Introduction

Geomorphic analysis of a river presents unique challenges and requires a systematic and organized approach in terms of system complexity involved. Progress in the study of fluvial geomorphology has been aimed to develop the capability to identify, investigate and understand the continuity and connectivity of the flow processes and fluvial landforms in a river system. This prescribes the need to recognize and explore the links that bind the fluvial system in space and time. Knowledge of the history and sequence of antecedent events and trends of morphologic evolution during the months, years and decades preceding the study will definitely help to understand the process–form relationships of a river. River engineers, policymakers and managers recognize the importance of accounting channel morphology and the dynamics of the fluvial system of alluvial rivers. Modern approaches to the river management require engineers to work with the natural process–form relationships of the river, by retaining as much as possible of the natural hydraulic geometry of the self-formed channel at the time of river regulation, channel training, navigation, flood defense and land drainage. It is essential to identify correctly the current morphological status of the river and to predict its future development with and without the proposed engineering interventions. The latter requires the knowledge of predicting the reaction of river channel, for example flow regulation (for hydropower or flood defense), or bank stabilization and training works (to control channel migration). Morphological impacts are assessed from the point of disturbance through process–form feedback mechanisms operating in the fluvial system. Channel migration perhaps is a very common phenomenon found profoundly in braided and meandering rivers over the earth’s surface characterized by wide and flat valley floors and erodible banks. In a general sense, channel migration is simply the movement of a river channel back and across its valley during a specific time span.

Riverbank erosion is an endemic and recurrent natural hazard. When rivers attain the mature stage, they become sluggish and form meander bends. These oscillations cause massive riverbank erosion (Rahman 2010; Das and Bhowmik 2013). Lateral migration is a process that can cause catastrophic local or regional changes (Hickin and Nanson 1984; Thakur et al. 2012). The comprehensive effect of such changes introduces a socioeconomic hazard to the floodplain dwellers in a respective river basin. A number of factors determine the lateral migration of a river along its pathways such as drainage basin area, topography, vegetation cover, tectonic activity, land use patterns and climate factors in a particular region. Erosion may be caused either by undercutting of the bank materials by channels. Such undercutting during catastrophic flood situation produces an overhanging cantilevered block that eventually fails or by over-steepening of bank materials due to the migration of the thalweg closer to the bank during the falling stages (Goswami 2002). Various studies have been carried out for some major rivers with the help of remote sensing and GIS technique for detecting spatiotemporal changes in river erosion (Nanson and Hickin 1986; Yang et al. 1999; Bhakal et al. 2005; Kotoky et al. 2005; Kummu et al. 2008; Thakur et al. 2012; Sarma and Acharjee 2012; Chakraborty and Datta 2013; Gogoi and Goswami 2013, Neog 2017). From the scientific point of view, the migration of a river channel with respect to space and time is an expression of response by a river channel to its natural and anthropogenic drivers. This response manifested the changes in the physical form of a river channel owing to the actions of various drivers. Such changes may incorporate 1D change (e.g., changes in width, depth and thalweg length) along with certain 2D changes (e.g., adjustments in channel planform) in the riverine morphology (Wallick et al. 2007).

Migration of a river channel comprehensively encompasses the temporal width–depth adjustments of a river as well as alterations in its planform geometry. The obvious act centering the study of channel migration is thus, unquestionably, the stipulated ‘timeframe’ over which a river channel is expected to move through a defined corridor. In fact, it is the timescale which extends the limits of channel migration. A period of 100 years is very often used as an appropriate timeframe (Bolton and Shellberg 2001). While considering the variables of such migration, Hickin and Nanson (1984) defined channel migration as the collective output of stream power, boundary resistance, bank height, meander bend radius and channel width. The dynamics of channel adjustment in terms of lateral migration can be lucidly illustrated through the application of modern tools of geoinformatics, i.e., the techniques of remote sensing and geographic information system (GIS) which are widely capable of detecting as well as representing the changes that took place over space and time. A meticulous endeavor has been made to assess the channel dynamics in 44 years and to unveil the nature of channel migration in terms of changes in the historical bank line, centerline positions and alterations in meander geometries through numeric and graphical methods coupled with field observations of the Raidak-I river.

