EXPERIMENTAL INVESTIGATIONS ON CHANNELIZATION OF RIVER KOSI FROM CHATRA TO NIRMALI

Abstract

The Kosi River is one of the most ancient rivers of India. It is also known as Kaushiki in Sanskrit literature. It is a perennial river. Its three main tributaries, viz. river Sun Kosi from West, river Arun from North and river Tamur from East, meet river Kosi at Tribeni to form the Sapt Kosi. River Arun, the longest of the tributaries, has cut through a gorge in the great Himalayan range and drains under the name of Phung Chu the entire Tibetan trough from Gosainthan to Kanchenjunga. Downstream of the confluence, the Sapt Kosi flows in a narrow gorge for a length of about 10 km and then debouches into the plains near Chatra. Further downstream, the river runs in a sandy, almost flat plain and flows southwards to join the Ganga near Kursela. Abundant supply of sediment from the Himalayas and low slope results in shifting of the river. The river Kosi was shifting towards west, but, it suddenly changed its course towards east in the year 1883 near Purnea. Mr. W. A. Inglis was deputed to make a reconnaissance survey of the area, who suggested to take suitable measure to control the river. After surveying, he came to the conclusion that it was not possible to interfere with the natural flow. He recommended maintaining proper records of changes in the river course and of its flood levels. Subsequently a conference was held in 1896-97 at Calcutta and the conclusion of that conference was that no steps were feasible for controlling the course of the big river with numerous channels having wide and shallow bed. So, to protect human being area, short length embankment was constructed, without adopting standard section. During flood, the embankment breached frequently, which damaged large area. Later in November 1937, conference of official and non-official was called by the newly formed congress government at Bihar to tackle the problem, but the conference could not arrive at any concrete conclusion. In 1940, Sir Claude Inglis visited that area and suggested to check the soil erosion in Nepal. In 1945, Lord Wavell visited the flood stricken area and in his address in the annual session of the Institution of Engineers (India) held at Calcutta, he suggested that the problem needs to be referred to newly formed Water Irrigation and Navigation Commission for advice. The Water Irrigation and Navigation Commission advised to construct a barrage at Chatra in Nepal along with canal system and 90 MW of hydro-electric power station on eastern canal, which is also known as Kosi project. In 1951, the dam site proposed by Kosi project at Barahakshetra was shifted downstream near Belka hills. But Central Water and Power Commission (CWPC) came to the conclusion that the Belka dam proposal would not be much useful and so, the proposal vi was dropped. Finally, in 1953 CWPC submitted proposal for construction of a barrage at Hannumannagar, 48 km downstream of Chatra to serve as a control point with a canal system and construction of embankments on both banks of the river to prevent flood and arrest westward shifting of the river course. The river Kosi carries about 187 million metric tonnes of sediment, which is coming from the Himalayas. About 70% of the sediment is settling between two embankments/levees and the balance is flowing to the river Ganga. During past two centuries, river Kosi has shifted laterally towards west. It has shifted by across a width of over 100 km. The river Kosi has been provided with levees along the banks, at a width ranging from 6 km to 16 km to curb further shifting. The general bed level at many places has risen by 0.1 to 3.0 m over a period of just 50 years. This is due to inadequate velocities in the river for carrying away the sediments. For minimizing the silting and lateral shifting, it is desirable to explore the possibility of channelizing the river Kosi by using various techniques, which is thought to be a possible method for enhancing the velocities in the river to boost its sediment carrying capability. The reach considered for the present study is about 88 km long from Chatra to Nirmali, where the deposition of sediment is the highest and the embankments have breached about 22 times since their completion in 1958. The most recent breach causing colossal destruction took place on 18.08.2008 near village Kushaha, Nepal. In view of the above, the following objectives have been envisaged for the present investigations, which are interrelated with a bearing on the proposed experimental model studies. i. Modelling of stage-discharge and sediment-discharge rating at selected river gauging sites; ii. Morphological study (plan form changes and cross-sectional changes) in the reach under consideration iii. Mathematical and physical model studies for investigating channelization of river through the use