eduzhai > Applied Sciences > Engineering >

A Hybrid Deep Learning Model for Predictive Flood Warning and Situation Awareness using Channel Network Sensors Data

  • Save

... pages left unread,continue reading

Document pages: 21 pages

Abstract: The objective of this study is to create and test a hybrid deep learningmodel, FastGRNN-FCN (Fast, Accurate, Stable and Tiny Gated Recurrent NeuralNetwork-Fully Convolutional Network), for urban flood prediction and situationawareness using channel network sensors data. The study used Harris County,Texas as the testbed, and obtained channel sensor data from three historicalflood events (e.g., 2016 Tax Day Flood, 2016 Memorial Day flood, and 2017Hurricane Harvey Flood) for training and validating the hybrid deep learningmodel. The flood data are divided into a multivariate time series and used asthe model input. Each input comprises nine variables, including information ofthe studied channel sensor and its predecessor and successor sensors in thechannel network. Precision-recall curve and F-measure are used to identify theoptimal set of model parameters. The optimal model with a weight of 1 and acritical threshold of 0.59 are obtained through one hundred iterations based onexamining different weights and thresholds. The test accuracy and F-measureeventually reach 97.8 and 0.792, respectively. The model is then tested inpredicting the 2019 Imelda flood in Houston and the results show an excellentmatch with the empirical flood. The results show that the model enablesaccurate prediction of the spatial-temporal flood propagation and recession andprovides emergency response officials with a predictive flood warning tool forprioritizing the flood response and resource allocation strategies.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×