eduzhai > Applied Sciences > Engineering >

A deep convolutional neural network model for rapid prediction of fluvial flood inundation

  • Save

... pages left unread,continue reading

Document pages: 45 pages

Abstract: Most of the two-dimensional (2D) hydraulic hydrodynamic models are stillcomputationally too demanding for real-time applications. In this paper, aninnovative modelling approach based on a deep convolutional neural network(CNN) method is presented for rapid prediction of fluvial flood inundation. TheCNN model is trained using outputs from a 2D hydraulic model (i.e. LISFLOOD-FP)to predict water depths. The pre-trained model is then applied to simulate theJanuary 2005 and December 2015 floods in Carlisle, UK. The CNN predictions arecompared favourably with the outputs produced by LISFLOOD-FP. The performanceof the CNN model is further confirmed by benchmarking against a support vectorregression (SVR) method. The results show that the CNN model outperforms SVR bya large margin. The CNN model is highly accurate in capturing flooded cells asindicated by several quantitative assessment matrices. The estimated error forreproducing maximum flood depth is 0 ~ 0.2 meters for the 2005 event and 0 ~0.5 meters for the 2015 event at over 99 of the cells covering thecomputational domain. The proposed CNN method offers great potential forreal-time flood modelling forecasting considering its simplicity, superiorperformance and computational efficiency.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×