Fluvio-geomorphic change of the Padma-Meghna river course using the NDWI and MNDWI techniques

ABSTRACT

This study intends to investigate the fluvial-geomorphic change and erosion-accretion processes of two significant rivers. The 2149 km² research area’s overall erosion and accretion were depicted using multi-temporal Landsat images and Normalized Difference Water Index, Modified Normalized Difference Water Index methods to measure and investigate change. However, focus has been placed on four unique sections that are prone to erosion, and the bank migration of the section has been measured in order to depict the actual state of erosion. The Padma-Meghna River eroded approximately 11.0, 77.0, and 13.0 km² between 1973–1978, 1978–1989, and 1989–1994. However, 78.0 and 48.0 km² of accretion occurred in the years between 1994 and 1999 and 1999 and 2004. Moreover, between 2004 and 2021, the river’s erosional activity was remarkably dynamic compared to deposition. Besides, 64 km² and 4 km² of land accretion occurred during 2008–2014 and 2014–2019, while 77 km² and 68 km² of land eroded during 2004–2008 and 2019–2021. The overall rates of erosion and accretion are 5.12% and 4.04%, respectively. Statistically, both banks of the Padma River eroded about 2 km from 1973 to 2021. Padma-Meghna confluences are more unstable and susceptible to change. The Meghna’s left bank moved 4.5 km east between Goshairhat and Haimchar in 1989, and in 2008, the river shifted 4.8 km west from the river of 1994. In 1989, it moved 5 km east from a previous location between Goshairhat and Haiderganj, and in 2008, it moved 6.3 km west from the 1989 river. The bank migrated 12 km eastward in 2014 from where it was in 2008 through the bar development. The Mehendiganj Upazila’s eastern side lost more than 7.4 km between 1973 and 2021, and the current major channel is simply 7.4 km west of the 1973 river. The findings of the dynamic changes might be useful to planners, developers, researchers, and government agencies as they develop and put into practice new development initiatives and river basin management.

GRAPHICAL ABSTRACT

KEYWORDS:

• NDWI
• MNDWI
• riverbank erosion
• Padma river
• Meghna river

Introduction

Bangladesh is a riverine country due to its extensive network of rivers that flow through it, including the Ganges, Brahmaputra, and the Meghna River. Major parts of the country are covered by flood plains. The rivers and streams of this country have continuously changed their course and developed many civilizations from time to time. However, this continuous river course shifting causes riverbank erosion due to soft flood plain deposits, which is a very old and serious problem in our country. Riverbank erosion is displacing thousands of people and robbing them of their land every year (Das, 2011). In Bangladesh erosion has resulted in the loss of agricultural land and landlessness, which is a contributory cause of poverty (Rana, 2017). Twenty out of Bangladesh’s 64 districts are subject to riverbank erosion, which erodes around 8700 hectares of land annually and impacts over 200,000 people by damaging their houses and agricultural land (Alam, 2017; Freihardt & Frey, 2023). River course change is significant for affecting riverbank erosion and the development of the flood plain from a geomorphological standpoint. Understanding the processes is crucial for explaining the formation of fluvial features because they are among the most dynamic parts of the landscape (Hooke, 1979). The bulk of Bangladesh’s rivers feature sandy bottoms, low slopes, a lot of meandering, and banks that are prone to erosion and shifting (Bangladesh Water Development Board BWDB, 2020). Bangladesh is situated in the largest Ganges delta in the world. The Ganges Delta is often referred to as the Ganges-Brahmaputra-Meghna (GBM) River Delta (Islam, 2016; Nicholls & Goodbred, 2005; Woodroffe, Nicholls, Saito, Chen, & Goodbred, 2006). The huge, active river courses in the deltaic zone, including the Ganges-Brahmaputra-Meghna (GBM) River move a substantial quantity of sediment (10⁹ tons/year) through discharge water (Figure 2), that building the Ganges delta (Islam & Gnauck, 2008; Sarker, Huque, Alam, & Koudstaal, 2003). The name Padma gets its name after the confluence of the Ganges and Jamuna Rivers, at Aricha-Goalundo area of Faridpur district, in the northern part of the study area. The Meghna River links the Padma near Chandpur to form the Padma-Meghna confluence in the middle and the Meghna estuary forms at the mouth of the Bay of Bengal in the south.

Figure 1. Location of the study area.

