Mesoscale Ensemble Forecasting and Predictability Studies

点击翻译按钮获取中文摘要

An ensemble forecast is a technique in which a numerical forecast model is used to generate several forecasts based on different initial conditions. It has been found that if the forecasts are consistent, then this indicates a more reliable forecast than if the forecasts diverge. While such techniques have been developed for synoptic and climate models, application of ensemble forecasts to local-scale numerical models is still a very new field.The fundamental goal of this research is to investigate the feasibility and usefulness of an ensemble approach in mesoscale prediction. Specifically, the Principal Investigator will investigate the advantages and limitations of ensemble versus deterministic forecasting in mesoscale models at high resolution. To this end, an advanced numerical forecast model will be used to address a range of issues pertaining to short-range ensemble forecasting, including those concerning perturbation strategies, ensemble configuration, and the structure and evolution of initial errors. The design of an ensemble forecasting system for mesoscale systems poses many challenges. Perhaps the most important concerns the initialization of an ensemble. The Principal Investigator will compare two different approaches for constructing the initial perturbations: a) the Monte-Carlo method based on observing systems simulation experiments, and b) the breeding of growing modes method. He will also investigate the importance of perturbing different parameters and coefficients in model physics, as well as perturbation of surface fields such as topography and sea surface temperature. The role of perturbing lateral boundary conditions will also be examined as part of this study. Other issues related to ensemble forecasting, such as minimum size of an ensemble, dependency of ensemble performance on resolution, and structure and growth of initial errors will be addressed. The aforementioned ensemble studies will be carried out on a total of ten cases, split evenly between summer and winter situations. A number of validation and evaluation techniques will be utilized to document the performance and characteristics of the ensembles. Particular attention will be paid to the reliability of ensemble-derived probability estimates and quantitative precipitation forecasts. Successful completion of this research will advance capabilities in short-term local weather forecasting.