Initial Error-induced Optimal Perturbations in ENSO Predictions, as Derived from an Intermediate Coupled Model

The initial errors constitute one of the main limiting factors in the ability to predict the E1 Nino-Southem Oscillation （ENSO） in ocean-atmosphere coupled models. The conditional nonlinear optimal perturbation （CNOP） approach was em- ployed to study the largest initial error growth in the E1 Nino predictions of an intermediate coupled model （ICM）. The optimal initial errors （as represented by CNOPs） in sea surface temperature anomalies （SSTAs） and sea level anomalies （SLAs） were obtained with seasonal variation. The CNOP-induced perturbations, which tend to evolve into the La Nifia mode, were found to have the same dynamics as ENSO itself. This indicates that, if CNOP-type errors are present in the initial conditions used to make a prediction of E1 Nino, the E1 Nino event tends to be under-predicted. In particular, compared with other seasonal CNOPs, the CNOPs in winter can induce the largest error growth, which gives rise to an ENSO amplitude that is hardly ever predicted accurately. Additionally, it was found that the CNOP-induced perturbations exhibit a strong spring predictability barrier （SPB） phenomenon for ENSO prediction. These results offer a way to enhance ICM prediction skill and, particularly, weaken the SPB phenomenon by filtering the CNOP-type errors in the initial state. The characteristic distributions of the CNOPs derived from the ICM also provide useful information for targeted observations through data assimilation. Given the fact that the derived CNOPs are season-dependent, it is suggested that seasonally varying targeted observations should be implemented to accurately predict ENSO events.

Relationships between the Limit of Predictability and Initial Error in the Uncoupled and Coupled Lorenz Models

In this study, the relationship between the limit of predictability and initial error was investigated using two simple chaotic systems： the Lorenz model, which possesses a single characteristic time scale, and the coupled Lorenz model, which possesses two different characteristic time scales. The limit of predictability is defined here as the time at which the error reaches 95% of its saturation level; nonlinear behaviors of the error growth are therefore involved in the definition of the limit of predictability. Our results show that the logarithmic function performs well in describing the relationship between the limit of predictability and initial error in both models, although the coefficients in the logarithmic function were not constant across the examined range of initial errors. Compared with the Lorenz model, in the coupled Lorenz model in which the slow dynamics and the fast dynamics interact with each other--there is a more complex relationship between the limit of predictability and initial error. The limit of predictability of the Lorenz model is unbounded as the initial error becomes infinitesimally small; therefore, the limit of predictability of the Lorenz model may be extended by reducing the amplitude of the initial error. In contrast, if there exists a fixed initial error in the fast dynamics of the coupled Lorenz model, the slow dynamics has an intrinsic finite limit of predictability that cannot be extended by reducing the amplitude of the initial error in the slow dynamics, and vice versa. The findings reported here reveal the possible existence of an intrinsic finite limit of predictability in a coupled system that possesses many scales of time or motion.

The Initial Errors in the Tropical Indian Ocean that Can Induce a Significant “Spring Predictability Barrier” for La Nina Events and Their Implication for Targeted Observations

Initial errors in the tropical Indian Ocean(IO-related initial errors) that are most likely to yield the Spring Prediction Barrier(SPB) for La Ni?a forecasts are explored by using the CESM model.These initial errors can be classified into two types.Type-1 initial error consists of positive sea temperature errors in the western Indian Ocean and negative sea temperature errors in the eastern Indian Ocean,while the spatial structure of Type-2 initial error is nearly opposite.Both kinds of IO-related initial errors induce positive prediction errors of sea temperature in the Pacific Ocean,leading to underprediction of La Nina events.Type-1 initial error in the tropical Indian Ocean mainly influences the SSTA in the tropical Pacific Ocean via atmospheric bridge,leading to the development of localized sea temperature errors in the eastern Pacific Ocean.However,for Type-2 initial error,its positive sea temperature errors in the eastern Indian Ocean can induce downwelling error and influence La Ni?a predictions through an oceanic channel called Indonesian Throughflow.Based on the location of largest SPB-related initial errors,the sensitive area in the tropical Indian Ocean for La Nina predictions is identified.Furthermore,sensitivity experiments show that applying targeted observations in this sensitive area is very useful in decreasing prediction errors of La Nina.Therefore,adopting a targeted observation strategy in the tropical Indian Ocean is a promising approach toward increasing ENSO prediction skill.

