CFSIE), wherein six copies of atmospheric components with different perturbed initial states of CFS are coupled with the same ocean model by exchanging heat, momentum and fresh water fluxes dynamically at the air-sea interface throughout the model integrations. The CFSIE experiments are designed to reduce weather noise and using a few ten-year long forecasts this study shows that reduction in weather noise leads to lower SST forecast skill. To understand the pathways that cause the reduced SST prediction skill, two twenty-year long forecasts produced with CFS and CFSIE for 1980-2000 are analyzed for the ocean subsurface characteristics that influence SST due to the reduction in weather noise in the North Pacific. The heat budget analysis in the oceanic mixed layer across the North Pacific reveals that weather noise significantly impacts the heat transport in the oceanic mixed layer. In the CFSIE forecasts, the reduced weather noise leads to increased variations in heat content due to shallower mixed layer, diminished heat storage and enhanced horizontal heat advection. The enhancement of the heat advection spans from the active Kuroshio regions of the east coast of Japan to the west coast of continental United States and significantly diffuses the basin-wide SST anomaly (SSTA) contrasts and leads to reduction in the SST prediction skill in decadal forecasts.
The variability in the SST on interannual time-scales in the North Pacific Region (NPR; 20∘-60∘N -120∘E-100∘W) is an important aspect in understanding the Pacific decadal variability (Mantua et al. 1997; Power et al. 1999; Deser et al. 2004). The SST variability in the NPR is generally characterized by strong atmosphere-ocean variations (Trenberth and Hurrell 1994) and this variability is best described as a combination of multiple physical modes (Deser et al. 2010).
Several mechanisms are proposed to explain the variability of SST in the NPR. These mechanisms are broadly generalized into three categories: (1) stochastic forcing of weather noise, (2) remote forcing of El Niño Southern Oscillation (ENSO) through atmospheric bridge phenomenon, and (3) subsurface ocean heat transport from Kuroshio and its extension. The view of the first category is that due to the higher heat capacity of oceans in the earth-climate system, stochastic forcing by weather noise acts to redden the low-frequency Ocean SST variability spectra (Hasselmann 1976; Frankignoul and Hasselmann 1977; Barsugli and Battisti 1998). The second category is based on the statistical correlations that lead to hypothesize that the variability in the North Pacific SSTs are dominated by air-sea interactions that are remotely forced through atmosphere by ENSO SST anomalies (Alexander et al. 2002). The third category stems from the simple conceptual models in association with analyzing in-situ datasets (Deser et al. 1996; Saravanan and McWilliams 1998; Qiu 2003) as recent advances in the satellite imagery and ocean data assimilation that allow to examine the volumetric heat transport from Kuroshio and Ryuku current streams (Andres et al. 2009); these subsurface heat transport processes are shown to influence the North Pacific SSTs on decadal time scales.
There are also few studies that examine the SST variability in the NPR as a combination of these categories. For instance, NPR variability is argued to be a combination of stochastic forcing due to the weather noise and the ENSO forcing through atmospheric bridge (Newman et al. 2003). SST variability in the NPR is also proposed as a combination of the weather noise forcing and the SST reemergence mechanism due to the seasonal variations in the mixed layer depths (Deser et al. 2003). The impact of subsurface heat transport variations on SST variability due to the wind stress variability and the weather noise forcing were also examined in the previous studies. A study by Deser et al. (1996) showed that a lead-lag correlation between mean curl of the wind stress and subsurface thermal anomalies in the NPR is established through the ventilated thermocline mechanism (Luyten et al. 1983). Though not directly linked to SST variability in the NPR, the interannual heat content variations in the Kuroshio and its extension were well captured by forcing a simplified ocean model by predominantly Rossby wave driven atmospheric momentum (Kawabe 2001). However, the pathways that link the subsurface heat content and the SST in the NPR could be well examined with coupled general circulation model (CGCM) experiments.
This study is motivated by advancements in the coupled modeling that allow to realistically separate impact of the atmospheric internal dynamics (weather noise) on North Pacific SST. The primary goal is to understand the pathways that influence the variability in the North Pacific SST in the absence of weather noise at the ocean-atmosphere interface in a CGCM. To achieve this a CGCM simulation should be compared against the exact same CGCM counterpart that has no weather noise. In this study, the impact of weather noise on the North Pacific SST variability is analyzed by comparing a twenty-year long forecast produced by state-of-the-art Climate Forecasting System version 2 (CFS; Saha et al. (2012)) against another identical forecast produced by the same model but with reduced weather noise.
Interactive Ensemble method (Kirtman and Shukla 2002) is employed to reduce the weather noise fluxes of heat, momentum and fresh water at the ocean-atmosphere interface in the CFS forecasts. This variant of CFS in Interactive Ensemble framework (CFSIE) produces a modified version of CFS by coupling the ocean component with the averaged state of fluxes from several atmospheric components. To completely eliminate the weather noise, fluxes produced by infinitely large number of atmospheric components at air-sea interface shall have to be coupled with the ocean component at an infinitesimal time intervals during the model integrations. However, this study utilizes six realizations of atmospheric components to interact with one ocean component at a model integration time step of every 30 mins as the previous studies show that coupling the mean of six atmospheric realizations with ocean component at half hour integrations can significantly reduce the weather noise (Kirtman et al. 2005; Stan and Kirtman 2008). This way of reducing weather noise was shown to influence the low-frequency SST variability in CGCMs (Yeh and Kirtman 2004; Schneider and Fan 2007).
The manuscript is organized as follows. Section 2 contains the numerical experiments that produced the 20-year long Climate Forecasting System forecast and a similar forecast with reduced weather noise and the decadal forecast data that is used in deducing the SSTA prediction skill in CFS and CFSIE forecasts. The analysis of the CFS and CFSIE experiments along with discussions are presented in Section 3 and conclusions of this study are summarized in Section 4.
