多普勒天气雷达原理与应用8-雷达探测算法-3课件.pptVIP

预报雷暴生成和演变的科学基础 雷暴生命史及其演化特征 雷暴生成过程频繁地发生在边界层辐合线附近 Auto-Nowcaster NCAR Thunderstorm Auto-nowcaster NCAR Thunderstorm Auto-nowcaster Auto-nowcaster 弱点: - automated detection of convergence lines needs considerable improvement. - detailed stability information is not available - need to improve methods for identifying growing cumulus Nowcasting Parameters 雷暴大小和增长 Storm size is color coded (m2). Large storms tend to live longer. 描述雷暴的参数 Trend in growth size is colored coded (m2/hr). Based on the visible and IR data clouds are automatically classified by type. Note color classification table on the right of the picture. Toggle with the visible picture to check classifications The difference in infrared cloud top temperatures between time periods is used to estimate if the cumulus clouds are growing or not. This picture indicates that the NW-SE line of cumulus is cooling (blue and green colors) and thus growing. 描述云的参数 Nowcasting Parameters NCAR Thunderstorm Auto-nowcaster 数据输入 预报参数 结合预报参数的方法 Methodology for Combining Nowcast Parameters Forecast Parameters are Combined using Fuzzy Logic Concepts. Each Forecast Parameter is converted to a number between -1 and 1 which relates to likelihood of storm initiation, growth and dissipation (-1 very unlikely and 1 very likely). Example given for boundary relative low-level shear. Low-level shear Membership function (converts low-level shear to thunderstorm likelihood) Likelihood Methodology for Combining Nowcast Parameters Each likelihood field is multiplied by a numerical weight relative to its importance. These fields are then summed to provide the final likelihood field. This process is done separately for the likelihood of storm initiation and likelihood of growth/decay for existing echoes. The membership functions and weights are defined by the forecaster and can be easily modified. The graph is an example of a membership function that converts the final growth/decay likelihood field