幸运飞艇计划

To deal with the inherent uncertainties in weather and climate聽predictions, the common practice is to apply ensemble methods to produce聽a probability distribution of variables of interest. Informatively聽summarizing such probability distributions is a non-trivial task and the聽commonly used means and standard deviations result in the loss of聽crucial information, especially in the case of multimodal distributions聽with distinct likely outcomes. In this project, we aim at machine聽learning and visualization methods to summarize results from ensemble聽models in a physically more meaningful and conceptually more intuitive聽way, while at the same time preserving significant information contained聽within the ensemble.