广州城市暴雨内涝模型模拟效果评估

EVALUATION OF SIMULATION EFFECT OF URBAN RAINSTORM WATERLOGGING MODEL IN GUANGZHOU

关键词:  广州  暴雨内涝模型  内涝点  积水深度  风险等级

Key words:  Guangzhou  rainstorm waterlogging model  waterlogging point  ponding depth  risk level

摘要:

基于水动力方法构建的广州城市暴雨内涝模型，结合精细化降水预报，对内涝点的积水深度及风险等级进行模拟，结果表明，模型对近两年内涝点内涝风险等级命中率达65%，对总降水量为50～100 mm的降水过程命中率最高为72.8%。模型对2020年“5.22”特大暴雨过程模拟的积水深度和实况相比偏弱，误差主要分布在30 cm以内，大约占64%，大部分模型模拟积水深度偏小，主要位于广州中北部地区；此外，模型对积水１ｍ以下内涝点的积水有不错的模拟能力，而对２ｍ左右的深积水模拟能力还有限。不同重现期雨情下，广州中心城区的降雨量和历时越大，积水面积越大。1 h重现期雨情下，积水深度一般在20 cm以下，部分在20～59 cm；3?h降雨情景下，积水明显加深，积水深度一般在20～59 cm，部分在0～20 cm和60～119 cm。总体而言，模型模拟结果与实测内涝积水情况基本一致，模型准确度可满足业务需求。

Abstract:

In the present study, a Guangzhou urban rainstorm waterlogging model which is developed using the hydrodynamic method, combined with refined precipitation forecast data, is used to simulate the ponding depth and risk level of waterlogging points in Guangzhou. The results demonstrate that the model has a hit rate of 65% for the waterlogging risk level of the waterlogging points in the past two years, while the highest hit rate is 72.8% for the precipitation processes with accumulated precipitation of 50～100 mm. The simulated ponding depth for the precipitation event on May 22 is relatively shallower compared with the real situation, and the error is mainly found in the simulation for the ponding depth within 30 cm, accounting for about 64%. Most of the simulated ponding depth is shallower and these ponding points are mainly located in the north-central part of Guangzhou. The model has good simulation ability for ponding with depth below 1 m, while the simulation ability for deep ponding around 2 m is still limited. Moreover, the waterlogging conditions are different in the central urban area under different return periods. Under the 1-hour rain condition, the ponding depth is generally less than 20 cm, and some range within 20～59 cm; under the 3-hour rain condition, the ponding depth is generally between 20～59 cm, with some ranging between 0～20 cm and 60～119 cm. Overall, the simulation results roughly match the actual waterlogging situation, and the model accuracy can meet general operational needs.