of: • Progressive series of T-shape groyenes along both the banks of river reach from Chatra to Kosi barrage; • Levees along both the banks of the river for a reach from Chatra to Nirmali; • Progressive series of hockey-stick shape spurs along both the banks of the river reach from Chatra to Kosi barrage; and vii • Regrading the bed of the river to suitable depth through excavation (Dr. S.V. Chitale’s recommendation). Rating curve based on regression analysis is used commonly by field engineers to estimate the stream flow as well as sediment load. The regression and curve fitting techniques are not adequate in view of the complexities involved in hydrological and fluvial processes. An inherent problem in the rating curve development using above techniques is the high degree of scatter, which may be reduced but not eliminated. To describe such processes, the applicability of Artificial Neural Network (ANN) to various problems of water resources has been amply demonstrated by different investigators. In the present study, ANN technique has been therefore used for modelling the stage-discharge and sediment-discharge rating relationships. The ANN model has been trained, validated and cross-validated for two important gauging sites, namely Kosi barrage and Barahkshetra (gorge area). Daily data of stage-discharge at the gauging site of Barahkshetra for 17 years (January 1949 to October 1966) has been used for the modelling. Out of this, nine years six month data are used for training, three years nine month for validation, and three years nine month for cross validation. Similarly, sediment-discharge at the gauging site of Kosi barrage for 9 years (May 2002 to January 2009) has been used for ANN modelling. Out of this, five years have been used for training, two years for validation, and two years for cross validation. The performance of ANN model has also been compared with the conventional regression technique. It is found that discharge as well as sediment concentration estimates at gauging sites obtained by ANN technique are more realistic than the sediment rating curves. Corrected data obtained by ANN technique for sediment-discharge at Kosi barrage gauging site have been transferred to Chatra by using SCS modelling techniques for further model studies. The study reach of river Kosi from Chatra to Nirmali (88 km) is also analysed to assess the channel morphological changes actuated by stream bank erosion process during recent times. Adopting Plan Form Index (PFI) developed by Nayan Sharma (2004), attempt has been made to assess the temporal and spatial variation of braiding intensities along the whole stretch from Chatra to Nirmali based on the remote sensing image processing. The basic data used for estimating Plan Form Index are derived from digital satellite images of Indian Remote Sensing Landsat5 sensor, comprising scenes for the years 1992, 2005, and 2011. The PFI values are seen to by and large decrease significantly with time, indicating increase in braiding viii intensities in majority of cross sections on both upstream and downstream sides of the Kosi barrage. Model studies on channelizing the river by providing T-shape groynes: Three-dimensional physical model of river Kosi, covering the reach from 40 km upstream to about 47 km downstream of Kosi Barrage, was constructed to a horizontal scale of 1/500 and a vertical scale of 1/70. The velocity and discharge scales derived from these scales are 1/8.36 and 1/292831, respectively. The various river structures, viz. Kosi barrage, proposed Dagmara barrage, Rail & Road bridge, Eastern embankment, Ring bund, afflux bund and spurs are reproduced in the model. The bed configuration of the river including deep channels, shoals and spill portion are reproduced as per the survey data of post-flood 2002. T-shape groynes are provided in the model to channelize the river in the following four ways. i) T-shape groynes of wing length of 2000 m were first provided with a spacing of 4000 m on both the banks to have a waterway of 3000 m. Using the dominant discharge of 15586 m3 ii) Next, the waterway in the model was reduced to 2000 m. In this case, though the flow was seen to touch the wings of the groynes, the velocity was inadequate to initiate the sediment motion. /s in the model, the flow of water was not seen to touch the wings of the T- shape groynes in most cases. iii) To increase the velocity, the waterway was further reduced to 1100 m. In this case, the flow velocity was adequate to wash out shoal and a regime channel was established. iv) To examine the possibility for optimizing the above, another trial was made by increasing the spacing between T-shape groyenes from 4000 m to 8000 m, but it was not found effective. Model studies on channelizing the river by providing levees: Although flow in natural rivers is three-dimensional (3D), however, one