Dynamic neural networks for real-time water level predictions of sewerage systems-covering gauged and ungauged sites

doi:10.5194/hess-14-1309-2010

Abstract

Abstract. In this research, we propose recurrent neural networks (RNNs) to build a relationship between rainfalls and water level patterns of an urban sewerage system based on historical torrential rain/storm events. The RNN allows signals to propagate in both forward and backward directions, which offers the network dynamic memories. Besides, the information at the current time-step with a feedback operation can yield a time-delay unit that provides internal input information at the next time-step to effectively deal with time-varying systems. The RNN is implemented at both gauged and ungauged sites for 5-, 10-, 15-, and 20-min-ahead water level predictions. The results show that the RNN is capable of learning the nonlinear sewerage system and producing satisfactory predictions at the gauged sites. Concerning the ungauged sites, there are no historical data of water level to support prediction. In order to overcome such problem, a set of synthetic data, generated from a storm water management model (SWMM) under cautious verification process of applicability based on the data from nearby gauging stations, are introduced as the learning target to the training procedure of the RNN and moreover evaluating the performance of the RNN at the ungauged sites. The results demonstrate that the potential role of the SWMM coupled with nearby rainfall and water level information can be of great use in enhancing the capability of the RNN at the ungauged sites. Hence we can conclude that the RNN is an effective and suitable model for successfully predicting the water levels at both gauged and ungauged sites in urban sewerage systems.

Highlights

The growth of urbanization has paralleled the growth of the global economy over the last decades
It should be noticed that the learning target values to the recurrent neural networks (RNNs) to estimate the fitness to the hydrological models in hydrois replaced by the synthetic data, assumed as the observa- logical applications, and to facilitate the comparition values, of water levels generated from the storm water management model (SWMM) at the son of different estimated/predicted results
The related outputs can provide important prior knowledge and/or information for the successful operation of pumping stations in an urban sewerage system, and the accurate water level predictions should offer an improvement in preventing a city from inundation

Summary

Introduction

The growth of urbanization has paralleled the growth of the global economy over the last decades. Owing to the rapid development of metropolitan areas, the natural hydrological mechanisms have been changed, such as the reduction of the infiltration and the concentration response time in a catchment, which leads to unexpected inundations and failures in operating pumping stations. Taiwan is located in the northwestern Pacific Ocean where the activities of subtropical air currents happen frequently. Due to the irregular timing and location of precipitations and the increase of impervious areas, the flood hydrographs during typhoon seasons result in large peak flows with fast-rising limbs which usually cause serious disasters in Taiwan. On 17 September 2001, Typhoon Nari struck northern Taiwan accompanied with heavy rainfalls, more than 500 mm/day, which caused 27 deaths. The flood inundated 4151 building basements that brought on countless economic losses

Methods

Results

Conclusion