1 Introduction

River confluences are the nodal points in the river system where main river channels are amalgamated, which in turn, may employ a significant control over the river migration and morphology (Richards 1980; Best and Rhoads 2008). Bangladesh is known as a low-lying riverine country with mighty Ganges–Brahmaputra–Meghna (GBM) major river system. Land erosion–accretion and channel migrations are very common phenomena in Bangladesh, taking place on both banks of a river. This process extensively wears away huge prolific land areas at the river confluence zones due to the intricate processes of sediment-water interactions. The regions close to the study area (Padma–Meghna and Ganges–Jamuna confluence) are facing extensive erosion problems adjacent to the several important urbanized and economically developed regions (Ahmed et al. 2014).

Several comprehensive field assessments have already been done basically on the flow process in the confluence zones (Biron et al. 1993; De Serres et al. 1999; Baranya et al. 2015). The hydro-morphodynamic evolution of mountain river confluences was documented by Guillén-Ludeña et al (2016). Confluence morphodynamics has been studied mostly on the small fluvial channels (Mosley 1976; Biron et al. 1993; Boyer et al. 2006) but with the advent of recent technologies, large-scale fluvial channels are now the most focusing arena of scientists around the world (Ashworth et al. 2000; Parsons et al. 2007; Sambrook Smith et al. 2009). Many researchers across the world documented confluence morpho-dynamics of small, large river (Parsons et al. 2008; Hackney et al. 2018; Stevaux et al. 2009), gravelly braided river confluence (Wheaton et al. 2013) and tidal channel confluences (Ginsberg et al. 2009; Xie et al. 2018). Current advance in remote sensing and GIS enables us understand the extent to which river confluences are dynamic, planform characteristics and decadal evolution of the confluence points (Trigg et al. 2012; Lewin and Ashworth, 2014; Mount et al. 2013). In recent time, many aspects of river confluence like sediment, flow structure, and morpho-dynamics are being focused throughout the world to understand the process and mitigate the possible hazards (Ali et al. 2019; Szupiany et al. 2019; Sankey et al. 2018; Rhoads and Johnson 2018). In the river confluence zone, erosion–accretion is considered a prime focal point. Erosional process interrupts water quality and consumption for human and industry and river ecosystems (Benda et al. 2004; Gazi et al. 2020; Leite et al. 2012). River capacity decreases due to sediment accumulation that finally results in the modification of morpho-dynamics in the river confluence zone (Julien et al. 2010; Mohammed et al. 2011). This reduced capacity and tributaries discharge intensify flood risks due to heavy precipitation (Wang et al. 2016).

Some remote sensing and GIS-based studies emphasized on the dynamic behavior of the world major confluences including GBM river system confluence (Dixon et al. 2018; Szupiany et al. 2019). More focusing work on the river bank erosion and shifting at Chandpur near lower Meghna River has also been done using satellite images (Nath et al. 2013). It is worth to mentioning that important initiatives have been taken at different times to assess the erosion–accretion scenarios in the river banks of Bangladesh (CEGIS 2003; SWMC 2002; IWM 2005). Very few studies conducted on the morpho-dynamics behavior of confluence systems in Bangladesh till now.

The prime objective of this research is to assess the morpho-dynamics and shifting direction of the confluences and interpret the vulnerable zones surrounding the confluences as erosion and accretion are very prominent in the bank of the confluences. Understanding the morpho-dynamics of these areas will be highly efficient for managing land areas close to the river confluence zones.

2 Study Area

The study area covers Padma–Meghna confluence at Chandpur town, Chandpur district and Ganges–Jamuna confluence at Aricha Ghat, Rajbari district, which are two major confluences in Bangladesh (Fig. 1). The confluences are morphologically very dynamic in nature during monsoon and post-monsoon season since the water discharge and sedimentation rate are very high because of low elevation. Sand, silt, and clay are comprising the sediments of the Ganges–Brahmaputra–Meghna (GBM) river system at the downstream part in the Bengal basin, Bangladesh, which is accumulating under unidirectional energy conditions (Datta and Subramanian 1998).

Fig. 1
figure 1

Map showing the study area (Confluence points). GJC indicates Ganges-Jamuna confluence and PMC stands for Padma-Meghna confluence

Full size image

The Padma is a very large alluvial river with the longitudinal slope of about 5 cm per km (Halcrow 1993), bed material is in the range D16 = 0.02 mm, D50 = 0.09 mm and D84 = 0.2 mm and average discharge is 28.4 (103 m3/s) (FAP 24 1996). Meghna can be categorized as a meandering river with an average discharge of about 33 (103 m3/s) and bed material size D50 is about 0.09 mm. Ganges river is one of the major sediment dispersal systems (mostly suspended sediment) in the Bengal basin with a channel length of about 2974 km (Wasson, 2003). Jamuna is a braided river with an average braid plain width of 11 km and a flow depth of around 5 m (Sarker and Thorne 2006).

