2012 Geomorphology Brahmap

GEOMOR-03977; No of Pages 12 Geomorphology xxx (2012) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Geomorphology

journal homepage: www.elsevier.com/locate/geomorph

Tectonic controls on the morphodynamics of the Brahmaputra River system in the upper Assam valley, India

Siddhartha K. Lahiri a,b,⁎, Rajiv Sinha a a Engineering Geosciences Group, Indian Institute of Technology, Kanpur 208016, India b Department of Applied Geology, Dibrugarh University, Dibrugarh 786004, India article info abstract

Article history: The Brahmaputra is one of the largest tropical rivers of the world and is located in an area of high structural Received 28 May 2011 instability as evidenced from the presence of a large number of earthquakes in the Himalayan catchment Received in revised form 10 April 2012 through which it flows. Syntectonic evidence of changes in the morphodynamics is difficult to identify for Accepted 11 April 2012 the large rivers. Nevertheless, we note that the Brahmaputra River has become astonishingly large in plan- Available online xxxx form in a historical timescale. Reconstruction of planform changes over a period of 90 years in the upper reaches of the Assam valley shows that the 240-km-long channel belt is widening all along its course in Keywords: Himalayan tectonics the region. From the average width of 9.74 km in 1915, the channel belt has widened to the average width Indo-Burmese Arc of 14.03 km in 2005 (44% widening), and in certain reaches the average widening is as high as 250%. However, River dynamics the bank line shift is not symmetric along both banks. Further, the planform characteristics of the Brahmaputra River reveal significant spatial and temporal variability from upstream to downstream reaches, and we attribute this variability to tectonogeomorphic zonation of the river based on subsurface configuration and channel slope. Further, the tributaries joining the northern and southern banks of the Brahmaputra differ remarkably in terms of river dynamics, and this is attributed to the differences in tectonic regimes of the Himalaya in the north and the Naga Patkai hills in the south. © 2012 Elsevier B.V. All rights reserved.

1. Introduction In contrast, the upper reaches of the river in Assam have received inadequate attention so far (Goswami, 1985; Sarma and Phukan, 2004, The Brahmaputra is a large mountain-fed river system (Sinha and 2006; Kotoky et al., 2005; Sarma, 2005). The Brahmaputra River divides Friend, 1994) draining the tectonically active Himalayan foreland the upper Assam valley into two distinct geographic zones, the north basin through Assam (India) and Bangladesh and has the distinction and the south banks. In addition, this reach of the river hosts one of of being the river with the highest sediment yield (852.4 t/km2/y; the largest alluvial islands in the world, the Majuli Island, which is Latrubesse, 2008) in the world. The Brahmaputra acts like an efficient known for its unique cultural heritage (Sarma and Phukan, 2004) conduit to transfer a large sediment flux from the Himalaya (the source) apart from its geomorphic significance. The Brahmaputra River shows (Singh, 2006; Singh et al., 2006)totheBayofBengal(thesink). With significant geomorphic diversity in this region, which is strongly mani- a mean annual discharge of 21,200 m3/s, measured at Bahadurabad, fested in the morphodynamics of the river at a historical timescale. Bangladesh (Latrubesse, 2008), the Brahmaputra is the seventh largest Moreover, the tributaries joining the main river from these two banks river in the world (Hovius, 1998; Tandon and Sinha, 2007) and has are quite different in terms of their morphometric characteristics and created a thick and extensive valley fill in its alluvial reach. The large- temporal dynamics. The present study documents the spatial diversity scale dynamics of the Brahmaputra River has fascinated geomorpholo- in morphodynamics of the Brahmaputra River in the upper Assam val- gists across the globe for over three decades; and a large number of ley as well as the tributaries joining the northern and southern banks. studies, particularly in the lower reaches of the river in Bangladesh, We argue that such spatial variability in morphodynamics has been bear testimony to the international attention this river has received strongly influenced by different tectonic domains in this region. (see Coleman, 1969; Bristow, 1987; Curray, 1994; Goodbred and Kuehl, 1998, 2000; Richardson and Thorne, 2001; Goodbred et al., 2. The study area 2003). The study area includes a 240-km×80-km corridor in the Brahma- putra valley that is sandwiched between the NE–SW bound Himalayan

⁎ Corresponding author at: Department of Applied Geology, Dibrugarh University, frontal thrust (HFT) and the Naga-Patkai thrust (NPT) (Fig. 1A, B). The Dibrugarh 786004, India. Tel.: +91 373 2370247; fax: +91 373 2370323. older topographic maps suggest that three rivers (the Siang, the Dibang, E-mail address: [email protected] (S.K. Lahiri). and the Lohit) used to meet at a place called Kobo to form the

1000 km

Dh Di Ti

Si Legend Prominent thrusts Blind thrusts Jo Faults Places Study area

Fig. 1. (A) Three major reaches of the Brahmaputra River and the valley shown by encircled numbers: I, the upper Assam reach; II, the lower Assam reach and III, the Bangladesh reach. (B) The study area in the upper reach of the Brahmaputra valley with three distinct units is shown on the IRS-P6-LISS-3 image. Unit 1 is from the oldconﬂuence of three rivers: the Lohit, the Dibang, and the Siang to Dibrugarh in the downstream direction. Comparatively steady, unit 2 continues up to the upper tip of Majuli Island. Segment 3 includes Majuli Island and extends towards the Mikir hills. Abbreviated place names are Dh — Dhemaji; Di — Dibrugarh; Ti — Tinsukia; Si — Sibsagar; NL — North Lakhimpur; Jo — Jorhat.

Brahmaputra during 1915. This conﬂuence point shifted by ~16 km multipattern river that has a tendency to very frequently generate an downstream to a place called Laikaghat by 1975. By 2005, a farther down- anabranching (Latrubesse, 2008) pattern in decadal scale. stream shift of 19 km took place as observed in the satellite imagery. Fig. 1B shows the tectonic setting of the study area. Apart from the Though the Brahmaputra has been described as a braided river, the regional-scale thrusts such as the MBT (Main Boundary Thrust), HFT, conventional deﬁnitions for braided rivers (Lane, 1957; Leopold and and NPT, several other faults have been mapped in this region based Wolman, 1957; Bristow, 1987) as anastomosing channels or the pres- on the interpretation of seismic data acquired by the oil companies ence of a number of bars and islands having an intertwining association (Murty, 1983; Ranga Rao, 1983; Bally, 1997; Narula et al., 2000; with channels is not adequate to describe it. In the study reach of the Kent et al., 2002; Goswami and Goswami, 2007). Some of these faults upper Assam area, the Brahmaputra appears to be a multichannel and are trending in the NE–SW direction and others are crisscrossing them.

