Research of urban river level forecast method based on InfoWorks ICM and LSTM
[Objective]The prediction of urban river water level is of great significance for the management of urban waterlogging risk.However,the urban water system in coastal areas is complex,and traditional numerical simulation models have low compu-tational efficiency and cannot achieve real-time calculation.[Methods]In response to the above issues,this article uses the water culture and hydrodynamic model data constructed by the Urban Integrated Basin Drainage Model(InfoWorks ICM)as a data-driven approach.Taking into account rainfall,urban surface elevation(DEM),land use,street distribution,and drainage net-work layout,an LSTM urban river water level prediction neural network model based on machine learning method is constructed.Taking the Jin'an River to Guangminggang Basin in Fuzhou City as an example,a case study is conducted.[Results]The results show that the average Nash efficiency coefficient(MNSE)of the model for predicting urban river water level during the 48 hour foresight period is above 0.7,the prediction accuracy reaches level B,and the error of peak water level prediction is less than 3％.[Conclusion]The model can provide reliable river water level evolution process and peak water level prediction result,indi-cating that the constructed model has good predictive performance and can be used for rapid water level prediction in urban river networks.
InfoWorks ICMLSTMflood forecasturban river networkurban waterloggingjoint scheduling of multiple gate pumpsFuzhou City
万方数据
维普