The average rainfall in the PMC and GJC regions is, respectively, 53.39 mm and 58.79 mm throughout the year. The water level varies in Jamuna, Ganges, Padma and Meghna river, respectively, 13.6–19.1 (mPWD), 5.4–13.7 (mPWD), 1.5–5.5 (mPWD) and 1.5–6 (mPWD), but during the monsoon and flood period, there is no specific range to remark (Sarkar et al. 2003).

3 Materials and Methods

Georeferenced multi-spectral satellite images from 1972 to 2019 (in the same month and season) were used to analyze the planform morpho-dynamics of the confluence points, and images were acquired from USGS (https://earthexplorer.usgs.gov/) (Table 1). The images were corrected radiometrically and geometrically. Geospatial software Erdas Imagine 14 and ArcGis 10.3 were utilized in this study for data analysis and interpretation. The individual bands of acquiring sensors have been added together by “Layer stacking”. To ease the processing and enhance spatial resolution, all the bands were stacked except the thermal band. Study area was selected for each image finally by “Subset” process. Using the ruler function in ArcGis, the confluence width was measured. The accuracy of the ArcGis ruler function was assessed by taking 100 repeat measurements at a known distance at the same virtual altitude and image resolution of the study area.

Table 1 Satellite images (used in this study) with their sensor, acquisition dates and spatial resolution (meter)
Full size table

Modified normalized difference water index (MNDWI) was calculated to differentiate between land and water boundary. Subsequently that was used to extract the erosional–accretional boundary and to locate the exact position of the confluence points. NDWI is generally calculated using (Xu 2006):

$$ {\text{MNDWI}} = \, \left( {{\text{Green}} - {\text{SWIR}}} \right) / \left( {{\text{Green}} + {\text{SWIR}}} \right) $$
(1)

Here,

SWIR = reflectance of short-wave infrared region and

Green = reflectance of green regions of the used Landsat images

Confluence angle was measured using the approach of Hackney and Carling (2011). The middle points of the channel center lines united and the angles of intersection (A1 and A2) were measured by means of a protractor with degree increments (Lublowe 1964). A3 was determined using the following formula:

$$ A3 \, = \, 360 \, - \, \left( {A1 + A2} \right) $$
(2)

4 Results and Discussion

4.1 Confluence Point Shifting

The major concern of the study is to enumerate the confluence point shifting. There was a rapid shifting of the confluence points over the years. Significant migration had been occurred in both confluence points affecting the surrounding region of the study area. The migration mainly occurred for the erosional and accretional activities of the river bank. High flow rate causes erosional action and low flow rate causes accretional activity.

4.1.1 Shifting of the PMC

The PMC moved about 9.17 km toward N 17.69° W from 1972 to 1980 (Table 2). In the next 10 years, the confluence migration was reversed and moved around 2.73 km toward S 13.79° E. From 1980 to 2010, the confluence migrated through the SE direction. But from 2010 to 2019, the confluence point was reversed and had chosen the direction from SE to NW. Eventually, from 1972 to 2019, the confluence point was migrated toward S 36.6° E and traveled about 6.77 km. Tracking the shifting trend of this confluence point is very crucial for the presence of important urbanized region that is very close to the eastern part of the confluence (Fig. 2).

Table 2 Direction and distance summary of confluence migration (PMC) from 1972 to 2019
Full size table
Fig. 2
figure 2

Temporal planform shifting of PMC and GJC from 1972 to 2019

Full size image

4.1.2 Shifting of the GJC

The GJC point was migrated about 2.37 km towards NE direction from the year 1972 to 1980, and from 1980 to 2019, the confluence point migrated towards the SE direction (Table 3). So, the migration is continuing towards the SE direction. The major migration was occurred from 1980 to 1990, migrated almost 6.07 km towards S 50.55° E. From 1972 to 2019, the confluence shifted around 13.15 km towards S 46.56° E. This confluence followed a specific trend from 1972 to 2019 towards S 46.56° E, about 13.15 km. This tremendous movement of the confluence point indicates the erosional activity of the riverbank in the SE direction (Fig. 3).

