Richard Kleeman

A review of the predictability and prediction of ENSO

A hierarchy of El Nino-Southern Oscillation (ENSO) prediction schemes has been developed during the Tropical Ocean-Global Atmosphere (TOGA) program which includes statistical schemes and physical models. The statistical models are, in general, based on linear statistical techniques and can be classified into models which use atmospheric (sea level pressure or surface wind) or oceanic (sea surface temperature or a measure of upper ocean heat content) quantities or a combination of oceanic and atmospheric quantities as predictors. The physical models consist of coupled ocean-atmosphere models of varying degrees of complexity, ranging from simplified coupled models of the “shallow water” type to coupled general circulation models. All models, statistical and physical, perform considerably better than the persistence forecast in predicting typical indices of ENSO on lead times of 6 to 12 months. The TOGA program can be regarded as a success from this perspective. However, despite the demonstrated predictability, little is known about ENSO predictability limits and the predictability of phenomena outside the tropical Pacific. Furthermore, the predictability of anomalous features known to be associated with ENSO (e.g., Indian monsoon and Sahel rainfall, southern African drought, and off-equatorial sea surface temperature) needs to be addressed in more detail. As well, the relative importance of different physical mechanisms (in the ocean or atmosphere) has yet to be established. A seasonal dependence in predictability is seen in many models, but the processes responsible for it are not fully understood, and its meaning is still a matter of scientific discussion. Likewise, a marked decadal variation in skill is observed, and the reasons for this are still under investigation. Finally, the different prediction models yield similar skills, although they are initialized quite differently. The reasons for these differences are also unclear.

A mechanism for generating ENSO decadal variability

A coupled ocean-atmosphere model of the Pacific basin is used to illustrate a mechanism by which El Nino and the Southern Oscillation (ENSO) may be modulated on decadal time scales. For reasonable choices of model parameters, solutions exhibit two types of oscillation, an ENSO-like interannual mode and a decadal one. The decadal mode affects the equatorial zone by means of an oceanic teleconnection that involves transport variations of the North Pacific Subtropical Cell. Since almost half of the cool, thermocline water that upwells in the eastern equatorial Pacific participates in this cell, these variations significantly alter the extent of the cold tongue, and hence provide an efficient mechanism for modulating ENSO.

Rectification of the Madden–Julian Oscillation into the ENSO Cycle

An ocean general circulation model, forced with idealized, purely oscillating wind stresses over the western equatorial Pacific similar to those observed during the Madden‐Julian oscillation (MJO), developed rectified low-frequency anomalies in SST and zonal currents, compared to a run in which the forcing was climatological. The rectification in SST resulted from increased evaporation under stronger than normal winds of either sign, from correlated intraseasonal oscillations in both vertical temperature gradient and upwelling speed forced by the winds, and from zonal advection due to nonlinearly generated equatorial currents. The net rectified signature produced by the MJO-like wind stresses was SST cooling (about 0.48C) in the west Pacific, and warming (about 0.18C) in the central Pacific, tending to flatten the background zonal SST gradient. It is hypothesized that, in a coupled system, such a pattern of SST anomalies would spawn additional westerly wind anomalies as a result of SST-induced changes in the low-level zonal pressure gradient. This was tested in an intermediate coupled

Stochastic forcing of ENSO by the intraseasonal oscillation

Abstract Using the ideas of generalized linear stability theory, the authors examine the potential role that tropical variability on synoptic–intraseasonal timescales can play in controlling variability on seasonal–interannual timescales. These ideas are investigated using an intermediate coupled ocean–atmosphere model of the El Nino–Southern Oscillation (ENSO). The variability on synoptic–intraseasonal timescales is treated as stochastic noise that acts as a forcing function for variability at ENSO timescales. The spatial structure is computed that the stochastic noise forcing must have in order to enhance the variability of the system on seasonal–interannual timescales. These structures are the so-called stochastic optimals of the coupled system, and they bear a good resemblence to variability that is observed in the real atmosphere on synoptic and intraseasonal timescales. When the coupled model is subjected to a stochastic noise forcing composed of the stochastic optimals, variability on seasonal–inte...

