Enso Forecast Historical Data Using The Method Of Numerical Simulation Studies

To reach the end of not only utilizing the state of the art contribution in dynamical method but also absorbing the information provided by the long term historic datasets including the historical hindcast and simulation data in the dynamical ENSO prediction field, based on the national climate center inter-mediate ocean-atmosphere coupled model(NCCo),and centered on the mainly affecting factors of ENSO prediction, in this paper we comprehensively explore the method of utilizing the historical information in several key fields as mentioned follows and get many exciting findings which imply the well applicable perspective: the model error correction, the initialization process, the data assimilation and the coupled process of ocean and atmosphere model. The major results in this paper are described as follows:1) To reduce the influence of model error on dynamic ENSO prediction, following the principle of skillfully utilizing the historical data, we use a model error correction method constrained by the historically analogical error information to develop the dynamically analogical error correction system on the Lorenz model and NCCo model, respectively. The correspondingly academic and technical issues are explored, and the main results exhibit that in theory the prediction error can be reduced greatly for a dynamic model by using the dynamically analogical error correction method. But in general, the method is greatly affected by the length of historical datasets and the degree of model error. As the relative impact of the above two factors on the method, the method is more sensitive to the former. Besides, in the dynamically analogical error correction system, there are several parameters as mentioned follows which can also exert their impact on the prediction skill: firstly, the effect of analogue degree including the part analogue and comprehensive analogue is compared, the results exhibit that in a coupled system the comprehensive analogue is much better than the part analogue for the model in this paper, because the former can really depict the analogue degree between the current initial value and its historical partners, thus leading to a exact estimation of model error. Secondly, the investigation on the effect of the re-estimate period of error (RPE) denotes that RPE is also a crucial parameter to this model. Usually, there is an optimal combination between the RPE of atmosphere and ocean model under different analogue degrees to make a good prediction. Thirdly, the results in this paper also display that there are finite analogue samples in the datasets that the authors hold, and the hindcasting skill has a linear response to the analogue sample sizes due mainly to the fact that more analogue samples can supply more error information to the model thus leading to better estimation of model error and more improvement of prediction skill. Furthermore, the effect of the different scaling methods on the prediction skill of the system is also compared, the results show that the normalization method has the best effect contrasted to standardization method and maximum method due mainly to the former can loose the impact of dimension on searching the analogue from the historic datasets.2)In order to improve the quality of model initial values, reduce the uncertainty in the ENSO prediction arose by the initial errors, and resolve the dynamically inharmonic problem in the original initialization scheme of the NCCo model system, we develop a new dynamically harmonic initialization method based on a data reconstructed method which can separate the model compatible part of information from the observational data by using the dynamically harmonic information contained in the long term coupled simulation data. The comprehensive verification on the prediction skill and analysis on the relative mechanism for the new scheme are further discussed. The results exhibit that the information reconstructed method can really reach the end of getting the model compatible part of information from the observational data and realizing the harmony between the dynamical model and dataset through inversion of the dynamically harmonic information in the long term coupled simulation results. Besides, the results also show that the scheme can not only keep the merit of utilizing many observational information but also avoid the defect of dynamical disharmony, which is due to the new scheme can greatly squeeze out the high frequency and small scale noise arose by the inharmonic shortcoming in the original initialization scheme, and preserve the ENSO scale signals in the initial field which can adapt to the NCCo model quite well.3)To avoid the failure in the data assimilation caused by the incompatibility between dynamic model and observational data, we explore impact of the model compatible part of observational information on the affect of assimilation, based on a method of getting the model compatible part of information from the observational data. The researches find that the failure of assimilating the SSTA data in Chen 1997's work is mainly caused by the simpleness of the ZC model. To ZC model, the assimilated observational SSTA data is so complex that it can not adapt to the data, thus the initial fields contain many small scale disturbances when the observational data is assimilated, which will grow up rapidly in the integrating process and contaminate the prediction result. Furthermore, to two kinds of assimilation circumstance as the uncoupled and coupled one, the assimilation experiment shows that the effect of the former is much poor than the latter whatever the data that will be assimilated in the model: the observational data containing full information or just the model compatible part of one. While, we also find that the effect of assimilating the model compatible part of observational information is always better than assimilating the observational data containing full information whenever the assimilate circumstance is coupled or not. Just because the initial fields that are formatted by assimilating the model compatible part of observational information exhibit primary ENSO scale observational information; however the initial fields formatted by assimilating the full information observational data blend both the ENSO scale and the small scale information which is incompatible with the model.4)To realize the aim of maintaining the beneficial signals in the atmosphere while squeezing out the noises that may contaminate the prediction result, and improving the forecasting skill of the model, we develop an atmospheric noises filtered method by utilizing the relative information in the diagnostic research on the physical mechanism of ENSO. The numerical verification on this method denotes that the prediction skill on the ENSO can be improved to a certain extent and the false sea surface temperature anomaly caused by nonlinear increase of the non-ENSO scale disturbs contained in the atmosphere can be reduced greatly after squeezing out this part of disturbs during the coupled process of ocean and atmosphere. However, by filtering the atmosphere noises, the improvement on the prediction skill of Nino3 index are rather small within the lead-time of 6 months, but great improvement occurs after this period, which may reflect slow varying property of the ocean. The improvement of prediction skill on ENSO after filtering of the atmospheric noises may depend mainly on restraining the excessive development of equatorial westerly anomaly, and improving the corresponding ocean vertical movement over this region, thus the prediction skill on the sea surface temperature of the region is greatly ameliorated, and the forecasting error is reduced accordingly. The above results imply that the utilizing of information containing in the mechanism simulation results of the ENSO can not only help us to improve the prediction skill on the ENSO but also to resolve the problem that the diagnostics results are always leaved unused that there is a great gap between the statistic and dynamic method, and may help to enhance the connection between those two methods.