对多普勒雷达集合交叉相关外推技术的构造与实例检验

Construction and Case Verification on Doppler Radar Ensemble Extrapolation of Cross-Correlation

本文构造了集合的线性交叉相关方法,对多普勒雷达回波进行外推。因为线性交叉相关算法仅根据两个时次的回波外推下一时次的回波分布,且对于单个外推回波单元是线性变化的;而基于集合的线性交叉相关方法,考虑了多时刻回波的演变过程、回波的移向和移速以及回波在移动过程中的形变,都具有非线性特征,与实际情况更为符合。本文利用集合的线性交叉相关方法和最小二乘拟合方法分别对两次暴雨过程进行回波外推,通过比较发现,利用集合交叉相关外推方法所推得的回波分布与实况更为接近;利用Z-I关系,进一步分析了基于集合交叉相关外推方法的降水评分。与最小二乘拟合方法外推相比,集合交叉相关外推技术所得的回波移动位置与实况较为一致,且1小时累积降水预报效果也较好,因此集合交叉相关外推技术在临近预报中具有一定的应用价值。

This study constructed a method of ensemble linear cross-correlation and conducted Doppler radar echo extrapolation.The linear cross-correlation algorithm can obtain the echo distribution based on the previous two echo times.As a result,the ensemble linear cross-correlation technique is more close to the observed and has been taken into account the nonlinear characteristics,which involve the evolution,moving speed and direction,and deformation during the movement of multi-time echoes.According to the methods of ensemble linear cross-correlation and least squares fitting,the echo extrapolation of two rainstorm processes has been carried out.By comparison,the echo distribution derived from the ensemble cross-correlation is more realistic.Further analysis of precipitation evaluation indicates that the echo movement derived by ensemble linear cross-correlation is more similar to the observed,and gets better 1 h accumulated precipitation forecast.Therefore,the ensemble cross-correlation extrapolation technique has certain practical value in nowcasting.