首页|期刊导航|中国水利|洪水风险图推演实时性提升技术研究与实践应用

洪水风险图推演实时性提升技术研究与实践应用OA

Research and practical application of techniques for improving real-time performance of flood risk map deduction

中文摘要英文摘要

2024-2025年全国洪水风险图编制工作新增实时动态推演能力建设要求,旨在提升水旱灾害防御"四预"(预报、预警、预演、预案)系统应用的实战支撑能力.针对传统洪水风险图推演串行处理模式存在的响应速度慢、实时性差,以及难以适配防洪调度决策及时、高效需求的问题,本文开展洪水风险图推演实时性提升技术研究.基于洪水风险图实时动态推演平台研发实践,系统梳理洪水风险图推演全业务流程,识别方案加载、模型计算、数据存取、可视化渲染等各环节的性能瓶颈.融合高性能计算、高效数据存取、服务器后台渲染等关键技术,创新构建计算流、数据存取流、渲染流的并行处理架构,突破传统全量计算完成后再渲染展示的技术模式.该优化技术方案改善了系统平台的交互体验与实践应用效果,已在水利部、太湖流域及安徽省等洪水风险图实时动态推演平台中得到应用.本文以安徽省蒙洼蓄洪区25万个网格的推演场景开展实例研究,结果显示,相较于传统模式15 min的洪水分析成果呈现等待时长,该架构可实现20 s内输出首帧推演可视化结果,后续每帧更新仅需2 s,同时支持推演过程随时暂停或终止计算,有效提升防洪决策支撑的灵活性与时效性,可为全国其他省份、流域的洪水风险图实时动态推演平台建设提供技术参考.

The national flood risk map compilation project for 2024-2025 has put forward new requirements for the development of real-time dynamic deduction capabilities,aiming to enhance the practical support capacity of the forecasting,warning,rehearsal,and contingency planning applications for flood and drought disaster prevention.Traditional serial processing modes for flood risk deduction suffer from slow response speed and poor real-time performance,which fail to meet the timely and efficient demands of flood control scheduling and decision-making.This paper conducted technical research on improving the real-time performance of flood risk map deduction.Based on the research and development practice of a real-time dynamic deduction platform for flood risk maps,this paper systematically sorted out the entire business process of flood risk deduction and identified performance bottlenecks in key links,including scheme loading,model calculation,data access,and visual rendering.By integrating key technologies such as high-performance computing,efficient data access,and server-side background rendering,this paper innovatively constructed a parallel processing architecture consisting of computation flow,data access flow,and rendering flow,which broke the traditional technical mode that performs rendering only after the completion of full model calculation.The optimized technical solution has improved the interactive experience and practical application effectiveness of the system platform and has been applied in real-time dynamic flood risk map deduction platforms of the Ministry of Water Resources,the Taihu Lake basin,and Anhui Province.By taking the 250,000-grid deduction scenario of the Mengwa Flood Detention Area as a case study,the results show that compared with the 15-minute waiting time for presenting flood analysis results in the traditional mode,the proposed architecture can output the first frame of deduction visualization results within 20 seconds,with each subsequent frame updated in only 2 seconds.It supports arbitrary suspension and termination of the deduction process,effectively improving the flexibility and timeliness of flood control decision support,providing a technical reference for the construction of real-time dynamic deduction platforms for flood risk maps in other provinces and basins nationwide.

陈胜;柴福鑫;刘业森;刘舒

中国水利水电科学研究院,100038,北京||水利部防洪抗旱减灾技术创新中心,100038,北京中国水利水电科学研究院,100038,北京||水利部防洪抗旱减灾技术创新中心,100038,北京中国水利水电科学研究院,100038,北京||水利部防洪抗旱减灾技术创新中心,100038,北京中国水利水电科学研究院,100038,北京||水利部防洪抗旱减灾技术创新中心,100038,北京

建筑与水利

洪水风险图实时性动态性数据缓存后台图像渲染动态推演平台并行计算架构

flood risk mapreal-time performancedynamic characteristicdata cachingbackground image renderingdynamic deduction platformparallel computing architecture

《中国水利》 2026 (11)

24-30,7

国家重点研发计划项目"蓄滞洪区立体感知与运用智能决策关键技术"(2024YFC3012300).

10.3969/j.issn.1000-1123.2026.11.003

评论