Table 3 Direction and distance summary of confluence migration (GJC) from 1972 to 2019
Full size table
Fig. 3
figure 3

Diagram showing the directional migration of confluence point (PMC and GJC) with 10 years interval from 1972 to 2019

Full size image

4.2 Confluence Width

The change in the width of river confluences both in the PMC and GJC infers the erosional and accretional trends surrounding the confluence zones along with river banks.

4.2.1 Temporal Variation of Width in the PMC

Variation in confluence width was very prominent in this zone. During the years of 1972, 1980, 1990, 2000, 2010, and 2019, the calculated confluence widths were, respectively, 6.87 km, 12.64 km, 11.63 km, 9.37 km, 6.03 km, and 6.98 km. From the year 1972 to 1980, the confluence width dramatically enlarged from 6.87 to 12.64 km accounting 5.77 km widening. It can be suggested that within this time frame, a tremendous amount of erosion occurred which indicates the high-water flow rate. However, from 1990 to 2010, confluence width decreased from 11.63 to 6.03 km representing a 5.3 km narrowing that indicated low water flow rate (Table 4). After that, the confluence width increased up to 6.98 km till 2019. So, the confluence width increased about 11 km throughout the study period.

Table 4 Confluence width in the PMC and GJC from 1972 to 2019
Full size table

4.2.2 Temporal Variation of Width in the GJC

The change of confluence width was also very pronounced in this region. Throughout the years of 1972, 1980, 1990, 2000, 2010, and 2019, the documented confluence width was about 8.1 km, 8.5 km, 5.38 km, 2.69 km, 3.01 km, and 2.8 km, respectively (Fig. 4). The maximum changes occurred from 1980 to 2000. The width decreased by nearly 5.81 km over the 20 years. The maximum width recorded about 8.50 km in the year of 1980. But, a slight increase in the confluence width occurred from 2000 to 2010. So, it can be expected that the confluence width might be decreased with time.

Fig. 4
figure 4

Showing the changes of the Confluence Width for both PMC and GJC from 1972 to 2019

Full size image

4.3 Sandbar Morphodynamics at the Confluences

The most important aspect of study areas is the dynamics of sandbar accumulation and recession. Mainly, sandbars are accumulating in the upstream channel of both confluences.

4.3.1 Sandbar Morpho-dynamics at PMC

There had been a rapid accumulation of sandbar in the PMC throughout the study period and redistribution of sandbars had also been noticed. Aerial extents of the sandbar in the confluence zone were 66.87 km2, 123.58 km2, 128.54 km2, 176.02 km2, 177.86 km2 and 211.96 km2 for the years of 1972, 1980, 1990, 2000, 2010, and 2019, respectively. Over the years, the study showed an increasing trend of sandbar accumulation in this confluence zone (Fig. 5). The total area increased from 1972 to 2019 was around 145.09 km2 (Fig. 6). The major sandbar in this confluence zone is in the Padma river just above the confluence point which indicates that the flow velocity through this Padma River is decreasing and the sediment flux is increasing day by day.

Fig. 5
figure 5

Showing the distribution and dynamics of Sand Bar in the PMC and GJC from 1972 to 2019

Full size image
Fig. 6
figure 6

Showing the distribution of sand bar at PMC and GJC from 1972 to 2019

Full size image

4.3.2 Sandbar Morpho-dynamics at GJC

A huge change had also been observed in the sandbar area in the GJC throughout the study period. In the years of 1972, 1980, 1990, 2000, 2010 and 2019, sandbar occupied about 69.96 km2, 88.57 km2, 16.11 km2, 38.61 km2, 15.42 km2 and 18.94 km2 area, respectively (Table 5). The result identified a decreasing trend of the sandbar deposition from 1972 to 2019 in the GJC that was reverse to the PMC. From 1972 to 1980, sandbar area increased, but from 1980 to 1990, a huge decrease occurred, again from 1990 to 2000, sandbar area increased and then followed a decreasing trend up to 2019 (Fig. 5). The major sandbar in this confluence zone is in the upstream of both the Ganges and Jamuna rivers which indicates that the flow velocity of both rivers is very high in this confluence zone than the zone above the confluence point.

Table 5 Sand bar areal statistics of both Confluences (PMC and GJC) from 1972 to 2019
Full size table

4.4 Erosion–Accretion Surrounding the Confluence Zones

Erosion–accretion is very prominent and has been continuing alternately along the banks of the river surrounding the confluence zones.