Study area

The Raidak River is one of the main right bank tributaries of the Brahmaputra River. It is a trans-boundary river and flows through countries like Bhutan, India and Bangladesh. The river rises in the Himalayas and is known as Thimphu Chhu in its upper reaches. It passes through various mountains and valleys in Bhutan. In its journey, it is joined by several small tributaries flowing from nearby mountains. The river is known as the Wong in its upper course in Bhutan. The river after traversing through the mountainous terrain in Bhutan comes back to the plains in India, into the plains in Alipurduar District and then flows through Cooch Behar District in the state of West Bengal. After flowing in the Indian subcontinent, the river enters Bangladesh. The Raidak River merges with the Brahmaputra River within the nation limits of Bangladesh. The confluence takes place at the chainage of 327 km.

Among the three course of river Raidak, the middle one is old Raidak, western flow is named as Raidak-I or Deepa Raidak and lastly, the eastern one is named as Raidak-II. The name of middle flow signifies that old Raidak is the ancient course of the Raidak River. The study area river Raidak-I is located in between two districts of West Bengal, which are Cooch Behar and New Alipurduar. The region is situated between 26° 34′ 30.18′′ to 26° 12′ 57.58′′ North latitude and 89° 43′ 12.12′′–89° 41′ 38′′ East longitude. The length of the river in the study area is 81.9 km and divided into part AB and BC (Fig. 1).

Fig. 1
figure 1

Location map of study river Raidak-I

Full size image

Methodology and database

Database preparation

The most important aspect of the present work is developing a morphology-independent methodology. The authors have identified channel migration by using satellite imagery with the help of remote sensing and GIS technique. GIS is the effective and accurate tool of quantifying channel changes both at medium-term and at short-term geomorphic scales (Winterbottom 2000). For identifying bank line migration, calculation of shape area migration and erosion and deposition of the study area, remote sensing techniques and satellite imageries have been used (Table 1). This work adopts to regularly use sophisticated software ArcGIS 10.1 and extensions to identify centerlines, centerline migration, shape area migration and calculation of erosion and accretion area.

Table 1 Characteristics of selected satellite images
Full size table

Delineation of channel centerline migration

The location of the riverbank in the six datasets was determined using the Arc GIS software and its extensions. The changes in the locations of both riverbanks were assessed during six periods, i.e., 1972–1978 (6 years), 1978–1990 (12 years), 1990–1996 (6 years), 1996–2002 (6 years), 2002–2009 (7 years) and 2009–2016 (7 years). For the determination of centerlines of the active river channels of different time periods, ArcGIS extension, namely ‘River Bathymetry toolkit,’ has been applied. The extension has been used in ArcGIS 10.1 to generate centerlines automatically from the active bank lines. These centerlines display the nature of lateral migration of the river. Another extension Channel Migration Toolbox (CMTB) was used in Arc GIS 10.1 to determine how to reach average channel migration, the average length of migration of channel and shape area migration of channel, which eventually helps to understand the lateral migrating nature of the river Raidak-I. The migration of the centerline was measured with the help of the migration area, which was defined as the area enclosed by successive channel centerlines (Fig. 2). The migration area of the centerline can be divided into the left and right parts according to the river flow direction by using the old centerline as the reference. If the left migration is larger than the right migration, then the river shifting direction is toward left, and vice versa.

Fig. 2
figure 2

Schematic diagram of centerline migration area

Full size image

Determination of erosion and accretion areas

Riverbank lines during different time periods have been superimposed, and through the transect generation, the river channels have been divided into several segments. The amount of erosion and accretion was calculated separately for each side of the river. Digitization of the active bank lines of each period of 1972 and 2016 superimposed into a single layer and processed them using the GIS software to create polygons that represented the difference between the first and second positions of different segments. If a polygon was positioned to the left of the 1972 bank line, it represented an erosion polygon; if a polygon was positioned to the right of the 1972 bank line, it represented a deposition polygon. The same process was adopted to assess erosion and deposition along the right bank. But following the changes of polygons to the left and right of the 1972 bank line represented deposition and erosion, respectively. Summing the areas of these polygons, the total eroded and deposited areas from 1972 to 2016 were assessed. After determining these areas, we added 1978, 1990, 1996, 2002 and 2009 bank lines sequentially to subdivide each of the polygons into smaller polygons that represented the bank lines of each period (Fig. 3). The GIS software was used to calculate the areas of overlap between the bank line in each year and the boundaries of the polygon that contained that portion of the bank line. The same approach we used for the period from 1972 to 2016 to calculate the erosion and deposition between consecutive years.