A Bankline shift (km)

Fig. 2. The bank lines and the nature of the bankline shift of the channel belt of the Brahmaputra River are shown during three different times over the last 90 years (1915–2005).

H1 Unit 1 L1 Mean elevation = 114m H2 Mean slope = 0.35m/km H3 L2 Unit 2 H4 H5 Mean elevation = 98m L3 Mean slope = 0.25m/km L4 L5 Unit 3 Mean elevation = 77m Mean slope = 0.21m/km Channel belt width (km)

Relative widthvariation–1915(%) =((Place width -5.89)/5.89)*100 Av. widthin1915=5.89km Minimum= -66.89304atX = 128.65 Maximum= 56.19694atX = 122.4

UNIT 1 UNIT 2 UNIT 3 Relative width variation–1975(%) =((Place width–8.88)/8.88)*100 NE SW Av. widthin1975 = 8.88km Upstream Downstream Minimum= -70.4955atX = 150.31 Maximum= 49.77477atX = 3.43 Width variability - 1915 Relative width variation–2005(%) Minimum= 2.34atX = 12.65 ((Place width–11.96)/11.96)*100 Maximum= 23.11atX = 146.56 Minimum= -72.15719atX = 152.37 Mean= 9.74 Maximum= 85.45151atX = 4.68

Widthvariability - 1975 (Note: In the above calculations, Minimum= 5.38atX = 236.86 Majuli Maximum= 21.93atX = a 146.56 was excluded from the channel belt. Mean= 11.60 As a result, average width decreased Considerably. For example, average Width variability - 2005 Width with Majuli in 1915 was 9.74km Minimum= 4.71atX = 236.86 C And without Majuli 5.89km) Maximum= 22.54atX = 146.56 Mean= 14.03

UNIT 1 UNIT 2 UNIT 3 NE SW Upstream Downstream

Fig. 3. (A) The planform variation in the widths of the channel belt of the Brahmaputra River (Majuli Island included) in three different times and units shown in the location map. (B) Longi- tudinal proﬁle and the channel slopes at different stretches. H — highs, L — lows stand for the geomorphic highs and lows. (C) Relative changes in widths of the channel belt from the average width without the Majuli Island. We observed that the relative changes in widths, planar as well as temporal, show an increasing trend caused by the ‘damming’ effect of the geomorphic highs.

Most of these faults are blind (no surface expression) but have influ- from severe bank erosion (Sarma and Phukan, 2004). Unit 3 is 121 km enced the river systems significantly (discussed later). long; and three important tributaries join in this unit: the Subansiri from the north bank and the Dikhau and Dhansiri from the south bank. 3. Data and approach Each unit was further subdivided into 37 reaches, 4.5–9kmlong,(unit 1: 9; unit 2: 9 and unit 3: 19 reaches), to measure bank line shifts and For this study, the IRS-P6-LISS-3 images acquired on 15 December all planform parameters such as sinuosity and braid–channel ratio 2005 with a spatial resolution of 23.5 m and older topographic maps (after Friend and Sinha, 1993), channel belt width, channel area, and of 1:253,440 scale corresponding to 1912–1926 and 1977 (scale: bar area for different periods. All data were integrated into a GIS environ- 1:250,000) have been used. Digital image processing of the satellite ment to document the morphodynamics of the Brahmaputra and its trib- images obtained from the National Remote Sensing Centre, Hyderabad, utaries flowing through the study area and to understand the causative India, was carried out to enhance the geomorphic features for mapping. factors of temporal variability of the geomorphic parameters men- Shuttle Radar Topographic Mission (SRTM) data with spatial resolution tioned above. The following section presents the results of the morpho- of ~90 m and vertical resolution of ~1 m were used to find point eleva- dynamic analysis of the Brahmaputra and its tributaries for these three tions and for computing slope. All temporal data were georeferenced units. and registered on a common platform for investigating the temporal variability in bankline, channel width, and planform parameters of the 4. Results and interpretation Brahmaputra and its tributaries for three different time periods: 1915, 1975, and 2005. In addition, several unpublished reports of the Oil 4.1. Bank line shift and changes in width of the Brahmaputra River and Natural Gas Corporation Limited (ONGCL) and Oil India Limited (OIL) were used to understand the distribution of subsurface faults in All three units are nearly linear with sinuosities ranging from 1.02 to the Brahmaputra valley in the upper Assam. The Brahmaputra River in 1.05 for the median courses (both the old and new courses). Fig. 2A–D the study area was divided into three major geomorphic units (Fig. 1B) shows the positions of the north and south banks of the Brahmaputra for analysis. The uppermost and widest unit 1 starts from the confluence River for different time periods for all three units. A careful analysis of of the Siang, Dibang, and Lohit rivers; extends to 51 km downstream; the maps shows that the nature and extent of the bank line shifts are and contains a newly formed large alluvial island (Dibru-Saikhoa Island, quite different in the three units. Unit 1 shows the maximum shift locally called ‘new Majuli’). The 68-km-long unit 2 is the narrowest; and (see Fig. 2D), and the south bank has been much more mobile in the re- two important tributaries, the Burhi Dihing and Disang, join the Brah- cent times compared to the north bank. However, the bank line shifts maputra on the southern bank in this unit. Unit 3 is Majuli Island, the are not uniform during the period of study for both north and south largest alluvial island in the world, which has been under serious threat banks. For example, the mean shift in the north bank for the period

Table 1 Temporal variation in the parameters like average channel belt widths, areas of the channel belts, channels, and sand bars, and the ratios of channel and channel belt areas as well as braid bar areas and channel areas are shown for three units of the Brahmaputra River in the upper reach of the Brahmaputra valley.

