Stochastic Parameterisation of Precipitation in Ensemble Simulations with an Intermediate Complexity General Circulation Model

点击翻译按钮获取中文摘要

In the recent years, weather prediction has begun to produce ensembles of individual forecasts which start with different initial conditions and are forced randomly during the forecast period. The different initial conditions correspond to uncertainties in the observations and the initialisation. The random forces, which correspond to uncertainties in the model and to the variability of the unresolved small scale processes, are not included in the deterministic parameterisations. The stochastic ensemble prediction approach has shown up to improve the forecast, in particular precipitation, and to increase the spread within the ensembles which is used as a measure of predictability. However, many questions related to the noise used in weather forecast models are open and cannot be answered within costly weather forecast models. Therefore, the aim of this project is to study the noise properties and to detect optimal settings by means of a numerically efficient general circulation model (GCM) of intermediate complexity. In particular we consider: (1) different noise amplitudes and spatial and temporal correlations, (2) the impact of noise on different dynamical variables, mainly on humidity, (3) the sizes of the ensembles which are necessary for a sufficient exploration of weather states, and (4) the statistical properties of extreme events. Subsequently, the findings of the initial part of this project are tested and applied in comprehensive climate and weather forecast models.