The Padma transports 900 Mt of sediment annually, whereas, the Upper Meghna transports 13 Mt (Mondal et al., 2020). The Padma is a very large alluvial river that has a 5 cm/km average slope and the Meghna River is a low-energy, multi-channel fluvial system with a system of linking channels and inter-channel zones (Alam, 1991; Chowdhury et al., 2021). Furthermore, the Upper Meghna has a 2 cm/km average slope, and the Lower Meghna has a 5 cm/km average slope (Gazi, Roy, Mia, & Akhter, 2020; Mondal et al., 2020).

Some recent research (Arefin, Meshram, & Seker, 2021; Chowdhury et al., 2021; Eshita, Bhuiyan, & Saadat, 2023; Ferdoush, Biswas, & Mondal, 2022; Halder & Mowla Chowdhury, 2021; Rashid, 2020) suggests that both the Padma and Meghna have substantially changes in the last 50 years due to erosion, bar development and migration of the riverbank. Variation of discharge water can influence the pattern of sedimentation, bank erosion, bar formation and bank shifting.

The seasonality of the three mighty rivers has a significant impact on water levels between high and low-flow seasons. Almost 80% of Bangladesh’s entire river flow occurs from July to October and 20% during the dry season (Bangladesh Water Development Board BWDB, 2020) (Figure 2). The difference of water level between monsoon and dry seasons is around 7 m on the Ganges, 5–6 m on the Padma, and 3.5 m on the lower Meghna (JICA, 2005).

Figure 2. Monthly water discharge flow of the major rivers in Bangladesh (modified after Islam, 2016).

Remote sensing-based satellite image analysis has a great deal of promises for researchers because of the many purposes of the study and has become a low-cost alternative for detecting characteristics and comprehending hydrogeological systems in well-mapped areas where field data is scarce (Acharya, Subedi, Huang, & Lee, 2019). It is currently the most popular in several research disciplines like urban planning, agriculture, and environmental monitoring. However, over the last few decades, various water body-mapping approaches using remote sensing data. The Normalized Difference Water Index (NDWI) is a popular approach based on the water spectral indices (Feyisa, Meilby, Fensholt, & Proud, 2014; Fisher & Danaher, 2013; Gao, 1996; Hollstein, Segl, Guanter, Brell, & Enesco, 2016; Li et al., 2019; McFEETERS, 1996; Xie, Luo, Xu, Pan, & Tong, 2016; Yang & Du, 2017; Yang et al., 2015). Present research applied NDWI and MNDWI technique to delineate the river, erosion-bar formation, bar dynamics and migration pattern of the river. In recent, many experts throughout the world, including this country, have used remote sensing technologies to investigate river morphology and dynamics. Several works are worth mentioning (Alam & Singh, 2021; Arefin, Meshram, & Seker, 2021; Chowdhury et al., 2021; Dewan et al., 2017; Gazi, Roy, Mia, & Akhter, 2020; Halder & Mowla Chowdhury, 2021; Islam & Islam, 2021; Kumar Pal, Rahman, & Anika Yunus, 2017; Langat, Kumar, & Koech, 2019; Mahmood, Ahmed, Zhang, & Li, 2020; Rashid, 2020; Rashid & Habib, 2022; Rashid, Habib, Khan, & Islam, 2021; Singha & Pal, 2021). Dewan et al. (2017) quantify the channel parameters like erosion, deposition, mean channel width, sinuosity, braiding index, etc. of two significant sections of the Ganges-Padma system using GIS.

Rashid (2020), used RS and GIS techniques to determine morphological aspects of the lower Padma River in Bangladesh, such as channel bar development, braiding index, sinuosity, and its relationship with bank line migration. However, few studies have used the NDWI and MNDWI indices to characterize the river course change, bank migration, and dynamic behavior of the major rivers of the Padma and the Meghna over a large area. Arefin, Meshram, and Seker (2021) investigated the Padma River’s channel shifting and its impact on land usage and land cover using RS and GIS techniques. The NDWI approach was used to distinguish between a water and land boundary. Even though Gazi, Roy, Mia, and Akhter (2020), used the MNDWI approach to identify the link between the dynamic character of confluences and vulnerability assessments at the Padma-Meghna and Ganges-Jamuna confluences, their research was limited to the two confluences of the Major River’s. Chowdhury et al. (2021) used the single-band image classification technique to analyze total erosion in Mehendiganj upazila in the lower Meghna estuary. However, the goal of this study is to represent gross erosion bar formation, bar dynamics, and channel shifting across the whole study area, with an emphasis on four distinct dynamic locations and associated change scenarios from 1973 to 2021. The research area covers 2149 km² from the confluence of the Padma and Meghna rivers, as well as a portion of the Lower Meghna that includes parts of Munshiganj, Madaripur, Shariatpur, Barishal, Chandpur, and Lakshmipur districts (Figure 1).