Units Geomorphological parameters 1915 1975 2005 Net change Net change (1915–1977) (1915–2005)

Unit 1 Average channel belt width (in km) 5.28 10.65 18.48 5.37 13.20 (+102%) (+250%) Channel belt area (in km2) 358.92 678.39 1186.27 319.68 827.56 (+89.12%) (+230.7%) Channel area (in km2) 122.04 144.13 270.24 22.09 148.2 (+18.1%) (+121.4%) Sand bar area (in km2) 234.08 534.26 916.03 300.18 681.95 (+128.24%) (+291.32%) Channel area/channel belt area 0.35 0.21 0.23 −0.14 −0.12 (−40%) (−34.3%) Sand bar area/channel area 1.92 3.71 3.39 1.79 1.47 (+93.2%) (+76.6%) Unit 2 Average channel belt width (in km) 6.25 8.80 9.42 2.55 3.17 (+40%) (+51%) Channel belt area (in km2) 460.32 655.47 698.42 195.15 238.1 (+42.4%) (+51.72%) Channel area (in km2) 128.33 164.49 205.64 36.16 77.31 (+28.3%) (+60.2%) Sand bar area (in km2) 331.98 490.98 492.77 159 160.79 (+47.9%) (+48.4%) Channel area/channel belt area 0.28 0.25 0.29 −0.03 0.01 (−10.7%) (+3.57%) Sand bar area/channel area 2.59 2.98 2.40 0.39 −0.19 (+15.06%) (−7.33%) Unit 3 Average channel belt width (in km) 13.93 13.70 14.38 −0.23 0.45 (−1.6%) (+3.26%) Channel belt area (in km2) 1789.16 1756.32 1855.49 −32.84 66.33 (−1.83%) (+3.71%) Channel area (in km2) 300.18 473.22 365.55 173.04 65.37 (+57.64%) (+21.78%) Sand bar area (in km2) 1489.0 1283.0 1490.0 −206 1 (−13.8%) (+0.06%) Channel area/channel belt area 0.17 0.27 0.20 0.10 0.03 (+58.8%) (+15%) Sand bar area/channel area 4.96 2.71 4.07 −2.25 −0.89 (−45.36%) (−17.9%)

1915–1975 was 1.45 km, whereas a shift of about 0.7 km (approximately (Fig. 3C) shows several geomorphic ‘highs’ and ‘lows’ in this stretch. half) was recorded for the south bank during the same period. However, The upper part of unit 1 (with an average slope of 0.38 m/km) shows the mean shift in the north bank during the period 1975–2005 was just the maximum temporal variability in channel belt width (250% during 0.06 km, whilst the mean shift in the south bank was 2.05 km (~34 the period 1915–2005; see Table 1). A dramatic change occurred between times!!) during the same period. Further, the reaches showing maxi- 1915 and 1975 during which the average width of the channel belt of unit mum shifts in both north and south banks are located in the most 1 doubled. Most of unit 2 with an average slope of 0.21 m/km shows upstream reaches (unit 1), whereas the minimum shifts are recorded the least variability. Table 1 shows that the average width of unit 2 in the downstream reaches (unit 2 or 3). Though the overall tendency was 6.25 km in 1915, more than the average width of unit 1 in the of shifts of the riverbanks was positive (widening), a significant same year (5.28 km). In contrast to the dramatic change in unit 1, the amount of negative shifts (narrowing) is recorded in selected reaches average channel belt width of unit 2 increased to 8.8 km in 1975 and at different times (see Fig. 2). The minimum south bank shift was then to 9.42 km (a total of 51% increase during 1915–2005). The 3.8 km (at X=206.57 km, unit 3) during 1915–1975, whereas the min- most downstream unit 3 has variable slopes (average 0.14 m/km) imum north bank shift was 5.62 km (at X=163.12 km, unit 3) during and shows moderate variability in width over the years. This unit has 1975–2005. remained relatively stable during the last 90 years or so (Table 1). A As a consequence of the overall positive bank line shift in both minor decrease in channel belt width between 1915 and 1975 (~2%) banks, the widths of the channel belt of the Brahmaputra River have is noted followed by an ~3% increase in 2005. Therefore, no net increase changed significantly over the years. Measured at several points along is recorded in the average channel belt width in this segment during the median course of the belt, the average channel width in the study the period of observation (1915–2005). area increased steadily from 9.74 km in 1915 to 11.6 km in 1975 and then further to 14.03 km in 2005 for the 240-km-long stretch of the 4.2. Morphodynamics of the Brahmaputra River river. However, the individual reaches showed varied patterns of increase. A continuous increase in width of the channel belt is observed If the sediment supply increases for some reason (e.g., tributary in unit 1. Unit 2 also shows an increasing trend, albeit with a lower contribution, deforestation in the catchment) for mountain-fed rivers rate, and unit 3 does not show much change. flowing through comparatively young valleys, the vertical incision di- Fig. 3B shows the temporal changes in the relative width of the minishes and the average thalweg depth decreases. To accommodate channel belt (relative to average width for the corresponding year the average discharge, lateral erosion dominates in the alluvial reaches of observation) in different units. A plot of longitudinal profile de- where channel banks are composed of unconsolidated sand. For situa- rived from elevation data from SRTM DEM for 179 points along the tions where all planform parameters such as channel belt width, chan- median path of the 240-km-long stretch of the Brahmaputra channel nel area, sand bar area, and channel belt area keep on increasing, the

CHB:358.92sq.km CHB:684.69sq.km BB: 244.31sq.km BB: 439.83sq.km CH: 114.61sq.km IF: 83.27sq.km CHB:1184.55sq.km BB/CH:2.13 CH: 161.59sq.km BB: 609.37sq.km IF: 261.54sq.km BB/CH:2.72 CH: 275.26sq.km BB/CH: 2.21 10km 10km AB10km C

CHB:460.76sq.km BB: 329.08sq.km CHB:656.38sq.km CHB:703.46sq.km CH: 129.29sq.km BB: 483.86sq.km BB: 494.06sq.km BB/CH:2.55 CH: 172.52sq.km CH: 209.40sq.km BB/CH:2.80 BB/CH:2.36

10km 10km D 10km E F

CHB:1859.22sq.km CHB:1763.28sq.km CHB:1787.68sq.km BB: 977.85sq.km BB: 642.65sq.km BB: 689.53sq.km IF: 508.21sq.km IF: 640.50sq.km IF: 787.87sq.km CH: 373.16sq.km CH: 480.13sq.km CH: 310.28sq.km BB/CH: 2.62 BB/CH: 1.34 BB/CH: 2.22 G H I 10km 10km 10km

Fig. 4. Changing morphology of the braid bars situated within the channel belt of the Brahmaputra River for three units shown in the location map during 1915–2005.