Fig. 3
figure 3

Schematic diagram of lateral erosion–accretion area

Full size image

Results and discussion

Graphic representations and illustrations have been generated based on computational derivations. The ArcGIS (Version 10.1) software and ArcGIS extension tools have been used to process and prepare the graphics outputs with regard to centerline migration, bank line movements, river meander dynamics and erosion–accretion and to delineate historical centerline position reach average centerline migration and shape area migration of channel.

Centerline migration

The centerline of the river channel was derived from the boundary based on the bank lines by using the mathematical morphology method which was an improved method to extract skeleton line of the polygon. The reach average length of channel migration and reach shape area migration of channel were measured by its centerline. The distance between two intersecting centerlines is considered a river reach. The centerline of the Raidak-I river from 1972 to 2016 is the historical migration zone (Fig. 4). Historical migration zone (HMZ) is the area of historical channel occupation recorded on photographic evidence.

Fig. 4
figure 4

Historical centerline positions of Raidak-I river (1972–2016)

Full size image

Zone-1

Zone-1 is showing centerlines of the period 1972–2016 in the historical migration zone. The centerline of 1972 is near to left bank, the centerline of 2009 is near the right bank, and other centerlines are located in between the two banks. Here, different centerline is displaying an oscillating nature.

Zone-2

Zone-2 clearly indicates the changing nature of centerlines of the Raidak–I River during 1972 to 2016. Within these time periods, it is evident that the meander bands have shifted toward a specific direction.

Zone-3 and Zone-4

The two zones are showing very well in the 1972 centerline on the right bank, and the centerline 2016 is on the left side. However, from 1972 to 2016, this change took place from right bank to left bank of the river, which means that this change is unidirectional.

The reach average length of 154.26 m, 100.13 m, 49.87 m, 75.60 m, 49.48 m and 30 m migration of the channel occurred during the periods of 1972–1978, 1978–1990, 1990–1996, 1996–2002, 2002–2009 and 2009–2016, respectively.

Channel migration

Following the criteria used in centerline migration, i.e., the methods of assessing reach average length of migration and shape area migration of channel, it was observed that in the past 44 years (1972–2016), the reach average length of migration and shape area migration of centerlines of Raidak-I river were 231.88 m and 10.53 km2, respectively.

It has been shown through various zones in 1972–2016, how the changes or migrate in different parts of the Raidak-I river and how to reach average channel migration would be calculated. Gray area is the reach average channel migration area of the river. The zones (zone-1, zone-2, zone-3 and zone-4), it show that the two centerlines intersect in each other and the area of division between them is the moving zone or migrated area, and this migration is calculated according to different river reach; it is called reach average channel migration.

Significant gradual decreasing trends of the average length of migration and shape area migration were observed during the period of 1972–2016 (Fig. 5). Compared with the earlier period (1990–1996), the average length of migration and shape area migration increased slightly during the period of 1996–2002. In this situation, the reach average length of migration decreased in periods (1972–1978, 1978–1990, 1990–1996, 1996–2002, 2002–2009, 2009–2016). During the periods of 1972–1978 and 1978–1990, the maximum reach average length of channel migration is displayed. But the study revealed that from 1972 to 2016, there is a decreasing tendency of reach average length of channel migration in Raidak-I river. The reach average shape area of migration in all the periods was 6.86 km2, 4.57 km2, 2.30 km2, 3.27 km2, 2.15 km2 and 1.31 km2 (Fig. 6a) which showed a decreasing trend.