Materials and method

The following tasks have been taken into consideration to complete the analysis and fulfill the objectives of the research: choosing a study area, gathering data, processing images, and extracting water bodies using NDWI and MNDWI methods (Figure 3). With the aid of the software programs QGIS 3.16 and ArcGIS 10.5, the study was accomplished. In order to delineate, analyze, and compare specific river characteristics, a five-year interval (1973–2021) of dry season, cloud-free Landsat MSS, Landsat TM Level 2 and Landsat 8 OLI Level 2 imagery was obtained from the USGS website (Table 1). NDWI and MNDWI-classified images were used to compute erosion and accretion in the research area.

Figure 3. Methodological flowchart.

Table 1. Specifications and lists for satellite images used for NDWI and MNDWI (source: USGS, https://earthexplorer.usgs.gov).

Satellite	Sensor	Path/Row	Year/month/day	Resolution (m)	Wavelength (µm)
Landsat MSS	MSS	147/44	1973/02/02 1978/02/21	60	Band 4: 0.5–0.6 Band 7: 0.8–1.1
Landsat-5	TM	137/44	1989/01/12 1994/01/02 1999/02/01 2004/02/15 2008/12/10	30	Band 2: 0.52–0.60 Band 5: 1.55–1.75
Land Sat-8	OLI	137/44	2014/01/25 2019/01/07 2021/11/28	30	Band3:0.533–0.590 Band6:1.566–1.651

Image processing

Image processing is a prerequisite for further satellite imagery investigation, such as land cover and land use mapping (LULC) and change detection analysis. The Dark Object Subtraction (DOS) atmospheric correction method is employed for image processing through the SCP plugin of QGIS, which was developed by Congedo (2021). Landsat (TM) and Landsat 8 images are of 30-m resolution; however, Landsat MSS images have a 60-m resolution. To maintain the uniformity and accuracy of the study, Landsat MSS images of 60-m resolution are regenerated to 30 m resolution by a resampling technique.

Normalized difference water index (NDWI) and modified normalized difference water index (MNDWI)

The normalized differential water index (NDWI) is a well-known method for locating surface water bodies. McFeeters first introduced NDWI in 1996 to detect surface waters in wetland environments and to allow for the measurement of surface water extent. Using remote sensing images’ green and near-infrared (NIR) bands, based on the phenomenon that the water body has strong absorbability and low radiation in the visible to infrared wavelength range (Du et al., 2016; McFEETERS, 1996). Although NDWI was initially developed using Landsat MSS data, it has now been effectively applied to other sensor data for surface water body identification (Chowdary et al., 2008; McFeeters, 2013).

$\mathit{NDWI}=\left(\mathit{Green}-\mathit{NIR}\right)/\left(\mathit{Green}+\mathit{NIR}\right)$

NDWI value ranges from -1 to 1, water body shows always a positive value and non-water body shows negative value. McFEETERS (1996) proposed zero as the threshold (Ji, Zhang, & Wylie, 2009). For water NDWI > 0 and for non-water NDWI ≤ 0. Though NDWI proposed by McFEETERS (1996) have a good result, in some cases, it is unable to discriminate built-up area to water feature, due to buildup area also shows positive value like water (Ji, Zhang, & Wylie, 2009).

Overcome this type of difficulty to discriminate built-up area to water feature Xu (2006) proposed a modified NDWI (MNDWI), where he replaced SWIR band instead of NIR band for Landsat TM band 5. The MNDWI is capable of extracting surface water while suppressing errors from built-up lands, vegetation, and soil (Xu, 2006; Yang & Du, 2017; Yang, Zhao, Qin, Zhao, & Liang, 2017).

$\mathit{MNDWI}=\left(\mathit{Green}-\mathit{SWIR}\right)/\left(\mathit{Green}+\mathit{SWIR}\right)$

To distinguish between rivers and land in the research region, NDWI is used on Landsat MSS and modified NDWI (MNDWI) on Landsat TM and Landsat OLI. NDWI and MNDWI were calculated by using raster calculator tools of ArcGIS 10.3. Bands 4 and 7 of the Landsat MSS were utilized as green and NIR, respectively. Green is assigned to Landsat TM band 2 and short-wave infrared (SWIR) to band 5, whereas green and SWIR are assigned to Landsat 8 band 3 and band 6, respectively. The natural break (Jenk) approach was used to classify NDWI and MNDWI images. The land-water boundary was distinguished by close visual interpretation of class values and then performed reclassification and raster-to-polygon approach due to the export final shape of the river and land. For change analysis, the main channel, as well as the accompanying bar/land and connecting tributary-distributary features, are considered, while other distracting tiny features are removed from the research region using the attribute table.