Please cite this article as: Lahiri, S.K., Sinha, R., Tectonic controls on the morphodynamics of the Brahmaputra River system in the upper As- sam valley, India, Geomorphology (2012), doi:10.1016/j.geomorph.2012.04.012 6 S.K. Lahiri, R. Sinha / Geomorphology xxx (2012) xxx–xxx temporal variability in the ratio of channel area to channel belt area can propensity to change in channel belt area except for a few erosion-prone give us additional information about the aggrading tendency. Highly sites. In unit 3, the channel belt area seems to be fairly stable during the negative change may indicate aggradation, and a positive trend should period of study (Fig. 5A), but the bar areas have changed significantly. be manifested in degradation. Similarly, a large positive change in the In addition to this, a reversal of trend is observed — the reaches where ratio of bar area to the channel belt area should be associated with ag- the channel belt decreased earlier were widened later and vice versa. gradation, and the negative change with degradation. This is most likely related to the anthropogenic intervention because Fig. 4 shows the major morphodynamic changes in the three units of the Majuli Island as well as the river banks in the adjoining reaches that the Brahmaputra River. We have also measured various planform pa- have been protected through embankments (existing embankment rameters such as channel belt area, sand bar area, and channel area for length is about 160 km; District Disaster Management Plan of Jorhat, three different years (1915, 1975, and 2005) to quantify the morphody- 2011) are often breached during the flood season. These embankments namic changes. constrain the bank erosion and lateral shifting for a few years. However, Fig. 5A–D shows the planform variability of the channel belt, braid sediment load is not distributed freely in the adjacent flood plains and bars, channels, and the braid bar/channel area ratio of the Brahmaputra the river bed rises very soon. Subsequently, the river breaches through River for two different periods (1915–1975 and 1975–2005). During the embankments and crevasse channels develop, adding to channel these periods, reach 4 in unit 1 shows a sharp rise of 21 km2 in channel multiplicity. Units 1 and 3 show a fluctuating trend over the two periods belt area (19 km2 in 1915 to 40 km2 in 1975) and 101 km2 (from mentioned above in terms of bar area, channel area, and their ratios 40 km2 in 1975 to 141 km2 in 2005) caused by the avulsive character (Fig. 5B–D). In general, unit 1 showed an aggrading trend and unit 3 of the Lohit River that brought the Dibru Saikhoa reserve forest (new showed a degrading trend from 1915 to 1975. During 1975–2005, these Majuli) within the Brahmaputra channel belt. Unit 2 shows a much lesser trends were reversed. Moreover, though the channel belt was widening

A B Increase Decrease Expansion Contraction Channel belt area per unit reach length

C D Aggradation Increase Decrease Degradation Degradation

UNIT 2 UNIT 3 UNIT 1 UNIT 2 UNIT 3 UNIT 1 SW NE SW NE Upstream Downstream Upstream Downstream

Fig. 5. Temporal variations of morphologic parameters like areas of channel belt, braid bars, channels, and braid bar/channel ratio per unit reach length during 1915–1975 and then 1975–2005. The aggradation trend shifts from unit 1 to unit 3.

Please cite this article as: Lahiri, S.K., Sinha, R., Tectonic controls on the morphodynamics of the Brahmaputra River system in the upper As- sam valley, India, Geomorphology (2012), doi:10.1016/j.geomorph.2012.04.012 S.K. Lahiri, R. Sinha / Geomorphology xxx (2012) xxx–xxx 7 in unit 1 during 1975–2005, unit 3 was undergoing a major aggradational Sinha, 1996). The unidirectional (westward) shifting of all the northern phase. tributaries is certainly remarkable. Three important southern tributaries of the Brahmaputra River 4.3. Dynamics of the Brahmaputra tributaries originate in the Naga Patkai hills, namely, the Disang, Burhi Dihing, and Dikhau. Out of these, the Disang has a much longer flow path The Brahmaputra River receives tributaries along the north bank (by about 20 km) in comparison to the other two before they join originating in the eastern Himalayan watershed as well as along the the Brahmaputra. Further, a series of subtle impressions of palaeo- south bank draining through the watershed of the Indo-Burman Range. channels is observed on the imagery along the southern bank of the We note significant differences in fluvial dynamics of the northern and Brahmaputra River. These palaeochannels do not show up on the his- southern tributaries. torical topographic maps; and therefore, they must be older. This im- Most of the major tributaries joining the Brahmaputra from the plies a larger avulsion frequency of these channels. Historical maps do north have distinct palaeochannels as revealed from a comparison show, however, a number of oxbow lakes (Fig. 7) formed during the of historical maps and recent satellite images (Fig. 6A). This suggests last 90 years (1915–2005); and this suggests that meander migration that avulsion is the dominant mechanism of fluvial dynamics at a his- has been one of the common mechanisms of channel shift in southern torical scale. For example, three rivers (viz. the Jiya Dhol, the Sisi, and tributaries. the Simen) have avulsed westward between 1915 and 2005, and Three prominent rivers of the south bank (namely, the Burhi Dihing, palaeochannels are marked to the east of the modern channels. The the Disang, and the Dikhau) have average sinuosities of 1.37, 2.06, and other tributaries have remained fairly stable during this period. A com- 1.96, respectively (Fig. 8A–C), within the valley after their emergence parison of sinuosity of the older and new channels reveals that, except from the Naga Patkai hills and then joining the Brahmaputra River. for the Simen, the older channels have distinctly lower sinuosities com- Fig. 8 also plots the reach-scale sinuosities for these rivers for 1915 pared to the new channels. This observation is contrary to the general along with slope of the individual reaches. The Disang River shows a belief that the new avulsion channels have morphological characteris- continuous increase (~50%) in sinuosity over a period of 90 years in ticssimilartotheolderchannels(Allen, 1965). Such morphological all reaches with high valley slopes. The Burhi Dihing and Dikhau Rivers changes owing to avulsion are fairly common in the Gangetic Rivers also show an increase in sinuosity in reaches of high channel slopes, and have been reported by previous workers (Richards et al., 1993; although we note an overall decrease in sinuosity temporally.