Fig. 5
figure 5

Channel migration of Raidak-I river. The channel centerlines and how to reach average channel migration would be calculated

Full size image
Fig. 6
figure 6

a Average length of migration of Raidak-I river. b Shape area migration of Raidak-I river

Full size image

Reach average channel migration

On this map showed different high channel migration parts of the Raidak-I river in the different zones according to the different period. Zone-1: In this section of the river, it is clearly shown that channel migration has decreased during the entire period and the meandering of the river has increased over the time. Zone-2: This zone depicted that the channel migration is increasing from 1996 to 2016. Zone-3: This part of the river showed increasing bending or meandering and unidirectional shifting during the period 1972 to 2016 and an oxbow lake is formed. Zone-4: This part of the river created several oxbow lakes due to river high dynamicity and oscillating change. Zone-5: Sequentially change and unidirectional shifting scenarios of a meander bend are represented during the period 1972 to 2016 in this zone. An oxbow lake was created from 1972 to 1978.

One thing can be concluded from the above analysis of centerlines position of different periods of Raidak-I river that Raidak-I river is very dynamic. But some parts are not dynamic. However, the parts which are very dynamic are also very sequential in meandering as noticed in Fig. 7. The number of river reach or intersection of centerlines increased over time like 115 in 1972–1978, 105 in 1978–1990, 175 in 1990–1996, 126 in 1996–2002, 149 in 2002–2009 and 218 in 2009–2016.

Fig. 7
figure 7

Reach average channel migration of Raidak-I river

Full size image

Erosion and accretion

The activity of erosion–accretion processes of Raidak-I river has been evaluated for five successive periods of 1972–1978, 1978–1990, 1990–1996, 1996–2002, 2002–2009 and 2009–2016.

The erosion–accretion is prevalent at different segments during different time periods (Table 2). The intensity of erosion–accretion decreased through time since 1972. It is revealed that the Raidak-I river experienced a higher intensity of erosion along its left bank than the right bank and accretion is higher along the right bank than the left bank (Fig. 8). During the period 1972–2016, the total average annual erosion and accretion were 250,000 and 260,000 m2/year, respectively, signifying a net gain of sediment/land area (Fig. 9).

Table 2 Erosion and deposition of Raidak-I river during 1972 to 2016.
Full size table
Fig. 8
figure 8

Various segments of the Raidak-I river created by Transect

Full size image
Fig. 9
figure 9

Erosion and accretion of Raidak-I river during 1972 to 2016

Full size image

The period 1972–1978 showed phenomenal river bank erosion where the deposition is very less (Fig. 10). The right and left bank eroded areas are 1,220,000 and 1,717,000 m2, respectively, during the period 1972–1978. During this period, accreted areas in both right and left bank are 1,772,000 and 1,250,000 m2 (Table 2). The erosion and accretion are higher in this period compared to other periods. During this period, the erosion and accretion are higher in both right bank and left bank, but the sifting direction is toward the left side. During the period 1978 to 1990, right bank and left bank eroded areas are 943,000 and 1,103,000 m2, respectively. But, the accreted areas in both right bank and left banks are 1,056,000 and 979,000 m2, respectively. Both side erosion and accretion decreased in this period. During the period of 1990–1996, 583,000 m2 area eroded from the right bank and 663,000 m2 area eroded from the left bank which are the lowest of both side erosion and accretion of corresponding periods of Raidak-I river. The pattern of erosion and accretion is different during the period of 1996–2002 compared to earlier periods. In this period, right and left bank erosions were 979,000 and 1,070,000 m2, where depositions were 1,358,000 and 1,224,000 m2, respectively. Erosion and accretion of both sides of the river increased in this period. The study showed that the accretion in the right bank is less in comparison with the left bank in the period of 2002–2009. Such a trend at erosion and accretion continued in the next period (Table 2). Right bank deposition of this period is lower than the left bank accretion. A decreasing trend is seen continuously in the period 2009–2016. More specifically, during this period, right and left bank erosion and accretion are 49,000, 369,000, 491,000 and 434,000 m2, respectively (Figs. 11, 12, 13, 14 and 15).