Robust Iterative Learning Controller for the Non-zero Initial Error Problem on Robot Manipulator

Industrial robot system is a kind of dynamic system w ith strong nonlinear coupling and high position precision. A lot of control ways , such as nonlinear feedbackdecomposition motion and adaptive control and so o n, have been used to control this kind of system, but there are some deficiencie s in those methods: some need accurate and some need complicated operation and e tc. In recent years, in need of controlling the industrial robots, aiming at com pletely tracking the ideal input for the controlled subject with repetitive character, a new research area, ILC (iterative learning control), has been devel oped in the control technology and theory. The iterative learning control method can make the controlled subject operate as desired in a definite time span, merely making use of the prior control experie nce of the system and searching for the desired control signal according to the practical and desired output signal. The process of searching is equal to that o f learning, during which we only need to measure the output signal to amend the control signal, not like the adaptive control strategy, which on line assesses t he complex parameters of the system. Besides, since the iterative learning contr ol relies little on the prior message of the subject, it has been well used in a lot of areas, especially the dynamic systems with strong non-linear coupling a nd high repetitive position precision and the control system with batch producti on. Since robot manipulator has the above-mentioned character, ILC can be very well used in robot manipulator. In the ILC, since the operation always begins with a certain initial state, init ial condition has been required in almost all convergence verification. Therefor e, in designing the controller, the initial state has to be restricted with some condition to guarantee the convergence of the algorithm. The settle of initial condition problem has long been pursued in the ILC. There are commonly two kinds of initial condition problems: one is zero initial error problem, another is non-zero initial error problem. In practice, the repe titive operation will invariably produce excursion of the iterative initial stat e from the desired initial state. As a result, the research on the second in itial problem has more practical meaning. In this paper, for the non-zero initial error problem, one novel robust ILC alg orithms, respectively combining PD type iterative learning control algorithm wit h the robust feedback control algorithm, has been presented. This novel robust ILC algorithm contain two parts: feedforward ILC algorithm and robust feedback algorithm, which can be used to restrain disturbance from param eter variation, mechanical nonlinearities and unmodeled dynamics and to achieve good performance as well. The feedforward ILC algorithm can be used to improve the tracking error and perf ormance of the system through iteratively learning from the previous operation, thus performing the tracking task very fast. The robust feedback algorithm could mainly be applied to make the real output of the system not deviate too much fr om the desired tracking trajectory, and guarantee the system’s robustness w hen there are exterior noises and variations of the system parameter. In this paper, in order to analyze the convergence of the algorithm, Lyapunov st ability theory has been used through properly selecting the Lyapunov function. T he result of the verification shows the feasibility of the novel robust iterativ e learning control in theory. Finally, aiming at the two-freedom rate robot, simulation has been made with th e MATLAB software. Furthermore, two groups of parameters are selected to validat e the robustness of the algorithm.

ENSO Predictions in an Intermediate Coupled Model Influenced by Removing Initial Condition Errors in Sensitive Areas： A Target Observation Perspective

Previous studies indicate that ENSO predictions are particularly sensitive to the initial conditions in some key areas（socalled ＂sensitive areas＂）. And yet, few studies have quantified improvements in prediction skill in the context of an optimal observing system. In this study, the impact on prediction skill is explored using an intermediate coupled model in which errors in initial conditions formed to make ENSO predictions are removed in certain areas. Based on ideal observing system simulation experiments, the importance of various observational networks on improvement of El Ni n？o prediction skill is examined. The results indicate that the initial states in the central and eastern equatorial Pacific are important to improve El Ni n？o prediction skill effectively. When removing the initial condition errors in the central equatorial Pacific, ENSO prediction errors can be reduced by 25%. Furthermore, combinations of various subregions are considered to demonstrate the efficiency on ENSO prediction skill. Particularly, seasonally varying observational networks are suggested to improve the prediction skill more effectively. For example, in addition to observing in the central equatorial Pacific and its north throughout the year,increasing observations in the eastern equatorial Pacific during April to October is crucially important, which can improve the prediction accuracy by 62%. These results also demonstrate the effectiveness of the conditional nonlinear optimal perturbation approach on detecting sensitive areas for target observations.

