数值预报经历了半个多世纪的发展,已成为当前主要的客观预报工具。在模式和资料状况给定的情况下,预报效果的改善很大程度上依赖于所采用的预报策略和方法。为此,全面回顾了国内外基于数值模式的预报策略和方法研究进展,认为采取统计一动力相结合、从历史资料中提炼信息的预报策略是提高数值预报水平的可行之路。最后在总结前人工作基础上,着重介绍了动力相似预报策略和方法的相关研究,特别是实际预报中的试验情况。
At present, numerical prediction has been a primary technique for objective forecast after half a century's development. Under the circumstance of model and data given, the improvement of predictive level is to great extent dependent on the prediction strategy and method that are employed. In the present paper, the study progress in prediction strategy and methodology on numerical model in the world, which primarily includes the correction of prediction errors, ensemble prediction techniques and so on, has been comprehensively reviewed. It is suggested that there exist a feasible approach to improving predictive level based on prediction strategy by combining statistical and dynamical methods together and extracting information from historical data. Moreover, on the basis of summarizing previous work, the studies on the strategy and methodology of dynamical analogue prediction developed in recent years are also introduced, especially including situations of prediction experiments in complex operational model.