Result

Erosion, accretion, and rate

The status of Padma-Meghna River in 1973 is taken as base area for all Erosion-Accretion/bar formation analysis for this study. Erosion/Accretion bar formation and average rate of all the changes of river course shifting described in the following. 2149 km² of the mapped area was considered the total study area.

A total 246 km² of land was lost due to river course change and 194 km² of land was gained due to bar formation, with yearly average change of 5.12 km² and 4.04 km² respectively between 1973 and 2021 (Table 2). Significant erosions observed during the comparison years between 1973 and 1978, 1978 and 1989, 1989 and 1994, 2004 and 2008 and 2019 and 2021.

Table 2. The erosion and accretion of the bank area and bar over the period of 1973–2021, (-) and (+) value indicate erosion and accretion, respectively.

Year	1973–78	1978–89	1989–94	1994–99	1999–004	2004–008	2008–14	2014–19	2019–21	1973–21
Erosion/Accretion (Km^2)	-11.0	-77.0	-13.0	+78.0	+48.0	-68.0	+64.0	+4.0	-77.0	-52.0
Rate (km^2/year)	-2.2	-7.0	-2.6	+15.6	+9.6	-17.0	+10.67	+0.8	-38.5	-1.08
Total Erosion (km^2)	-246	Erosion rate= Total Erosion/Total year Accretion rate= Total Accretion/Total year
Total Accretion (km^2)	+194
Total year	48
Erosion rate	-5.12
Accretion rate	+4.04
Net erosion rate	-1.08

While for the corresponding periods of 1994–1999, 1999–2004, 2008–2014, and 2014–2019, substantial accretion was observed in the study area. The rates of erosion and accretion/deposition for the same duration were -2.2, -7.0, -2.6, -17.0, -38.5, and 15.6, 9.6, 10.67, and 0.8 km², respectively. From 2019 to 2021, land erosion was much higher than the rest of the time. In just 2 years, 77 km² of land was lost at a rate of 38.5 km² per year (Table 2, and Figures 4, 5, and Figures 6).

Figure 4. The figure represents the rivers’ year-by-year changes (a) 1973 and (b)1978, (c) 1989 (d) 1994 (e) 1999, (f) 2004 (g) 2008, (h) 2014, (i) 2019, (j) 2021 (k) 1973 and 2021.

Figure 5. Comparative graph of river and land area in km² over the period of 1973 to 2021.

Figure 6. Erosion and accretion of land in the study region from 1973 to 2021, computed using landsat MSS, landsat TM, and landsat 8 OLI NDWI images and MNDWI classified technique.

Analysis of the river course shifting

Four separate cross-section areas are examined in order to estimate and measure the bank line migration from a certain site and to illustrate the true scenario of change in an area (Figure 7). Using the ArcGIS 10.5 and QGIS measuring tools, the change detection from a location for all section lines was calculated overlaying the river courses of all years. The comparison and estimation are displayed in Figures 8–11.

Figure 7. Map showing course of the river from 1973 to 2021, with change dynamics in section AB (A, Naria speedboat station to B, Hasali bazar), CD (C, Goshairhut UPDB to D, Haimchar UHC), EF (E, Ekota bazar to F, Haiderganj), and GH (G, Mehediganj fire station to H, Moju chowdary launch ghat).

Figure 8. It depicts the Padma river’s left and right bank distances, river width and bar length in different years, along the section AB (see Figure 7), from Naria speed boat station to Hasali Bazar.

Figure 9. The section CD (see Figure 7) depicts the Meghna river’s left and right bank distance, including river width and bar, from goshairhut upazila Parishad Dak Banglo (C) to Haimchar Upazila health complex (D) from the year 1973 to 2021.

Figure 10. The section EF (see Figure 7) depicts the Meghna River’s left and right bank distances including river width and bar, from Ekota Bazar (E) to Haiderganj (F) for the year 1973 to 2021.

Figure 11. The section GH (see Figure 7) shows the left bank and right bank distances, including river width from Mehendiganj fire station (G) to Mojuchowdary launch ghat (H) of the Meghna river for the year 1973 to 2021.

Erosion-accretion and river course shifting in section AB

The Padma is a mighty and very dynamic river, which changes its course at many spots over time. The Naria Upazila, located on the southern bank of the Padma River, is one of the most erosion-prone areas in the study area. The southern bank of Padma migrated more than 1.5 km to the north of the Naria in 1989, compared to the bank’s position in 1973. From 1989 to 1994, erosion started toward the south and continued until 2021 (Figure 4).