Diagnosing SST Error Growth during ENSO Developing Phase in the BCC_CSM1.1(m) Prediction System

In this study, the predictability of the El Nino-South Oscillation(ENSO) in an operational prediction model from the perspective of initial errors is diagnosed using the seasonal hindcasts of the Beijing Climate Center System Model,BCC;SM1.1(m). Forecast skills during the different ENSO phases are analyzed and it is shown that the ENSO forecasts appear to be more challenging during the developing phase, compared to the decay phase. During ENSO development, the SST prediction errors are significantly negative and cover a large area in the central and eastern tropical Pacific, thus limiting the model skill in predicting the intensity of El Nino. The large-scale SST errors, at their early stage, are generated gradually in terms of negative anomalies in the subsurface ocean temperature over the central-western equatorial Pacific,featuring an error evolutionary process similar to that of El Nino decay and the transition to the La Nina growth phase.Meanwhile, for short lead-time ENSO predictions, the initial wind errors begin to play an increasing role, particularly in linking with the subsurface heat content errors in the central-western Pacific. By comparing the multiple samples of initial fields in the model, it is clearly found that poor SST predictions of the Nino-3.4 region are largely due to contributions of the initial errors in certain specific locations in the tropical Pacific. This demonstrates that those sensitive areas for initial fields in ENSO prediction are fairly consistent in both previous ideal experiments and our operational predictions,indicating the need for targeted observations to further improve operational forecasts of ENSO.

Role of Parameter Errors in the Spring Predictability Barrier for ENSO Events in the Zebiak–Cane Model

ABSTRACT The impact of both initial and parameter errors on the spring predictability barrier （SPB） is investigated using the Zebiak Cane model （ZC model）. Previous studies have shown that initial errors contribute more to the SPB than parameter errors in the ZC model. Although parameter errors themselves are less important, there is a possibility that nonlinear interactions can occur between the two types of errors, leading to larger prediction errors compared with those induced by initial errors alone. In this case, the impact of parameter errors cannot be overlooked. In the present paper, the optimal combination of these two types of errors [i.e., conditional nonlinear optimal perturbation （CNOP） errors] is calculated to investigate whether this optimal error combination may cause a more notable SPB phenomenon than that caused by initial errors alone. Using the CNOP approach, the CNOP errors and CNOP-I errors （optimal errors when only initial errors are considered） are calculated and then three aspects of error growth are compared： （1） the tendency of the seasonal error growth; （2） the prediction error of the sea surface temperature anomaly; and （3） the pattern of error growth. All three aspects show that the CNOP errors do not cause a more significant SPB than the CNOP-I errors. Therefore, this result suggests that we could improve the prediction of the E1 Nifio during spring by simply focusing on reducing the initial errors in this model.

A Case Study of the Error Growth and Predictability of a Meiyu Frontal Heavy Precipitation Event

The Advanced Regional Eta-coordinate Model （AREM） is used to explore the predictability of a heavy rainfall event along the Meiyu front in China during 3-4 July 2003. Based on the sensitivity of precipitation prediction to initial data sources and initial uncertainties in different variables, the evolution of error growth and the associated mechanism are described and discussed in detail in this paper. The results indicate that the smaller-amplitude initial error presents a faster growth rate and its growth is characterized by a transition from localized growth to widespread expansion error. Such modality of the error growth is closely related to the evolvement of the precipitation episode, and consequently remarkable forecast divergence is found near the rainband, indicating that the rainfall area is a sensitive region for error growth. The initial error in the rainband contributes significantly to the forecast divergence, and its amplification and propagation are largely determined by the initial moisture distribution. The moisture condition also affects the error growth on smaller scales and the subsequent upscale error cascade. In addition, the error growth defined by an energy norm reveals that large error energy collocates well with the strong latent heating, implying that the occurrence of precipitation and error growth share the same energy source-the latent heat. This may impose an intrinsic predictability limit on the prediction of heavy precipitation.