D Jiya Dhol E Sisi C Subansiri F Simen

Upstream Downstream Upstream DownstreamUpstream Downstream Upstream Downstream 94 AB

50 km km

G Dikari H Depi I Silli J Remi

Upstream Downstream Upstream Downstream Upstream Downstream Upstream Downstream

Fig. 6. Subansiri and seven other smaller tributary rivers (namely the Jiya Dhol, the Sisi, the Simen, the Dikari, the Depi, the Silli/Leko Jan, and the Remi) of the north bank of the Brahmaputra valley. The recent course of all the rivers, just after entering into the valley area, witnessed a drastic southwestward avulsion in the last 90 years. A comparative study of the channel slope variation and the temporal change in sinuosity.

5. Discussion particularly in the frontal parts. In such tectonically active regions, a common manifestation is a very dynamic river regime — but it is 5.1. Tectonogeomorphic zonation often difficult to relate the two without a sound geophysical data and characterisation of major tectonic domains. Based on the tectonic map Seismic data indicate that the thickness of the alluvium along the of the area interpreted from the seismic data acquired by the oil compa- Himalayan Frontal Thrust in the Brahmaputra valley reaches about nies (see Supplementary Figs. I–III),wehaveattemptedafirst-order 2km(Das Gupta and Biswas, 2000), whilst the average surface eleva- tectonogeomorphic zonation of the valley: (i) the central uplift, (ii) the tion in these parts of the valley is ~100 m amsl (above mean sea level). slopes, and (iii) the depressions (Fig. 9). This zonation has helped us to Accumulation of thick alluvial fills requires not only a high sediment flux explain the spatial variability in fluvial dynamics. from the hinterland but also continuous creation of accommodation The central uplift is a NE–SW trending block, around 7466 km2 in space in the foreland. In active foreland basins such as the Himalaya, area that can be divided into upper and lower uplift zones. The upper crustal readjustments through tectonic movements (subsidence) are uplift zone with an area of ~4286 km2 is closer to the Naga-Patkai obvious mechanisms for creation of a large accommodation space, thrust, and it seems to have influenced the southern bank line of the

2005 1915 (d) Much older than 1915

D C

0 5 km 0 2 km

ge ran illy i h tka Pa ga Na 0 50 km

E B

0 5 km 0 5 km

Fig. 7. (A) The Burhi Dihing River, a tributary belonging to the south bank of the Brahmaputra River. Sixteen stretches of this river have been studied. Different windows of this river are blown up to project some of the interesting morphological characteristics. (B) In the hilly region (stretch-3) the old and the new courses show almost identical characteristics; (C) the valley part (stretches 4–7) shows a clear imprint of a number of palaeochannels. The degree of variation of the river course is apparently only caused by the usual mechanism of meander migration; (D) a drastic reduction (stretch 11) in sinuosity; and (E) before joining the River Brahmaputra (stretch 16); the river almost maintained the same signature the way it was maintained in the hilly area.

A) Burhi Dihing

E W Upstream Downstream

Fig. 9. First order morphotectonic classiﬁcation of the upper reach of the Brahmaputra B) Disang valley based on three types of features: uplift, slopes and depressions. Abbreviated place names are Dh — Dhemaji; Di — Dibrugarh; Ti — Tinsukia; Si — Sibsagar; NL — North Lakhimpur; Jo — Jorhat. Major thrusts; MCT: Main Continental Thrust; MBT: Main Boundary Thrust; HFT: Himalayan Frontal Thrust.

to the HFT, and the southern margin runs approximately parallel to the Naga thrust. The northeastern end is marked by another blind fault parallel to the Mishmi thrust. Of the two slopes, the northern slope is principally controlled by the Himalayan orogeny, and the southern slope is controlled by the Indo-Burman plate dynamics. Further, thick accumulation of sediments has been recorded in three major depressions, the Subansiri, the Dhansiri, and the Lohit, in the seismic sec- tions (see Supplementary Fig. I). Presently, unit 1 and most of unit 2 of the Brahmaputra channel belt are mostly confined within the northern slope. Unit 3 falls in the lower part of the central uplift zone and is flanked by the southern slope. E W Based on the measured data on morphodynamics of the different Upstream Downstream units (Table 1), the interpreted response for the period of study (1915–2005) is summarised in Table 2. Unit 1 with a high valley slope shows the maximum bank line shift and channel belt widening, C) Dikhau albeit with varying rates during the two time slices (1915–1975 and 1975–2005). Sediments eroded from the banks, mostly through slumping during falling stages (Coleman, 1969; Goswami, 1985), are deposited in the channel belt and are manifested as the overall increase in sand bar area in this unit (Table 1). Unit 1 is therefore characterised as a major zone of aggradation. Unit 2 shows a much lower but increasing trend in channel belt widening and minimal bank line shift and continuous increase in sand bar area. We have therefore characterised this unit also as a zone of aggradation, except that some minor degradation in some reaches may have occurred in the last ~20 years as reflected from a negative change in the ratio of sand bar area to channel area. We argue that the morphodynamic response of units 1 and 2 is a manifestation of northern slopes that resulted in a higher sediment supply. Higher sediment supply in turn results in the ‘wandering of thalweg’ (Coleman, 1969), undercutting of banks, and slumping. As a result, unit 1 has shown a very high degree of morphodynamics during the E W last 90 years, and the effects have gradually diminished in unit 2 because Upstream Downstream of distal position of this unit. In contrast, unit 3 has a significant lower average channel slope and shows a distinctly different trend in morpho- Fig. 8. A comparative study of the channel slope variation and the temporal change in dynamics. The most significant temporal changes observed in this unit sinuosity of three prominent tributaries in the south bank of the Brahmaputra River, are significant channel narrowing (and deepening?) in certain reaches namely the Burhi Dihing, the Disang, and the Dikhau. as evidenced from negative bank line shift (Fig. 2). Although we observe Brahmaputra River. The lower uplift zone with an area of ~3258 km2 is an increase in channel area, sand bar area decreased by ~14% during closer to the Himalayan front, and the Majuli Island is located in this 1915–1975, but the trends reversed during 1975–2005 and nearly zone. The northern boundary of this zone is a blind fault that runs parallel reverted to 1915 conditions. We characterise this unit as a zone of

Table 2 Differential response of the Brahmaputra channel belt as interpreted from the morphodynamics during the period 1915–2005.