From 1989 to 2021, the Padma shifted a few hundred to more than 1.5 km of its bank by moving in an arc-shaped manner toward the south (Figures 7 and 8). During 1994–1999 and 2014–2019, the southern bank of this area shifted almost 1.5 km toward the south, which was significantly greater compared to other years. The river water distance from point “A” in 1973 was 2.76 km, but at present, it is just about 400 m from the same location. Between 1973 and 2021, about 2 km of land was lost due to erosion, and the river shifted to the south, close to the point “A” (Figures 7 and 8). In 1973, the right/north bank of the river was 2.47 km away from point “B;” however, the distance has continuously decreased. The right bank has lost over 2 km of land between 1973 and 2021.

Erosion-accretion and bank migration in section CD

The left bank of the main channel of Meghna shifted toward the east between (C) and (D) during 1973–1989. The bank distance of the channel from (C) in 1973 was 8.28 km, and in 1978, the distance increased by more than 4.5 km until 1989 (Figures 7 to 11). At this time, the left bank gained land through bar development and shifted 4.5 km east of the main channel of 1973. While, during the periods of 1989–1994, 1994–1999, 1999–2004, and 2004–2008, the same bank of the channel lost 2.18, 1.92, 2.3, and 0.6 km of land, respectively, shifted toward the west by 4.8 km in 2008 from the river of 1994 (Figures 7 and 9).

A massive channel migration happened between 2008 and 2014. The left bank of the main channel migrated by bar development around 11 km to the east in 2014 from the river in 2008. Between 2014 and 2019, the riverbank was stable compared to the period of 2019–2021. In 2019–2021, the left bank lost nearly 1 km of land to erosion (Figures 7 and 9). Noticeable erosion also happened on the right bank. The right bank was comparatively stable from 2008 to 2021.

Erosion-accretion and bank migration in section EF

The left bank and right bank distances of the main channel were measured between (E) and (F) (Figure 7). The left bank lost 1.3 and 2.0 km of land between 1973–1978 and 1978–1989, respectively, by erosion. However, minor accretion occurred from 1994 to 2004, but the left bank was almost stable from 2008 to 2021 (Figure 10). If we consider the left bank from the edge of the bar in 1989, there was a substantial change that occurred in 1999, and in that time the river shifted to the west by 3.8 km of bar erosion. In contrast, 3.8 km eastward shifting occurred from 1999 to 2021 between “E” and “F” by a new bar formation.

The right bank of the channel shifted more than 1.0 km toward the west by accretion from 1973 to 1989. The right side of the channel enlarges its width by 5.7 km of bar erosion during that time. However, a new bar formation occurred to the west, and the right side of the river gained a straighter pattern compared to 1973 and 1978. The right bank of the channel gradually shifted to the west until 1999, and in 1999, it was shifted around 3 km from 1978. In 1999, the bank started shifting to the east, and 2.7 km of shifting occurred from 1999 to 2021.

Erosion-accretion and bank migration in section GH

The Mehendiganj subdistrict of Barishal district is one of the most erosion-prone areas in the study area, which is separated by the main channel of Meghna to the east and a small channel to the west. Erosion takes place all around the landmass during this span of time and forms the present shape of the island. It is observed that the left bank distances of the main channel from (G) to (H) (Figure 6) have gradually decreased over the years and almost 1 km² of land was lost in every span of time from 1973 to 2014. Significant erosion occurred during the years 1973–1978, 1978–1989, 1989–1994, 1994–1999, 1999–2004, 2004–2009, and 2009–2014 (Figures 7, 11, and 12). From 2014 to 2021, the left bank of the main channel was almost stable. However, during 1973–2021, this island lost nearly 7.4 km of land from (G) to (H), and in the present day, the main channel of the river flows just 7.4 km westward from the channel of 1973. The right bank of the channel from (H) gained 1.1 km of land during 1973–1978 by bar development. However, in 1978–1989, the right bank lost 2.1 km of land, including the newly formed bar of 1973–1978 (Figures 7, 11, and 12).

Figure 12. (a) Decadal changes of rivers between 1973–1989, 1989–1999, 1999–2008, and 2008–2021, (b) river rearrangement for better visualization of the changes, layer 2021 in the upper and 1973 in the lower.

