摘要
高分辨率数值模式在台风预报业务中面临计算量、预报时效性、预报技巧的平衡难题。为了在能保证台风预报技巧前提下节约计算机资源、满足高时效性业务要求,将一种自适应的台风模拟区域客观计算方法与处理同时存在多个台风的技术相结合,提出台风有限区域自适应设置方法。将该方法应用于高分辨率CMA-MESO模式建立了一个高分辨率台风数值预报系统(HRTYM),选取2019年的典型台风利奇马及2020、2021年的16个重点台风过程开展试验与对比分析。台风利奇马的试验结果显示,与9 km分辨率的台风业务模式(CMA-TYM)相比,3 km分辨率的HRTYM需要的计算机资源更少,台风路径预报误差减少20.7%~61.0%,暴雨和大暴雨24 h地面降水TS评分明显提高。2020、2021年的批量试验结果显示HRTYM的台风路径预报技巧在27 h后明显优于CMA-TYM。
The development of refined tropical cyclone(TC)forecasting operations relies on high-resolution regional TC numerical forecasting.However,when faced with both the characteristic of expensive computational consumption of high-resolution models and the characteristic of the wide range of TC forecasting responsibility areas,it is difficult for the existing numerical forecasting techniques to meet the high timeliness requirements in operational high-resolution TC forecasting.High-resolution numerical models face challenges in balancing computational demands,forecast timeliness,and prediction accuracy for TC forecasting operations.To save computational resources and meet high timeliness requirements while ensuring TC forecast accuracy,this study proposes an adaptive setting-up approach of a limited area for the short-term forecast of TC by combining an adaptive objective calculation method for the TC simulation domain with multi-TC coexistence processing technology.This approach dynamically adjusts the position and size of the high-resolution TC model simulation domain based on the forecast timeliness and the actual situation of the TC.By automatically providing a scientifically reasonable simulation domain,the computer resource consumption is reduced without losing forecasting skills.Applying this approach to the high-resolution CMA-MESO model,a High-Resolution Typhoon Numerical Prediction System(HRTYM)is established.The numerical experiments and comparative analysis are carried out using Typhoon Lekima(2019)and 16 major typhoon events in 2020 and 2021.The experimental results of Typhoon Lekima show that,compared with the 9 km-resolution operational model(CMA-TYM),the 3 km-resolution HRTYM requires fewer computational resources with model integration time reduced by 11.2%-17.5%and storage space reduced by 58.6%-66.7%,the TC track forecast error decreases by 20.7%-61.0%,and the TS scores for 24-hour rainstorm and heavy rainstorm predictions improve significantly.The results of batch experiments in 2020 show that after 24-48 hours of mode integration,the track error of the 3 km-resolution HRTYM decreases by 3.44-34.91 km compared to the 9 km-resolution CMA-TYM,and the results of batch experiments in 2021 show that after 27-48 hours of mode integration,the track error of the 3 km-resolution HRTYM decreases by 0.6-22.35 km compared to the 9 km-resolution CMA-TYM.The results of batch experiments in 2020 and 2021 demonstrate that HRTYM exhibits superior track prediction skill over CMA-TYM beyond 27 hours.The high-resolution TC numerical prediction system applying the adaptive setting-up approach of a limited area for the short-term forecast of TC proposed in this study effectively reduces the cost of computing resources and computer storage space resources and ensures the skill of TC numerical prediction at the same time.
作者
谭晓伟
TAN Xiaowei(CMA Earth System Modelling and Prediction Centre,Beijing 100081;Key Laboratory of Earth System Modelling and Prediction,CMA,Beijing 100081)
出处
《气象科技》
2025年第4期497-507,共11页
Meteorological Science and Technology
关键词
数值预报
台风
高分辨率
numerical prediction
tropical cyclone
high resolution