Units Tectono-geomorphic setting Observed morphodynamics Interpreted response and terrain characteristics (1915–2005)

Unit 1 Northern slope, average Maximum bank line shift, south bank more mobile in recent time Zone of major aggradation and channel channel slope 0.35 m/km; Continuous increase in channel belt width and channel area (CA/CBA=−34%) widening; highly dynamic in response mean elevation 114 m Sand bar area increased drastically (SBA/CA=+76%) to higher sediment supply from the northern slopes Unit 2 Northern slope, average Minimum bankline shift Zone of aggradation (with some reaches channel slope 0.25 m/km; Increasing trend in channel belt width and channel area but with a of minor degradation) and minimal mean elevation 98 m lower rate (CA/CBA=+3.57%) dynamics; gradual diminishing of the Sand bar area decreased (SBA/CA=−7%) effect of slopes Unit 3 Central uplift zone, average Moderate shift in bank line, pockets of signiﬁcant negative shift (narrowing) Zone of degradation, channel narrowing channel slope 0.21 m/km; Channel belt width stable, some increase in channel area (CA/CBA=+15%) (and deepening?) in response to uplift mean elevation 77 m Sand bar area nearly unchanged (SBA/CA=−18%)

degradation and argue that this is the response of the uplift in this zone. by increasing sinuosity in that reach to maintain the average gradient However, this unit has seemingly started experiencing aggradation in (Holbrook and Schumm, 1999; Schumm et al., 2000; Marchis and recent times. A recent large magnitude earthquake (M=8.7) in 1950 Napoli, 2008). Our data show that most of the southern tributaries of and subsequent readjustment of the valley (post-1950 interseismic the Brahmaputra have experienced a systematic increase in sinuosity changes) as evidenced from a number of earthquake events of M>5 over a period of 90 years, particularly in the reaches of high valley slopes. after the 1950 earthquake (Tandon, 1954) may have been responsible In addition, many of these rivers show sharp bends in the mountainous for such reversals in morphodynamics. reaches (see Fig. 7B). The Burhi Dihing River (Fig. 7C) also shows a lateral Further, the reach-scale degradation–aggradation in each unit is shift in a N–NW direction (Fig. 7D). Further, the Disang River, in spite of probably caused by the second-order tectonic differentiation. Variable having a highly sinuous course throughout, flows through a nearly sediment dispersal and the variable geologic control may also be generat- straight path for a 60-km-long stretch; and a closer examination re- ing variable stress patterns. As a result, the Brahmaputra is crossing a veals that this stretch nearly coincides with the boundary of the central number of second-order faults that are manifested as ‘highs’ and ‘lows’ uplift and the southern slope. We suggest that these features are mani- (Fig. 3C). The ‘damming’ owing to the regional highs (Schumm et al., festations of active tectonics related to thrust belt tectonics of the Naga 2000) increases aggradation, lateral erosion, and relative increase in the Patkai Range or the orogeny associated with the Mikir hills. width of the river, which we observe when the temporal and planar trend of relative width variation is compared with the valley elevation 5.3. Regional tectonics and morphodynamic model and slope (Fig. 3B and C). The exact relationship between these second- order faults and the first-order major faults is not clear yet, and this Some of the workers have identified the upper Assam valley as a needs further investigation using high resolution subsurface data. ‘seismic gap’ (Khattri, 1987) between the epicentres of two great earthquakes of 1897 and 1950; others have called it ‘aseismic’ (Nandy and 5.2. Spatial variability in morphodynamics and tectonic controls Das Gupta, 1991). Mukhopadhyay (1984) endorsed the ‘aseismic’ hypothesis and opined from his seismotectonic studies that the tectonic The northern and southern tributaries of the Brahmaputra River in stress gets distributed within the thrusts of the Himalayan and Burmese the upper Assam valley show significant difference in fluvial dynamics. arcs lying on either side of the upper Assam valley and that this keeps Whilst the northern tributaries are marked by frequent avulsions (mostly the valley aseismic. The area along the major thrusts and the Mikir westward), meander migration is more frequent in the southern tribu- hills is much more active than the Naga Patkai thrust as reflected from taries. We argue that this is a manifestation of different tectonic regimes the distribution of epicentres of recent moderate earthquakes (M>5) of these tributaries. (Fig. 9). The presence of blind faults (interpreted from the deep seismic Several workers have shown that unidirectional avulsions occur data, see Supplementary Figs. I–III) within the valley suggests that, though owing to cross-valley tilting in areas of high tilt rates (Alexander and the thrust planes at the valley margin are the major zones of stress re- Leeder, 1987; Holbrook and Schumm, 1999). Apart from active tectonic lease, stress is also partially transmitted to the valley. As a result, coseis- movements along faults, such down tilt lateral migration can be caused mic subsidence as well as interseismic uplift in the valley (Burbank and by subsidence where the down tilt direction is generally the site of mini- Anderson, 2012) will keep on uplifting its central part and generating mum elevation (Holbrook and Schumm, 1999). As discussed above, the accommodation space in the depressions. Brahmaputra valley in the study area has several large depressions that Fig. 10 presents a morphotectonic model for the study area sum- provide the accommodation space of the sediment accumulation, and marising the coseismic and interseismic forcings influencing the mor- the continued sediment loading in these depressions causes further sub- phodynamics of the valley. Fig. 10A shows the pre-1950 earthquake sidence and reorganisation of the hinterland–basin relationship. Most scenario depicting the highly dynamic and aggradational regime in of the northern tributaries of the Brahmaputra River flow through the the upper two units. Frequently avulsing tributaries along the north Subansiri depression, and we argue that sediment loading in the Subansiri bank in contrast to the older meander bends of the south bank tribu- depression might have resulted in a southwestward tilt of the adjoining taries reflect the influence of two different tectonic domains, namely parts of the Himalayan foreland, which in turn caused unidirectional avul- the Himalayan thrust belt and the Naga Patkai thrusts. Fig. 10Bshows sion of the tributaries of the north bank (Figs. 6Aand10C). We also note the morphodynamic changes caused by the 1950 earthquake (M 8.7) that the avulsive shifts in these rivers diminish as we move northeast that resulted in coseismic subsidence in the Subansiri and Lohit depres- away from the Subansiri depression. sions and generated accommodation space. Fig. 10Cshowsthepost- In contrast, the southern tributaries are primarily characterised by 1950 earthquake (present-day) scenario. Aggradation in the Lohit de- slow meander-scale migration (Fig. 7) and subtle manifestation in pression tremendously increased channel instability and lateral erosion planform parameters that are quite sensitive to slope changes. Several in unit 1. The site of effective aggradation shows a switch over from unit workers have suggested that if the slope increases, the channel responds 1 to unit 3. Subsidence in the Subansiri depression continued the north