Discussion

Change dynamics of the Padma and the Padma-Meghna confluence (PMC)

Change dynamics in the years 1973 to 1978

In 1973, the Padma River was bifurcated by a single big channel bar between the left and the right bank of the river section AB and the channel width was 10.3 km (including the bar). The channel was narrowed to the right to meet the confluence of two rivers between Bhederganj and Chandpur, creating a new bar between the eastern edge of sections AB and PMC. In 1973, the main channel width between the bank and section AB’s eastern side ranged from 2.5 to 3.0 km. In 1978, the bifurcated channel bar extended to the east, while another bar had completely gone due to erosion, resulting in two Padma-Meghna confluence places. Both banks along AB eroded greatly, and the channel expanded in area compared to 1973. However, the right side extended more than the left (Figures 12(a,b)).

Change dynamics in the years 1978 to 1989

Between 1978 and 1989, a large bar at Bhederganj emerged and expanded between Bhederganj, Matlab, and Chandpur Sadar (left and right sides of the PMC). As a result of the newly formed bar, the channel narrowed to the south, while the main channel and PMC migrated north between Munshiganj Sadar-Mohonpur and Matlab. The Padma River changed shape from Janjira-Naria to Munshiganj-Matlab (Section AB to PMC) by eroding the left bank around Janjira and Bhederganj, as well as territory gained by Naria. The right banks of the Padma near B, as well as Meghna’s right bank of Matlab, and have experienced significant erosion. During that time, the confluence with the lower Meghna eroded an area of nearly half a kilometer to 2 km. The massive outflow of water caused by a significant flooding event in 1988 (Dewan, Nishigaki, & Komatsu, 2003; Rashid, 2020).

Change dynamics in the years 1989 to 1994

In the span of 1989–94, nearly 2–9 km in width and a 24 km long braided bar formed between AB and Chandpur, and the Padma dissected into north and south channels. The North Channel meets the Meghna near Monoharpur, and the South Channel meets near Chandpur. In that period, huge erosion occurred on the south bank along the Naria-Bhederganj and Lohajang-Tongibari-Munshiganj to the right bank. The channel migrated to the south, which is nearly 1–2 km to A and 7–10 km to Bhederganj.

Change dynamics in the years 1994 to 1999

A comparison of the channels between 1994 and 1999 revealed that both Padma banks expanded their banks between Janjira-Naria, Lohajang, and Munshiganj (AB). The north bank migrated to the north by 1.5 to 2 km along the Lohajang bank line, as well as 0.1 to 1.5 km in the Munshiganj Sadar, while the right bank shifted to the south by about 1.0 to 1.5 km in Tongibari. This North Channel caused a meandering bend between Lohajang and Munshiganj. The south bank extends about 1.0 to 2.0 km south of Janjira and 0.3 to 1.0 km south of Naria. The bar between that area and the next, which was formed in the previous era, also lost its length and width during that period and formed a nearly rectangular pattern in 1999.

The erosion and accretion of the river and adjacent floodplain are influenced by water level and discharge. When the river is flowing at high levels, energy accumulates, and the river bank erodes even more. In 1998, the country experienced severe flooding, which had a substantial impact on river morphology. Erosion is related to water levels and discharges in a progressive manner. Reduced water levels and outflow contribute to erosion (Sarker & Rahman, 2018).

Change dynamics in the years 1999 to 2004

Observed that in 2004, the bar extended its length by around 5.3 km N-E from the position of 1999 and formed an approximately 5–6 km wide and 23-km-long bar, which divides the Padma into two sub channels (north channel and south channel) between Janjira and Lohajang (Figure 13). The bar extended between the Janjira-Lohajang-Tongibari-Naria and Bhederganj-Munshiganj and the North Channel, which meets the Meghna near Monoharpur, Matlab. The south channel, bifurcated again by a small bar, meets the Meghna downstream from Monoharpur and jointly meets near Chandpur. The south channel extended its route more to the south by 1.3 km at Janjira compared to the position in 1999.

Figure 13. The figure represents the rivers’ year-by-year changes between 1973 and 1978 (a), 1978 and 1989 (b), 1989 and 1994 (c), and 1994 and 1999 (d).

Huge erosion occurred at Janjira, Bhederganj, and part of the Naria. The north channel reshapes its flow pattern by extending its right bank a few hundred meters to a kilometer to the north at Tongibari and half a kilometer to 1.5 km to the south by an eroding bar and continues to Tongibari-Munshiganj Sadar. The south channel began its meandering lope in 1999, which continued through 2004, and into the later period. The morphological features of the Padma River were straighter, with the exception of minor meandering loops in different areas (Dewan et al., 2017) and Rashid (2020) remarked that the statement does not apply to the entire Padma River.