Greenhouse Warming, Decadal Variability, or El Niño? An Attempt to Understand the Anomalous 1990s

The dominant variability modes in the Tropics are investigated and contrasted with the anomalous situation observed during the last few years. The prime quantity analyzed is anomalous sea surface temperature (SST) in the region 308S-608N. Additionally, observed tropical surface wind stress fields were investigated. Further tropical atmospheric information was derived from a multidecadal run with an atmospheric general circulation model that was forced by the same SSTs. The tropical SST variability can be characterized by three modes: an interannual mode (the El Nino-Southern Oscillation (ENSO)), a decadal mode, and a trend or unresolved ultra- low-frequency variability. The dominant mode of SST variability is the ENSO mode. It is strongest in the eastern equatorial Pacific, but influences also the SSTs in other regions through atmospheric teleconnections, such as the Indian and North Pacific Oceans. The ENSO mode was strong during the 1980s, but it existed with very weak amplitude and short period after 1991. The second most energetic mode is characterized by considerable decadal variability. This decadal mode is connected with SST anomalies of the same sign in all three tropical oceans. The tropical Pacific signature of the decadal mode resembles closely that observed during the last few years and can be characterized by a horseshoe pattern, with strongest SST anomalies in the western equatorial Pacific, extending to the northeast and southeast into the subtropics. It is distinct from the ENSO mode, since it is not connected with any significant SST anomalies in the eastern equatorial Pacific, which is the ENSO key region. However, the impact of the decadal mode on the tropical climate resembles in many respects that of ENSO. In particular, the decadal mode is strongly linked to decadal rainfall fluctuations over northeastern Australia in the observations. It is shown that the anomalous 1990s were dominated by the decadal mode. Considerable SST variability can be attributed also to a linear trend or unresolved ultra-low-frequency vari- ability. This trend that might be related to greenhouse warming is rather strong and positive in the Indian Ocean and western equatorial Pacific where it accounts for up to 30% of the total SST variability. Consistent with the increase of SST in the warm pool region, the trends over the tropical Pacific derived from both the observations and the model indicate a strengthening of the trade winds. This is inconsistent with the conditions observed during the 1990s. If the wind trends reflect greenhouse warming, it must be concluded that the anomalous 1990s are not caused by greenhouse warming. Finally, hybrid coupled ocean-atmosphere model experiments were conducted in order to investigate the sensistivity of ENSO to the low-frequency changes induced by the decadal mode and the trend. The results indicate that ENSO is rather sensitive to these changes in the background conditions.

Measuring Dynamical Prediction Utility Using Relative Entropy

A new parameter of dynamical system predictability is introduced that measures the potential utility of predictions. It is shown that this parameter satisfies a generalized second law of thermodynamics in that for Markov processes utility declines monotonically to zero at very long forecast times. Expressions for the new parameter in the case of Gaussian prediction ensembles are derived and a useful decomposition of utility into dispersion (roughly equivalent to ensemble spread) and signal components is introduced. Earlier measures of predictability have usually considered only the dispersion component of utility. A variety of simple dynamical systems with relevance to climate and weather prediction is introduced, and the behavior of their potential utility is analyzed in detail. For the climate systems examined here, the signal component is at least as important as the dispersion in determining the utility of a particular set of initial conditions. The simple ‘‘weather’’ system examined (the Lorenz system) exhibited different behavior with the dispersion being more important than the signal at short prediction lags. For longer lags there appeared no relation between utility and either signal or dispersion. On the other hand, there was a very strong relation at all lags between utility and the location of the initial conditions on the attractor.

The Linear Response of ENSO to the Madden–Julian Oscillation

The possibility that the tropical Pacific coupled system linearly amplifies perturbations produced by the Madden–Julian oscillation (MJO) is explored. This requires an estimate of the low-frequency tail of the MJO. Using 23 yr of NCEP–NCAR reanalyses of surface wind and Reynolds SST, we show that the spatial structure that dominates the intraseasonal band (i.e., the MJO) also dominates the low-frequency band once the anomalies directly related to ENSO have been removed. This low-frequency contribution of the intraseasonal variability is not included in most ENSO coupled models used to date. Its effect in a coupled model of intermediate complexity has, therefore, been studied. It is found that this “MJO forcing” (MJO) can explain a large fraction of the interannual variability in an asymptotically stable version of the model. This interaction is achieved via linear dynamics. That is, it is the cumulative effect of individual events that maintains ENSOs in this model. The largest coupled wind anomalies are initiated after a sequence of several downwelling Kelvin waves of the same sign have been forced by MJO. The cumulative effect of the forced Kelvin waves is to persist the (small) SST anomalies in the eastern Pacific just enough for the coupled ocean–atmosphere dynamics to amplify the anomalies into a mature ENSO event. Even though MJO explains just a small fraction of the energy contained in the stress not associated with ENSO, a large fraction of the modeled ENSO variability is excited by this forcing. The characteristics that make MJO an optimal stochastic forcing for the model are discussed. The large zonal extent is an important factor that differentiates the MJO from other sources of stochastic forcing.