dimensional (1D) or two dimensional (2D) mathematical models are often used in engineering practice for shallow open channel flow. In particular, when simulating a very long river of multichannel in cross-section for a long period, 1D models (e.g., HEC-RAS) are more cost effective than 2D or 3D models, however, they are unable to simulate momentum exchange between main channel and floodplain, turbulence around engineering structures (e.g. bridges piers, spur dykes), and flow in highly sinuous channels (Duan et al., 2001; Duan and Julien 2005). To overcome these limitations, engineers commonly enhance 1D model with empirical formulas to approximate energy losses ix attributable to in-stream structures or meandering bends. The site of the present study is the Kosi River, a reach consisting of braided, transitional, and meandering channels with levees and dykes, and highly overloaded sediment supply, which requires an improved 1D model to simulate the hydraulic aspects of channelizing the river. Due to very high annual sediment to the tune of 187 million metric tonne carried by the river, 70% of sediment settles between the embankments, it is desirable to explore the possibility of channelization, adopting same bed profile. With this objective in view, mathematical model studies were first conducted for channelization of river Kosi for a reach from Chatra to Nirmali by providing levees. This study was based on 1D mathematical model HEC-RAS 4.1 version to get an idea about hydraulic aspects of channelizing the river. Various channelization widths ranging from 800 m to 2000 m were examined adopting the same bed profile, which showed that the width of 1100 m would minimize the aggradations and degradation of the river for a dominant discharge of 15586 m3 The finding of mathematical model for waterway of 1100 m with levees along both the banks was also studied on the physical model; in the first phase for the reach from Chatra to Kosi barrage, which was extended upto Nirmali in the second phase. The physical model studies also confirmed the efficacy of channelization with 1100m water way over the complete reach from Chatra to Nirmali. However, in the first phase of the studies restricted upto the barrage, some deposition was observed on the downstream due to decrease in velocity as a result of sudden widening of the river. But, this deposition was not seen to have any adverse effect on the river hydraulics upstream of the barrage. Only the morphology of the river on the downstream was seen to be affected in this case. In the second phase, average velocity of 3 m/s was attained over the entire reach, and it was observed to be in regime similar to that in the mathematical model study. /s. Physical model studies were also conducted using levees to have 1100 m waterway from Chatra to Nirmali, which showed similar satisfactory performance. Model study on channelizing the river by providing hockey-stick shape spurs: Hockey-stick shape spurs were provided with a spacing of 1500 m on both the banks to achieve a waterway of 1100 m. For the dominant discharge of 15586 m3/s, it was seen that the hockey-stick shape spurs guide the flow of water with an average velocity of 3 m/s and form bed regime. x Mathematical model to study Channelization by re-grading the river bed through excavation (S. V. Chitale’s recommendation):. S.V. Chitale suggested that by re-grading the stream bed by 0.44 m, velocity required to transport the sediment could be generated. In view of this, mathematical model studies were conducted for a reach from Chatra to Kosi barrage using the data of River cross-section post 2002 flood and sediment-discharge relationship as upstream boundary condition and normal depth as a downstream condition. The required river cross-sectional data were available in the month of May 2002. The cross sections of the river within the embankment were at a regular interval of 1.0 km, from 41 km upstream of Kosi barrage. Data pertaining to sediment and discharges recorded at the Kosi barrage were used for this study. Mathematical model (HEC-RAS) was run with the input data as a discharge variation of nine year hydrograph. The mathematical model study indicated that the velocity of flow along the channel increases and the bed level further deepens due to transportation of sediment by increased velocity, reducing shifting behaviour of river. Detailed mathematical and physical model investigations have been carried out for channelization of river Kosi considering various alternatives for this purpose. The results of the investigations have brought out that the following three measures are equally effective for achieving channelization on technical grounds. (i) Providing T-shape groynes along both the banks (ii) Providing levees to have a constricted waterway of 1100 m (iii) Providing hockey stick shape spurs along both the banks