Change dynamics in the years 2008

Huge changes were observed in 2008 on both banks of the Padma. In this period, the channel bifurcates as usual like the previous year but reshapes its route simultaneously by straightening the south channel and developing (upward loop in Lohajang-Tongibari and downward in Tongibari-Munshiganj) two meandering loops in the north channel (M shape/sinusoidal). The right bank of the south channel shifted 0.5 km to 1.5 km to the north by eroding along the NW-to-SE elongated bar. As a result, the bar decreased its width but extended its length by rejoining a small bar and forming a narrow tip of land to the meeting point of the two rivers between Bhedarganj and Chandpur. On the left bank, a retreat is also about a few hundred meters to 1.3 km to the south between Naria and Bhedarganj (A to PMC), and a few hundred meters to more than 2.0 km to the north by the right bank eroding along the bar. The river is readjusting due to widening and migration on both sides of the south channel, and it turns from a meandering loop to a straight flow pattern.

Change dynamics in the years 2014 to 2021

The channels were restored to their previous pattern in 2014, with the exception of some erosion in some areas, such as the downward loop being extended a few hundred meters to the south, and the left and right banks of the south channel being extended by a few hundred meters and nearly half a kilometer, respectively, at Janjira-Naria and Bhederganj (left bank close to A).

On both banks of the south channel, massive erosion occurred in 2019 and 2021. In 2019, the left and right banks of the south channel extended 1.0 to 1.5 km and roughly 1.0 km to a few hundred meters between Janjira-Naria and Naria-Naria, respectively, compared to their 2014 positions. In comparison to the previous year, the right bank of the south channel extended its courses left to right from northwest to the south-east in 2021 by an eroding bar (left side) (Figure 14). Between Lohajang and Janjira-Naria (Figure 13), the south channel eroded around 3.8 km to 1.7 km and 1.5 to a few hundred meters alongside the bar in a north-west to south-east direction. Huge bar erosion occurred between Bhederganj and Chandpur during that period in the south of the left portion of the bar.

Figure 14. The figure indicates a year-by-year comparison of river changes from 1999 to 2004 (e), 2004 to 2008 (f), 2008 to 2014 (g), 2014, 2019, and 2021 (h).

Change dynamics of the lower Meghna

Change dynamics in the years 1973 to 1989

In 1973, the Meghna River developed a triangle shape between Goshairhat, Hizla, and Mehendiganj on the left bank and Haimchar, Haidarganj, and Ramgati on the right bank. It was bifurcated by a NW-SE elongated braided bar (Char Haim), which was roughly 20 km long and 6 km wide (Figure 7, section CD, EF). In comparison to the left side of the bar, the channel narrowed to the right and the bar degraded 1.5 km in the direction it was heading in 1973 and 1978. A delta-shaped bar (N to S 11–19 km long, E to W 2–19 km wide) was developed by extra sedimentation and channel turned to the left side. By this point, the channel on the right was broadening in comparison to the left of the previous year (Figure 12(b)). Nearly half of the previous era’s delta bar eroded right to left along N-S (2–9 km) in 1989. The Meghna reshaped itself by straightening its route and comparatively shifting to the left by enlarging its channel (Figure 12(b)).

The left bank of the Meghna shifted a few hundred meters to approximately 3.0 km to the west between Goshairhat and Bhederganj, and the right bank along Haimchar shifted a few hundred meters to more than 1.0 km to the east. Resulting from this erosion in 1989, the river got a straighter segment compared to 1978 from Goshairhat to Haimchar. It is worth mentioning that the river in 1973–78 was bifurcated between Goshairhat and Haiderganj, but in 1989 it migrated its route straight due to massive erosion. The massive migration is most likely caused by runoff from upstream during the 1988 flooding event.

Change dynamics in the years 1994 to 2021

From 1994, the left bank of the river started shifting to the west by developing a meandering lope along Bhederganj, Goshairhat, and Hizla, and simultaneously, a sporadic mid-channel bar (Char Haim) also started forming between Goshairhat and Haimchar. Moreover, the right bank shifted to the east by linear erosion between Haimchar and Haiderganj, which continued for every span of time until the present (Figures 6,Figures 9, and 12).

In 1999, the mid-channel bar increased its length and width, and the channel shifted to the west by 1.9 km and 3.3 km along the line’s CD and EF, respectively. In 2004, the channel started bifurcating (West channel and East channel) between the lines CD and EF by increasing the width of the bar, and the left bank shifted to the west like the previous year. In that period, the bar increases its width on both sides of the bar, as well as forming a sporadic bar beside the left bank, and the mid-channel bar turns into a braided bar. The bar formation process continued into the later period but increased its width to the west by squeezing the width of the west channel compared to the east.

