A HYDROMETRIC DATA-BASED FLOOD FORECASTING MODEL USING A SIMPLIFIED ROUTING TECHNIQUE

Abstract

Adverse effects are experienced when high river flows occur in the form of floods causing loss of life and damage to property which have to be mitigated by employing economically feasible structural measures such as levees, flood walls and channel improvement. However, these types of measures cannot eliminate completely the hydraulic risk, given the impossibility of building larger and larger structures to cope with extremely low probability events. Therefore, an important role remains for non-structural measures to be compared, evaluated and implemented in real-time. Flood forecasting is an important non-structural measure for flood damage reduction and for minimizing flood-related deaths and, hence, its implementation as an effective tool requires accurate forecasting with sufficient lead-time. Therefore, it is essential that flood forecasting methods should be physically based, less data intensive and, over and above, should be easily understood by the field engineers for troubleshooting of the problems related to these methods during real-time operation. Typically, the flood forecasting models have two components: The deterministic flow component and the stochastic flow component. While the former is determined by the hydrologic/hydraulic model, the latter is determined based on the error series of the difference between the forecasted flow for a specified lead-time and the corresponding observed one. The residual series reflects both the model error, due to the inability of the deterministic model to correctly reproduce the flow process, and the observational error while measuring the flow. It is imperative, therefore, to use an appropriate model to reduce the model error. The hydrometric databased flood forecasting model studied herein is employed for forecasting flood for a given lead-time at a gauging station knowing the evolving flood hydrograph at an upstream gauging station without involving rainfall, the causative factor for runoff generation. Accordingly, the deterministic model employed herein is the river routing method. The emphasis of this study is on the development of a routing procedure for the application of a Variable Parameter Muskingum method, known as the Variable Parameter McCarthy-Muskingum Discharge-routing (VPMMD) method which has been directly derived from the Saint-Venant equations by Price and Perumal, [2011], for the purpose of real-time flood forecasting in natural rivers under data deficient conditions, especially the morphometric data. It is considered that the morphometric data of the river reach required for the study is available only at the river gauging stations, where also the rating curves are available. The proposed routing procedure using the VPMMD method envisages the development of a reach-averaged rating curve for the river reach using the rating curves available at the upstream and downstream ends of the study reach and the development of the reach averaged cross-sectional geometrical elements information using the cross-sections data available at both the reach ends. The parameters of the VPMMD method required for channel routing are estimated based on these reachaveraged rating curve and channel cross-section information. The routing parameters of the VPMMD method are linked to the channel and flow characteristics which enable the variation of these parameters at every routing time step. The routing procedure of the VPMMD method employs the aforesaid reach averaged rating curve and channel crosssection data supplied in the form of look-up tables for linking uniquely the normal flow depth with the flow characteristics such as the normal discharge, the normal velocity and the normal celerity, and with the geometrical elements such as the area and top width of the reach-averaged flow section for determining the variable parameters of the routing method. This routing procedure enables the routing of floods in hypothetical channels as well as in natural rivers, covering the main channels as well as the floodplains. This n method is studied herein for its strengths and weaknesses by routing hypothetical inflow hydrographs in two synthetic channels: 1) a natural river section look alike artificial uniform channel as employed by Price [2009], and 2) a uniform compound channel section reach consisting of main and floodplain trapezoidal section. While ten channel types based on the former artificial sections were used for conducting numerical experiments of the proposed routing approach, it was tested in 72 artificial channels of the latter type. A number of hypothetical inflow hydrographs were routed through these artificial channels using the VPMMD method based routing procedure described herein and the solutions obtained were compared with the corresponding benchmark solutions of the Saint-Venant equations. Successful simulations of the benchmark solutions using the routing procedure formulated herein based on the VPMMD method demonstrate the theoretical correctness of the VPMMD method as well as that of the suggested routing procedure. These simulations also verified the ability of the VPMMD method to estimate the stage hydrographs by closely reproducing the corresponding benchmark stage hydrographs obtained from the solutions of the Saint-Venant equations. In addition to the verification of the VPMMD method routing capability, the utility of the method for field applications was also investigated by simulatingten past recorded flood events of a 15 km reach between Pierantonia and Ponte Felcino stations of Tiber River in Central Italy. All the ten events, except one could be reproduced with the Nash-Sutcliffe efficiency tj >99%, thus, demonstrating the immense usefulness of the method for routing floods in river reaches. It was considered appropriate to investigate the applicability limits of the VPMMD routing method to bring out the practical limitations of the method. This was carried out by simulating 11200 hypothetical routing solutions based on the Saint-Venant equations and reproducing these 11200 benchmark solutions using the VPMMD method. This study iii reveals that the VPMMD method is able to produce 95% of successful simulations of the discharge hydrograph solutions with 5% error in reproducing the pertinent characteristics of the benchmark solutions. Based on the applicability limit estimation study, the recommended criterion limit to be satisfied by the discharge hydrograph at the inlet of the reach is (l/SA(dy/dx) <0.57, and for similar successful simulation of the benchmark stage hydrographs only, the criterion limit to be satisfied is (l/S0)(dy/dx) <0.61. These range of applicability limits of the VPMMD method brings out the immense practical usefulness of the VPMMD method. The reach-averaged channel flow and the cross-sectional information required for the estimation of the routing parameters of the VPMMD method was supplied in the tabular form by relating the flow depth uniquely with the discharge, velocity and the celerity, and the top width of the flow section. No inbetween channel section information was used in the developed routing procedure which enables the channel routing between the upstream inflow section and the downstream outflow section, and enabling routing through the main and floodplain sections of the channel reach. Considering the practical usefulness of this routing procedure developed using the VPMMD method, a Variable Parameter McCarthy-Muskingum Discharge Real time Flood-Forecasting (VPMMDRF) method is developed using the VPMMD method as a component model of a hydrometric data-based deterministic forecasting model for real time flood forecasting, particularly considering routing through multiple sub reaches of a river reach. A two parameter autoregressive forecast error estimation model forms the other component of the VPMMDRF method. Extensive investigations were made to verify the suitability of the VPMMD method for real-time forecasting applications. Unlike, the simulation mode of routing, the routing is done by marching in time after routing along the entire routing reach for the current inflow discharge. This way of routing procedure is desirable especially for real-time flood operations in the river IV reaches. In order to verify this method, an application study was conducted for the Pierantonio (upstream) and Ponte Felcino (downstream) reach of the Tiber River in Central Italy by studying 10 recent flood events in forecasting mode. The forecasting results were arrived at by considering the 15 km long Pierantonio and Ponte Felcino reach, first as a single reach, and secondly considering as 2 sub-reaches (each of 7.5 km). For all these forecasting experiments, the varied forecasting lead times such as 1.00 h, 1.50 h, 2.00 h, 2.50 h, and 3.00 h were used. The performance evaluation of the proposed model is carried out in conjunction with an error forecasting model developed based on a simple Autoregressive (AR) model. From all the investigation results obtained from this forecasting study, it is found that the model produces accurate forecasting results along with the corresponding stage estimates for a lead time up to 3.00 h, with the warm up period considered for developing the error forecast AR model being 5.00 h. Therefore, this newly proposed VPMMDRF model can be conveniently used for discharge forecasting up to 3.00 h lead time in the considered Pierantonio-Ponte Felcino reach.