4.4.1 Erosion–Accretion Surrounding the PMC

In and around the PMC, the most erosion-prone period was 2010–2019 but erosion was most prominent in comparison with accretion in the period from 1990 to 2000. It was very distinct that land loss due to erosional activities had been increasing (57–89 Sq. Km) compared to accretion from 1972 to 2019 at the PMC (Table 6). Throughout the study period, the right bank of the lower Padma River was the most unstable area on the left side of the confluence zone (Fig. 7). However, in the recent decades, the area has undergone accretional activities and erosion, mainly approaching the central charland area within the confluence zone. The upper part of the confluence is showing an accretional trend, whereas the lower part of the confluence signifies an erosional trend.

Table 6 Erosion-accretion statistics in the both confluences from 1972 to 2019
Full size table
Fig. 7
figure 7

Spatio-temporal erosion-accretion trend in and around the PMC from 1972 to 2019

Full size image

Based on the shifting trend and erosion–accretion, the most erosion-prone areas around the PMC are Sakhipur, Bilaspur, Chandpur sadar, Srimandi, Sakhua, and char Atra. Chandpur Sadar and Srimandi. The present erosional trend suggests that within the next few years, maximum parts of the area will be eroded into the river (Fig. 8).

4.4.2 Erosion–Accretion Surrounding the GJC

Erosional activities decreased from 1972 to 1990, then increased slightly up to 2000, and declined again from 2000 to 2010 (Fig. 9). But, in the recent decades, erosion was most

Fig. 8
figure 8

Showing the most vulnerable zones around PMC due to shifting and erosional activities

Full size image

at the GJC compared to accretion. Both rivers carry huge amount of sediments but a high-flow velocity impedes sediments to be deposited in this confluence zone. So, accretion is very active in the upper part of the confluence zone, whereas erosion is evident in the lower part of the confluence zone, as evidenced from the absence of any sandbar in the lower part (Fig. 10). So, it can be predicted that erosion is very prominent in the lower part of the confluence zone. Aricha Ghat, Daulatdia, Masundia, Khanganj, and Nayakandi regions are more vulnerable to erosion in and around the GJC (Fig. 11).

Fig. 9
figure 9

Spatio-temporal erosion-accretion trend in and around the GJC from 1972 to 2019

Full size image
Fig. 10
figure 10

Erosion-accretion pattern in the PMC and GJC with 10 years interval from 1972 to 2019

Full size image
Fig. 11
figure 11

Showing the most vulnerable zones around GJC due to shifting and erosional activities

Full size image

5 Conclusion

The planform morphodynamics of both river confluences showed a wide range of adjustments in case of channel changes, accretions (bar depositions), and erosions within the confluence zone, channels migrating within a defined belt via braiding. The confluences are highly mobile that migrate an order of magnitude greater than the channel width. PMC and GJC demonstrate the dynamic trend of the confluence Point shifting. GJC followed particular movement of the confluence point, whereas PMC did not. GJC confluence point was moved NW direction from year 1972 to 1980 but from 1980 to the present, the confluence is moving through SE direction. On the other hand, PMC confluence moved through NW direction from the year 1972 to 1980 but for the next 30 years, from 1980 to 2010, the confluence moved through SE direction and most importantly, from 2010 to present, the confluence point is moving through NW direction.

Confluence width in the PMC varied significantly throughout the study time frame. The calculated width was 6.87 km2, 12.64 km2, 11.63 km2, 9.37 km2, 6.03 km2, 6.98 km2, respectively, in the years of 1972, 1980, 1990, 2000, 2010, and 2019. But the GJC zone had been followed a decreasing trend of confluence width change over the 47 years, the confluence width declined from 8.10 to 2.80 km2. Sandbar area of the PMC increased from 1972 to 2019 over 47 years about 147.09 km2 and the increment was more than three times in 2019 than that of 1972. In contrast, GJC is showing a decreasing pattern where the total sandbar area decreased about 51.02 km2 from 1972 to 2019. From the analysis of erosion–accretion, it can be clearly predicted that erosion is very prominent in the recent decades from 2010 to 2019 around the both confluences. The left bank of lower Padma in the lower confluence area of PMC is highly vulnerable due to erosion and downstream part of the Ganges river, the left side of GJC confluence zone has been recognized as the most vulnerable zone.

The study has been successful in delineating the relationship between the dynamic nature of confluence and the vulnerability assessment of the PMC and GJC. The study could be taken as a reference in identifying the vulnerable zones and would help in further mapping and planning of Government projects or industrialization around the two major confluences (PMC and GJC).