A new intermediate coupled model for El Niño simulation and prediction

A new intermediate coupled model (ICM) is developed and used to simulate and predict sea surface temperature (SST) variability in the tropical Pacific. The ocean component is based on an intermediate complexity model developed by Keenlyside and Kleeman [2002] that is an extension of the McCreary [1981] baroclinic modal model to include varying stratification and partial nonlinearity effects, allowing realistic simulation of the mean equatorial circulation and its variability. An empirical procedure is developed to parameterize subsurface entrainment temperature (Te) in terms of sea surface pressure (SSP) anomalies. The ocean model is then coupled to a statistical atmospheric model. The coupled system realistically produces interannual variability associated with El Nino. Hindcasts are made during the period 1980-1997 for lead times out to 12 months. Observed SST anomalies are the only field to be incorporated into the coupled system to initialize predictions. Predicted SST anomalies from this model do not show obvious systematic biases. Another striking feature is that the model skill beats persistence at all lead times over the central equatorial Pacific.

On the Dependence of Hindcast Skill on Ocean Thermodynamics in a Coupled Ocean-Atmosphere Model

Abstract Three different mechanisms for the generation of ENSO SST anomalies within a simplified tropical Pacific Ocean model are examined: thermocline depth changes, Ekman-induced upwelling anomalies, and zonal advection changes. The effect of varying the relative influence of these terms on the realism of tropical Pacific coupled models is analyzed. The principal tool used to assess such realism is hindcast skill, with forced ocean and oscillatory behavior also being examined. Of the mechanisms considered, thermocline perturbations are shown to be crucially important for high coupled-model hindcast skill. Furthermore, it is concluded that the realism of the model (as measured by hindcast skill) deteriorates markedly when the influence on SST of Ekman upwelling becomes greater than a small fraction of the thermocline influence. This provides strong evidence for the hypothesis that Ekman upwelling anomalies (which are essentially a local response to wind stress anomalies) have only a small influence on th...

The Response of a Coupled Model of ENSO to Observed Estimates of Stochastic Forcing

In this work the role that observed intraseasonal atmospheric variability may play in controlling and maintaining ENSO variability is examined. To this end, an asymptotically stable intermediate coupled model of El Nino- Southern Oscillation (ENSO) is forced with observed estimates of stochastic forcing, which are defined to be the part of the atmospheric variability that is apparently independent of the ocean circulation. The stochastic forcing (SF) was estimated from 51 yr (1950-2000) of NCEP-NCAR reanalyses of surface winds and net surface heat flux, 32 yr (1950-81) of reconstructed sea surface temperatures (SST), and 19 yr (1982-2000) of Reynolds SST in the tropical Pacific. The deterministic component of the surface wind and heat flux anomalies that can be linearly related to SST anomalies was estimated using the singular value decomposition of the covariance between the anomaly fields, and was then removed from the atmospheric anomaly fields to recover the stochastic component of the ocean surface forcing. Principal component analysis reveals that the stochastic component has no preferred mode of variability, exhibits decorrelation times of a few days, and has a spectrum that is indis- tinguishable from red noise. A 19-yr stochastically forced coupled model integration qualitatively shows some similarities with the observed equatorial SST. The robustness of this result is checked by performing different sensitivity experiments. The model mostly exhibits a linear (and nonnormal) response to the low-frequency tail of SF. Using the ideas of generalized linear stability theory, the dynamically important contributions of the SF are isolated, and it is shown that most of the variability in the stochastically forced model solution is produced by stochastically induced Kelvin waves forced in the western and central Pacific. Moreover, the two most dynamically important patterns of stochastic forcing (which account for 71% of the expected variance in the model response) describe eastward propagation of the forcing similar to the MJO. The results of this study support the hypothesis that a significant fraction of ENSO variability may be due to SF, and suggest that a better understanding of the influence of SF on the ocean surface in the western/central Pacific may be required in order to understand the predictability of ENSO.