A Short Term Regional Uncertainty Exploration Framework for Numerical Weather Model Ensembles
Numerical Weather Simulation Data has been used in weather forecasting widely.It comprises multiple meteorological elements and has been used to assist professionals in weather prediction.In practical applications,data from several different numerical models are simultaneously used to generate a better forecasting result.Therefore, in order to explore the multi-variable and multi-model data, uncertainty visualization for the data variance and the data accuracy are requested.In this paper, we propose a novel visual uncertainty exploration framework on the basis of analyzing the deviations of the historical numerical simulation data and the practical observation data.A stamp-based guiding tool is designed and implemented to explore the regional distribution of the uncertainty patterns within these data.The exploration pipeline originates from a single meteorological element, for example the rainfall amounts in 24 hours, and extends to the multiple elements under the guidance of the designed stamp map.By analyzing the distinctive stamps on a story board using the Boolean set operations,the regional distributions are generated, of which the regional information can be applied to identify from which the uncertainty comes and explore the relationship among elements in different areas and different ensembles models that otherwise would be hidden.Experimental results demonstrate that the regional distribution obtained from the pipeline can show some meteorological phenomena more clearly, and more regional information can be combined in a single map.
Uncertainty visualization multivariate data visualization numerical weather model ensemble weather forecasting geographic visualization
Hongsen Liao Li Chen Hui Zhang
Tsinghua University
国际会议
昆明
英文
193-211
2014-05-01（万方平台首次上网日期，不代表论文的发表时间）