A) Pre -1950

B) 1950 (Coseismic) Coseismic subsidence of the Lohit depression

Coseismic subsidence of the Subansiri depression

Dramatic increase in lateral erosion and planform expansion C) Post-1950 onward of the channel belt (interseismic)

Shift in the site of effective aggradation from unit 1 to unit 3

Interseismic upper central uplift continues

Interseismic uplift of the Mikir hills influences Dhansiri depression and it is free from subsidence as well as generation of new accommodation space

Fig. 10. Models (not to scale) summarising the coseismic and interseismic forcings inﬂuencing the morphodynamics of the valley. (A) Pre-1950 earthquake scenario. (B) 1950 earthquake (M 8.7) caused subsidence in the Subansiri and Lohit depressions which helped to generate more accommodation space. (C) Post-1950 earthquake to the present-day scenario. Aggra- dation in the Lohit depression decreased the bedload/suspended load ratio and increased lateral erosion tremendously in unit 1. Unit 2 does not show much change in braid bar/channel area ratio. The site of effective aggradation shows a switch over from unit 1 to unit 3. Subsidence in the Subansiri depression caused a number of channels of the north bank to avulse in the southwest direction. The south bank tributaries remained mostly unaffected.

bank tributaries to avulse in the southwest direction. The south bank • In a 240-km-long stretch of the Brahmaputra, the longitudinal profile tributaries remained mostly unaffected during the entire period. shows several ‘highs’ and ‘lows’ that have led to uneven sediment dispersal resulting in reach-scale aggradation and degradation. 6. Conclusions • The general trend of avulsion of the north bank tributaries of the Brahmaputra River is towards the southwest, and a definite struc- • The Brahmaputra valley in upper Assam is strongly influenced by the tural control of very recent origin is inferred. The palaeochannels Himalayan and Naga Patkai belt orogeny and can be broadly subdivided of the southern tributaries on the other hand are much older and into three types of first-order tectonogeomorphic units, namely the cen- suggest stability at shorter timescales. These differences in dynamics tral uplift,theslopes,andthedepressions. The spatial variability in mor- of the northern and southern tributaries are in response to different tec- phodynamics of the Brahmaputra River is strongly linked to these units. tonic regimes of the Himalayan and Naga Patkai thrusts, respectively.

