台风风暴潮异模式集合数值预报技术研究及应用

台风风暴潮数值预报的准确性在很大程度上取决于台风路径预报和强度预报的精度以及风暴潮预报模型的计算精度。目前,国际上24/48 h台风路径预报平均误差分别约为120/210 km左右[1],对于走向异常的台风误差更大;更有,根据单一的台风路径和单族的风暴潮数值预报模式并不能保证获得可靠的风暴潮预报结果。考虑多重网格法原理具有在疏密不同的网格层上进行迭代以达到平滑不同频率的误差分量,使得计算快速收敛,精度提高的特性。在前期研究基础上基于业务化高分辨率(结构网格/有限差分算法)和精细化(非结构网格/有限元算法)台风风暴潮集合数值预报模型构建多模型台风风暴潮集合数值预报系统。采用"非同族"模型进行集合预报很大程度上降低了误差相似遗传的可能性。应用该方法对典型台风风暴潮过程进行了试应用,试报结果表明:该方法对风暴潮增、减水预报效果高于单一集合预报,具有一定的应用前景。

Research and applications on multi-model ensemble numerical typhoon surge forecast technology

The accuracy of typhoon surge numerical forecasting results mainly depends on the precision of forecasted Typhoon tracks and intensity,but also the precision of storm surge forecast model. Currently, the errors for 24/48 h typhoon forecasted tracks are about 120/210 km.The errors were large for some typhoons with abnormal tracks. The reliability of storm surge forecasting results can not be guaranteed relying solely on single Typhoon track and single-class numerical Typhoon surge forecast models. Multi-grids method can smooth different frequency deviation components in order to accelerate iteration convergence and improve precision by iterative computations among coarse grid and finer grid. A multi-model ensemble numerical typhoon surge forecast system was implemented based on high resolution and refine storm surge ensemble numerical forecast models in this paper. In do so, it can reduce genetic probability for single-class model generating similar errors. The system was valid by testing with typical typhoon surge events which impact China coasts. The results indicate that this technology improves the effects of surge forecast effectively than using single ensemble model, and may be used in the future.