2 Description of numerical experiments
Several ten-year long forecasts are produced with CFS and CFSIE to examine the prediction skill in the decadal forecasts are detailed below. Two twenty-year long forecasts with CFS and CFSIE are also performed to understand the differences of underlying processes in the absence of weather noise. All the forecasts produced in this study are initialized only in the beginning and no initialization or data assimilation is performed during the course of the model integration.
2.1 Twenty-year long forecasts
The twenty-year long forecast produced from the state-of-the-art coupled global forecast model CFS is compared against similar forecast produced by CFSIE configuration (Kirtman and Shukla (2002)). A brief discussion on the atmosphere and ocean model components used in the CFS is documented by Narapusetty et al. (2012).
In the Interactive Ensemble configuration, the momentum, heat and fresh water fluxes that atmospheric component exchanges with ocean component are controlled in a way that the higher frequencies in these fluxes are reduced at the ocean-atmosphere interface. This is achieved by concurrently coupling the mean of the fluxes simulated by several atmospheric components with a single ocean component that in turn forces each atmospheric component with the same SST. This setup leads to a reduction in the atmospheric internal variability in the coupled system. For a more detailed discussion on how the Interactive Ensemble approach reduces the weather noise in a CGCM, refer to Kirtman et al. (2005).
The twenty-year long CFS and CFSIE forecasts are produced from 1 November 1980 to 31 October 2000. The atmosphere and the ocean initial conditions for the forecasts are derived from CFS reanalysis products (available online at http://cfs.ncep.noaa.gov/cfsr/). The initial conditions needed for the six copies of the atmospheric components in CFSIE integration are obtained from the CFS reanalysis data separated by 6 hours starting from 00Z of 1 November 1980.
2.2 Ten-year long forecasts
3 Results and discussion
3.1 Prediction skill of North Pacific SSTs
3.2 Relationship between Kuroshio current on North Pacific SSTs
In this section, the influence of heat transport by Kuroshio and its extension on north Pacific SSTs is explored to understand the lower prediction skill that is detected in the IEf. The 20-year long integrations for CFS and CFSIE experiments as explained in Section 2a are used to examine the physical processes that lead to the change in the prediction skills between CFSf and IEf forecasts.
The LHS of Eq. 1 explains the variation of heat in the mixed layer (referred to as VH hereafter), second term in the RHS explains the heat fluxes by advection and diffusion, and the third term includes the advective and diffusive fluxes at the interface z=H that are accounted for the change in the mixed layer depths from October to February. The second and third terms together in the RHS explain the heat transfer due to advective and diffusive fluxes (referred to as HT hereafter). Therefore, from the heat budget equation (Eq. 1), ocean heat transport in the mixed layer could be estimated as HT = VH −Q∗, where Q∗ is the first term in the RHS of Eq. 1 that represents the average heat flux at the sea surface.
The motivation for this study stems from the reduced prediction skill in North Pacific region (20∘-60∘N and 120∘E-80∘W, referred as NPR) SSTA forecasts as produced by CMIP5-style decadal forecasts with state-of-the-art global climate forecasting system model (CFS) in an Interactive Ensemble set-up (CFSIE), which reduces atmospheric internal variability. The main goal of this study is to examine and understand the pathways that prompted the reduced prediction skill in the CFSIE forecasts in comparison with the regular CFS forecasts. Two 20-year long forecasts produced with CFS and CFSIE for 1980-2000 are compared for the subsurface characteristics that influence SST in the North Pacific Region. The detailed heat budget analysis in the oceanic mixed layer of the CFS and CFSIE experiments reveals that the heat transport in the oceanic mixed layer is driven significantly by the atmospheric internal variability (weather noise). In the CFSIE, the reduction in the weather noise leads to shallower mixed layer depths and the increase in the heat content variations associated with lower heat flux result in diminished heat content storage and enhanced horizontal heat advection throughout the basin. Due to the deeper mixed layer in CFS, the available heat flux channels to raise the temperature in the the mixed layer and this diminishes the heat available for the basin-wide horizontal transport.
The enhanced HT in the CFSIE is basin-wide in the spatial extent and spans from the active Kuroshio regions of east coast of Japan to the west coast of continental United States. This enhanced horizontal advection of heat in the oceanic mixed layer acts in a way to diffuse the SSTA contrasts and leads to reduced skills in decadal forecasts.
This research is funded by NSF AGS-1338427, NOAA NA14OAR4310160, and NASA NNX14AM19G grants. Author wishes to thank Dr. Cristiana Stan of George Mason University for implementing the Interactive Ensemble method in the CFS forecasts used in this study. Author also wishes to thank W. Lapenta and L. Uccellini for enabling the collaborative activities. Computing resources provided by NCAR are also gratefully acknowledged.
- Andres M, Park J-E, Wimbush M, Zhu X-H, Nakamura H, Kim K, Chang K-I (2009) Manifestation of the Pacific decadal oscillation in the Kuroshio. Geophys Res Lett, 36 (L16602)Google Scholar
- Kang Y, Noh Y, Yeh S-W (2010) Processes that infuence the mixed layer deepening during winter in the north pacific. J Geophys Res, 115 (C12004)Google Scholar
- Levitus S (1982) Climatological atlas of the world ocean. NOAA Prof. Paper, (13), 173 ppGoogle Scholar
- Narapusetty B, Stan C, Kumar A (2014) Bias correction methods for decadal sea-surface temperature forecasts. Tellus A, ( 10.3402/tellusa.v66.23681), ISSN 1600–0870
- Saha S, et al. (2012) The NCEP Climate Forecast System veresion 2. J. Clim., (To be submitted)Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.