Supplementary data to this article can be found online at http:// Hovius, N., 1998. Controls on sediment supply by larger rivers. In: Shanley, K.W., McCabe, P.J. (Eds.), Relative Role of Eustasy, Climate, and Tectonism in Continental dx.doi.org/10.1016/j.geomorph.2012.04.012. Rocks: Society for Sedimentary Geology (SEPM) Special Publication, 59, pp. 3–16. Kent, W.N., Hickman, R.G., Das Gupta, U., 2002. Application of a ramp/flat-fault model Acknowledgements to interpretation of the Naga thrust and possible implications for petroleum exploration along the Naga thrust front. American Association of Petroleum Geologists Bulletin 86 (12), 2023–2045. The authors are highly thankful to the reviewers for their critical Khattri, K.N., 1987. Great earthquakes, seismicity gaps and potential for Earthquake comments and minute observations. It helped us to rethink and reor- disaster along the Himalayan plate boundary. Tectonophysics 38, 79–92. ganise some of the crucial issues of the earlier draft and redefine the Kotoky, P., Bezbaruah, D., Baruah, J., Sarma, J.N., 2005. Nature of bank erosion along the Brahmaputra River channel, Assam, India. Current Science 88 (4), 634–640. focus of the paper. We are thankful to IIT Kanpur for providing the in- Lane, E.W., 1957. A study of the shape of channels formed by natural streams flowing in stitutional support to conduct this study. We acknowledge the service erodible material. Missouri River Division Sediment Series No. 9. U.S. Army Engineer provided by the USGS website for the DEM data from the SRTM Division, Missouri River, Corps of Engineers, Omaha, Nebraska. fi Latrubesse, E., 2008. Patterns of anabranching channels: the ultimate end-member source. We are also thankful to the India Of ce Library and Records, adjustment of mega rivers. Geomorphology 101, 130–145. London, UK, for providing the topographic map of the study area pre- Leopold, L.B., Wolman, M.G., 1957. River channel patterns: braided, meandering and pared during the 1912–1926 seasons. straight. US Geological Survey Professional Papers, 262B, pp. 39–85. Marchis, M.De., Napoli, E., 2008. The effect of geometrical parameters on the discharge capacity of meandering compound channels. Advances in Water Resources 31, References 1662–1673. Mukhopadhyay, M., 1984. Seismotectonics of transverse lineaments in the Eastern Alexander, J.A., Leeder, M.R., 1987. Active tectonic control on alluvial architecture. In: Himalaya and its foredeep. Tectonophysics 109, 227–240. Ethridge, F.G., Flores, R.M., Harvey, M.D. (Eds.), Recent Developments in Fluvial Murty, K.N., 1983. Geology and hydrocarbon prospects of Assam Shelf. Recent advances Sedimentology. : Special Publication, 39. Society of Economic Paleontologists and and present status. Petroleum Asia Journal 6 (4), 1–14. Mineralogists, Tulsa, OK, pp. 243–252. Nandy, D.R., Das Gupta, S., 1991. Seismotectonic domain of northeastern India and ad- Allen, J.R.L., 1965. A review of the origin and characteristics of recent alluvial sedi- jacent areas. Physics and Chemistry of the Earth 18, 371–384. ments. Sedimentology 5, 89–191. Narula, P.L., Acharyya, S.K., Banerjee, J., 2000. Seismotectonic Atlas of India and its Bally, A.W., 1997. Hydrocarbon potential of Assam-Arakan and in India's Tertiary Environs. Special Publication, 59. Geological Survey of India, pp. 1–40. folded belt and foreland basins – a ranking of exploration opportunities, Unpub- Ranga Rao, A., 1983. Geology and hydrocarbon potential of a part of Assam-Arakan lished report Oil & Natural Gas Corporation Limited, India, 1–109. basin and its adjacent region. Petroleum Asia Journal 1–14. Bristow, C.S., 1987. Brahmaputra River: channel migration and deposition. In: Ethridge, F.G., Richards, K., Chandra, S., Friend, P., 1993. Avulsive channel systems: characteristics and Flores, R.M., Harvey, M.D. (Eds.), Recent Developments in Fluvial Sedimentology. examples. In: Best, J.L., Bristow, J.L. (Eds.), Braided Rivers. : Special Publication, 75. Special Publication, 39. Society of Economic Paleontologists & Mineralogists, Tulsa, Geological Society of London, London, pp. 195–203. OK, pp. 63–74. Richardson, W.R., Thorne, C.R., 2001. Multiple thread flow and channel bifurcation in a Burbank, D.W., Anderson, R.S., 2012. Tectonic Geomorphology, 2nd ed. Wiley-Blackwell, braided river: Brahmaputra–Jamuna River, Bangladesh. Geomorphology 38, West Sussex, UK. 185–196. Coleman, J.M., 1969. Brahmaputra River channel processes and sedimentation. Sedi- Sarma, J.N., 2005. Fluvial process and morphology of the Brahmaputra River in Assam, mentary Geology 3, 129–239. India. In: Latrubesse, E.M., Stevaux, J.C., Sinha, R. (Eds.), Tropical Rivers: Geomor- Curray, J.R., 1994. Sediment volume and mass beneath the Bay of Bengal. Earth and phology, Special issue, 70, pp. 226–256. Planetary Science Letters 125, 371–383. Sarma, J.N., Phukan, M.K., 2004. Origin and some geomorphological changes of Majuli Das Gupta, A.B., Biswas, A.K., 2000. Geology of Assam. Geological Society of India, Bangalore, Island of the Brahmaputra River in Assam, India. Geomorphology 60, 1–19. pp. 45–83. Sarma, J.N., Phukan, M.K., 2006. Bank erosion and bankline migration of the river Brahma- Friend, P.F., Sinha, R., 1993. Braiding and meandering parameters. In: Best, J.L., Bristow, putra in Assam, India, during the twentieth century. Journal of Geological Society of C.S. (Eds.), Braided Rivers. : Special Publication, 75. Geological Society of London, India 68, 1023–1036. London, pp. 105–111. Schumm, S.A., Dumont, J.E., Holbrook, J.M., 2000. Active Tectonics and Alluvial Rivers. Goodbred Jr., S.L., Kuehl, S.A., 1998. Floodplain processes in the Bengal Basin and the Cambridge University Press, pp. 21–42. storage of Ganges–Brahmaputra river sediment: an accretion study using 137Cs Singh, S.K., 2006. Spatial variability in erosion in the Brahmaputra basin: causes and and 210Pb geochronology. Sedimentary Geology 121, 239–258. impacts. Current Science 90, 1272–1276. Goodbred Jr., S.L., Kuehl, S.A., 2000. The significance of large sediment supply, active Singh, S.K., Kumar, A., Lanord, C.F., 2006. Sr and 87Sr/86Sr in waters and sediments of

tectonism, and eustasy on margin sequence development: Late Quaternary stratig- the Brahmaputra river system: silicate weathering, CO2 consumption and Sr flux. raphy and evolution of the Ganges–Brahmaputra delta. Sedimentary Geology 133, Chemical Geology 234, 308–320. 227–248. Sinha, R., 1996. Channel avulsion and floodplain structure in the Gandak-Kosi interfan, Goodbred Jr., S.L., Kuehl, S.A., Steckler, M.S., Sarker, M.H., 2003. Controls on facies distribu- north Bihar plains, India. Zeitschrift für Geomorphologie, N.F. Supplementband tion and stratigraphic preservation in the Ganges–Brahmaputra delta sequence. 103, 249–268. Sedimentary Geology 155, 301–316. Sinha, R., Friend, P.F., 1994. River systems and their sediment influx, Indo-Gangetic Goswami, D.C., 1985. Brahmaputra River, Assam, India: physiography, basin denunda- plains, Northern Bihar, India. Sedimentology 41, 825–845. tion and channel aggradation. Water Resources Research 21, 959–978. Tandon, A.N., 1954. Study of the great Assam earthquake of August 1950 and its after- Goswami, P.C., Goswami, P., 2007. Structural style & timing of structural deformation in shocks. Indian Journal of Meteorology and Geophysics 5, 95–137. Naga foothills region—a discussion. Association of Petroleum Geologists, India, Bulletin Tandon, S.K., Sinha, R., 2007. Geology of large rivers. In: Gupta, A. (Ed.), Large Rivers: 153–160. Geomorphology and Management. John Wiley & Sons, pp. 7–28. Holbrook, J., Schumm, S.A., 1999. Geomorphic and sedimentary response of rivers to tectonic deformation: a brief review and critique of a tool for recognizing subtle epeirogenic deformation in modern and ancient settings. Tectonophysics 305, 287–306.