Fig. 10
figure 10

Right and left bank erosion–accretion of Raidak-I river during 1972–1978

Full size image
Fig. 11
figure 11

Right and left bank erosion–accretion of Raidak-I river during 1978–1990

Full size image
Fig. 12
figure 12

Right and left bank erosion–accretion of Raidak-I river during 1990–1996

Full size image
Fig. 13
figure 13

Right and left bank erosion–accretion of Raidak-I river during 1996–2002

Full size image
Fig. 14
figure 14

Right and left bank erosion–accretion of Raidak-I river during 2002–2009

Full size image
Fig. 15
figure 15

Right and left bank erosion–accretion of Raidak-I river during 2009–2016

Full size image

From the year 1972, lateral migration of the river has decreased although both the side amount of erosion and accretion of the river have been increased. Thus, the river migrated laterally, but in the later phase the amount of erosion and accretion has been increased in both the banks; as a result, the amount of erosion in the left bank and accretion in the right bank and the amount of erosion in the right bank and accretion in the left bank balance each other (Table 3). Among the different segments, 03-05, 22–25 and 30–35 segments are very active for erosion and accretion during the study period 1972 to 2016.

Table 3 Right and left bank erosion–accretion of Raidak-I river during 1972–1978 and 2009–2016.
Full size table

Direction of shifting

Erosion and accretion over the river beds and transport of sediment are the common characteristics of the alluvial river channel. All these processes occur throughout a river system, but certain processes may dominate in particular areas. In case of an idealized fluvial system, erosion dominates in the upper reaches, whereas accretion dominates in the lower reaches. The sediment transport process dominates in the so-called transfer zone connecting the upper and lower extremes. The part of the fluvial system known as the transfer zone consists of meandering streams. The meandering part usually occurs in the lower reaches where slopes are less pronounced. The transfer zone approaches toward the stability condition as a result of the process of aggradation and degradation processes. In this situation, the amount of sediment entering the zone will be equal to that leaving the zone.

The Raidak-I river shifting is mostly oscillating, but some locations are unidirectional in nature (Fig. 16). In most of the locations, the river shifted to the opposite side of the concrete embankment. The length of the river is 81.9 km in the present study, out of which 7.95 km long is attributed to embankments (Fig. 17). Riverbank erosion is absent along the embankment. Zone-1, zone-2, zone-3, zone-4, zone-5, zone-6 and zone-7 are depicting the unidirectional shifting part of the Raidak-I river. This river meandering patterns are followed by neck cutoffs, chute cutoffs and meandering shifting. Zone-1 and zone-4 are experienced with neck cutoff. Neck cutoff usually occurs in a highly sinuous part of this river and are often initiated by floods. During this process, the long meandering path is by past. Zone-2, zone-5 and zone-6 are represented by some chute cutoffs. A chute cutoff is less drastic than a neck cutoff and tends to increase the radius of the bend. Zone-3 and zone-7 showed the normal trend of meander shifting of Raidak-I river. Migration by meander shifting usually occurs in the downstream direction. These movements can be quite drastic, and a shift of up to 288 m during 1996–2002 has been reported (Fig. 16).

Fig. 16
figure 16

Unidirectional shifting of Raidak-I river. The different windows (window 1, 2, 3, 4, 5, 6 and 7) display the unidirectional shifting

Full size image
Fig. 17
figure 17

Boulder-made embankment along the Raidak-I river

Full size image

The trend of lateral migration and erosion–accretion of the Raidak-I river demonstrated that the shifting direction has been changed through time (Table 4). Since 1972–1990, the river shifted to the left, but there has been no change since 1990–2002. But since 2002–2009, the river shifted toward the right and again there has been no change from 2009 to 2016. It is evident that lateral migration of the river occurred on both sides.

Table 4 Shifting direction of Raidak-I river at different time periods.
Full size table

Conclusion

From the above discussion, it can be concluded that the concerned Raidak-I river is very dynamic in nature throughout the study period. During the study period, the river shifted its course in both sides. From the year 1972, the amount of lateral migration of the river has been decreased although both side amount of erosion and accretion of the river have been increased. Though the river migrated laterally in earlier periods, in the later phase the amount of erosion and accretion has been increased in both the banks; as a result, the amount of erosion in the left bank and accretion in the right bank and the amount of erosion in the right bank and accretion in the left bank balance each other. Thus, the river now becomes less vulnerable in terms of lateral shifting and presenting steady phase. According to different historical evidence and results of the study about this river, both bank lines are stabilized with the time but within the bank line, the oscillations of concave and convex slopes increased. The present work deals with the channel shifting of the river only and does not seek the causes of that shifting although it can be said that shifting of the Raidak-I river may be the result of construction of embankment along the banks of the river and decreased discharge of the river.