Both banks of Meghna increased their width from 1994 to 2021 in between the line’s CD and EF by a meandering loop in the west and by linear erosion on the right bank. However, the main channel of the present-day river increased its width, shifted to the east by linear bar erosion, and narrowed its left channel by continuous land gain. The south of the main channel broadened its path by continuously eroding Mehendiganj Island over the study period (Figure 15). It is worth noting that the forms of the rivers did not alter much between 1973 and 1978, with the exception of a broadening in 1978 in certain regions. During the years 1978–1989, both rivers experienced significant natural and human-caused alteration, including the development of new bars, the relocation of existing bars, and the reshaping and enlargement of channels.

Figure 15. Represent the overall changes of the river between 1973 and 2021, here purple color indicate area common in both of year.

Probable causes of significant fluvio-geomorphic changes in the rivers

Many factors could have contributed to the dramatic alterations in rivers after 1980. Intense rainfall and catastrophic flooding can induce rapid changes in river morphology due to momentous bank erosion, migration, and sedimentation. Substantial flooding occurred in this country in 1987, 1988, 1998, 1999, 2004, 2010, and 2017, which may have resulted in significant erosion, sedimentation, and geomorphic changes in the Padma-Meghna River and its catchments. Upstream dams, such as the Farrakkha Barrage, impede the normal flow of water and sediment downstream and this could be one of the reasons, because it acts as a barrier to natural water flow from upstream and sedimentation throughout the downstream in the Bengal Delta. Other major causes of erosion include bank undercutting, bank slumping due to seasonal fluctuations in water level and material contrast, local scour from turbulence caused by the bank protection dam’s blockage, and bar development due to excessive sedimentation. Furthermore, the lack of gradient in the river bed limits water-holding capacity, and both rivers have undergone considerable modifications due to human activity.

Conclusion

According to the study analysis and findings, both the Padma and Meghna rivers have changed significantly throughout time. However, the patterns of river course change and movement behaviors of those rivers, as well as erosion and accretion, differ from place to place and time to time. In terms of erosion-accretion, bar development and relocation, channel migration and pattern change the Padma and Meghna are quite dynamic. Between 1973 and 2021, the left bank (north) of the Padma lost almost 2 km of land due to erosion and river course shift to the south. The right bank also lost nearly 2 km of land between 1973 and 2021. Within the cross-section line EF, it is observed that the main channel of the Meghna River migrated 5 km to the east in 1989, compared to its position in 1973. The period of shifting began in or before 1989 and lasted until 2008. Between 1973 and 2021, the eastern boundary of Mehendiganj Upazila lost over 7.4 km of land (see section GH). The Meghna River’s main channel now runs 7.4 km westward compared to the 1973 main channel. During the years 2008–2014, there was a significant shift. Due to bar growth to the east, the left bank of the main channel of Meghna changed almost 12 km from its position in 2008. The left bank began to erode again in 2014–2019 and 2019–2021. During this time, between 1.1 and 1.4 km of land were eroded, and the channel changed back to the west due to the erosion of more than 1 km of bar per year. It is also noted that as well as shifting and reshifting the meeting point with the construction of bars and changing their positions over time and space, the confluence of the two rivers is dynamic. In comparison to the gain process, the total rate of erosion remains dominant. This research will assist planners, developers, researchers, and government organizations in taking the next step in development.

However, the remote sensing investigations do not fully support this assumption because of image resolution and seasonal fluctuations were not taken into account. Additional research, such as geological, geotechnical, and tectonic investigations, is required to determine the causes of the major changes throughout time. This study advises that the river basin be managed in an integrated and comprehensive manner, including cooperation from the government, non-government organizations, society, and foreign states that share the river.

Author contributions statement

Faruk Hossain: Conceptualization, Methodology, Software, Writing Original draft manuscript. Mohammad Ashraful Kamal: Supervision, Writing, Reviewing and Editing. Tahera Afrin: Data curation, Visualization, draft manuscript preparation.

Geolocation 

This research focused on the Padma and Meghna River course and Confluence of the Padma and Meghna River, Bangladesh.

Acknowledgments

The authors would like to express their sincere appreciation to the Director-General of the Geological Survey of Bangladesh (GSB) for allowing them to conduct this research. Dr. Bazlar Rashid, Deputy Director (Geology), GSB, also contributed useful research papers and comments to the writer. The authors also expressed gratitude to the USGS officials for allowing the use of previous satellite images.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author upon reasonable request.

Additional information

Funding

The authors reported there is no funding associated with the work featured in this article.