Error Sensitivity Analysis in 10–30-Day Extended Range Forecasting by Using a Nonlinear Cross-Prediction Error Model

Extended range forecasting of 10-30 days, which lies between medium-term and climate prediction in terms of timescale, plays a significant role in decision-making processes for the prevention and mitigation of disastrous met- eorological events. The sensitivity of initial error, model parameter error, and random error in a nonlinear cross- prediction error （NCPE） model, and their stability in the prediction validity period in 1 0-30-day extended range fore- casting, are analyzed quantitatively. The associated sensitivity of precipitable water, temperature, and geopotential height during cases of heavy rain and hurricane is also discussed. The results are summarized as follows. First, the initial error and random error interact. When the ratio of random error to initial error is small （10＂5-10-2）, minor vari- ation in random error cannot significantly change the dynamic features of a chaotic system, and therefore random er- ror has minimal effect on the prediction. When the ratio is in the range of 10-1-2 （i.e., random error dominates）, at- tention should be paid to the random error instead of only the initial error. When the ratio is around 10 2-10-1, both influences must be considered. Their mutual effects may bring considerable uncertainty to extended range forecast- ing, and de-noising is therefore necessary. Second, in terms of model parameter error, the embedding dimension m should be determined by the factual nonlinear time series. The dynamic features of a chaotic system cannot be depic- ted because of the incomplete structure of the attractor when m is small. When m is large, prediction indicators can vanish because of the scarcity of phase points in phase space. A method for overcoming the cut-off effect （m 〉 4） is proposed. Third, for heavy rains, precipitable water is more sensitive to the prediction validity period than temperat- ure or geopotential height; however, for hurricanes, geopotential height is most sensitive, followed by precipitable water.

Spatial Propagation of Different Scale Errors in Meiyu Frontal Rainfall Systems

The spatial propagation of meso-and small-scale errors in a Meiyu frontal heavy rainfall event,which occurred in eastern China during 4-6 July 2003,is investigated by using the mesoscale numerical model MM5.In general,the spatial propagation of simulated errors depends on their horizontal scales.Small-scale（L 〈 100 km） initial error may spread rapidly as an isotropic circle through the sound wave.Then,many scattered convection-scale errors are triggered in moist convection zone that will spread abroad through the isotropic,round-shaped sound wave further more.Corresponding to the evolution of the rainfall system,several new convection-scale errors may be generated continuously by moist convection within the propagated round-shaped errors.Through the above circular process,the small-scale error increases in amplitude and grows in scale rapidly.Mesoscale（100 km 〈 L 〈 1000 km） initial error propagates up-and down-stream wavelike through the gravity wave,meanwhile migrating down-stream slowly along with the rainfall system by the mean flow.The up-stream propagation of the mesoscale error is very important to the error growth because it can accumulate error energy locally at a place where there is no moist convection and far upstream from the initial perturbation source.Although moist convection plays an important role in the rapid growth of errors,it has no impact on the propagation of meso-and small-scale errors.The diabatic heating could trigger,strengthen,and promote upscaling of small-scale errors successively,and provide ＂error source＂ to error growth and propagation.The rapid growth of simulated errors results from both intense moist convection and appropriate spatial propagation of the errors.

EFFECT OF INSTABILITIES OF FLOW ON MESOSCALE PREDICTABILITY OF WEATHER SYSTEMS

The numerical solution of Boussinesq equations is worked out as an initial-value problem to study the effect of the instabilities of flow on the initial error growth and mesoscale predictability. The development of weather systems depends on different dynamic instability mechanisms according to the spatial scales of the system and the development of mesoscale systems is determined by symmetric instability. Since symmetric instability dominates among the three types of dynamic instability, it makes the prediction of the associated mesoscale systems more sensitive to initial uncertainties. This indicates that the stronger instability leads to faster initial error growth and thus limits the mesoscale predictability. Besides dynamic instability, the impact of thermodynamic instability is also explored. The evolvement of convective instability manifests as dramatic variation in small spatial scale and short temporal scale, and furthermore, it exhibits the upscale growth. Since these features determine the initial error growth, the mesoscale systems arising from convective instability are less predictable and the upscale error growth limits the predictability of larger scales. The latent heating is responsible for changing the stability of flow and subsequently influencing the error growth and the predictability.