RS-GIS Based Assessment of River Dynamics of Brahmaputra River in India ()
1. Introduction
The river Brahmaputra has been the lifeline of northeastern India since ages. This mighty river runs for 2880 kms through China, India and Bangladesh. Any alluvial river of such magnitude has problems of sediment erosion-deposition attached with it; the Brahmaputra is no exception. The problems of flood, erosion and drainage congestion in the Brahmaputra basin are gigantic. The Brahmaputra River is characterized by its exceedingly large flow, enormous volume of sediment load, continuous changes in channel morphology, rapid bed aggradations and bank line recession and erosion. The river has braided channel in most of its course in the alluvial plains of Assam. The lateral changes in channels cause severe erosion along the banks leading to a considerable loss of good fertile land each year. Bank oscillation also causes shifting of outfalls of its tributaries bringing newer areas under waters. Thousands of hectares of agricultural land is suffering from severe erosion continuously in the Brahmaputra basin covering parts of states like Assam, Arunachal Pradesh, Meghalaya, Nagaland and Manipur.
Stream-bank erosion and its effects on channel evolution are essential geomorphic research problems with relevance to many scientific and engineering fields. Remote sensing satellite data having ability to provide comprehensive, synoptic view of fairly large area at regular interval with quick turn-around time integrated with Geographical Information System (GIS) techniques makes it appropriate and ideal for studying and monitoring river erosion and its bank line shifting. Various studies in this regard have been carried out for some major rivers all over the world [1-6]. Several investigators have used remotely sensed data for ascertaining channel changes of Brahmaputra River and its tributaries. NRSA has done the river migration study of the Brahmaputra using airborne scanner survey and has carried out repetitive survey to monitor changes in landuse, river channels and banks to provide a base for estimating the response of the rivers to flood events [7]. Bardhan [8] studied the channel behavior of the Barak River using satellite imagery and other data to identify the river stretches, if any, which remained reasonably stable during the period 1910-1988.
SAC (Space Application Centre), Ahmedabad and Brahmaputra Board jointly took up a study to access the extent of river erosion in Majuli island in order to identify and delineate the areas of the island which have undergone changes along the bank line due to dynamic behavior of the river [9]. Naik [10] studied the erosion at Kaziranga National Park using remote sensing data. Goswami [11] carried out a study on river channel changes of the Subansiri (northern tributary of Brahmaputra River) in Assam, India. Mani [12] studied the erosion in Majuli Island using remote sensing data. Bhakal [13] have quantified the extent of bank erosion in Brahmaputra River near Agyathuri in Assam, India over a period of thirty years (1973-2003) using remote sensing data integrated with GIS. Kotoky [14] studied selected reach of Brahamputra (from Panidihing Reserve Forest to Holoukonda Bil of the Brahmaputra River) with two sets of Survey of India toposheets (1914 and 1975) and a set of IRS satellite images (1998, IRS-1B, LISS II B/W geocoded data. Sarma [15] studied the nature of bank line migration of Burhi Dihing River (southern tributary of Brahmaputra River) course during a period of time from 1934 to 2004. Das and Saraf [16] made a study in respect to a trend in river course changes of Brahmaputra River and influence of various surrounding geotectonic features for varying period between 1970-2002 for different sections of the river using Landsat MSS, TM and ETM images. However, a comprehensive study of the bank erosion and channel migration of the entire Brahmaputra in India with most recent satellite data has not yet been reported in the literature.
The present paper briefly describes a study of the Brahmaputra river—its entire course in Assam from upstream of Dibrugarh up to the town Dhubri near Bangladesh border for a stretch of around 620 kms for a period of 18 years (1990-2008) using an integrated approach of Remote Sensing and GIS. The satellite data has provided the information on the channel configuration of the river system on repetitive basis revealing much needed data on the changes in river morphology, erosion/deposition pattern and its influence on the land, stable and unstable reaches of the river banks, changes in the main channel of the Brahmaputra River, etc. It is envisaged that the conjunctive use of the information generated out of this study with other ground based data will contribute substantially in more meaningful approach towards planning and execution of means and measures to combat recurring floods and erosion.
2. Study Area
The Brahmaputra River, termed a moving ocean, is an antecedent snowfed river which flows across the rising young Himalayan Range. Geologically, the Brahmaputra is the youngest of the major rivers of the world. It originnates at an altitude of 5300 m about 63 km south-east of the Mansarovar lake in Tibet. The river is known as Tsangpo in Tibet. Flowing eastward for 1625 km over the Tibetan plateau, the Tsangpo enters a deep narrow gorge at Pe (3500 m) and continues southward across the east-west trending ranges of the Himalayas, viz. the Greater Himalayas, Middle Himalayas and sub-Himalayas. After crossing the Indo-China border near Pasighat the river is called as the Siang or the Dihang. Two major rivers namely the Dibang and the Lohit join the Dihang at a short distance upstream of Kobo to form the river Brahmaputra. The river flows westward through Assam for about 700 km distance from Dhola until downstream of the town Dhubri, where it abruptly turns south and enters Bangladesh. The gradient of the Brahmaputra River is as steep as 4.3 to 16.8 m/km in the gorge section upstream of Pasighat, but near Guwahati it is as flat as 0.1 m/km. In the course of its 2880 km journey, the Brahmaputra receives as many as twenty two major tributaries in Tibet, thirty three in India and three in Bangladesh. The northern and southern tributaries differ considerably in their hydro-geomorphologic characteristics owing to different geological, physiographic and climatic conditions. Due to the colliding Eurasian (Chinese) and Indian tectonic plates, the Brahmaputra valley and its adjoining hill ranges are seismically very unstable.
For the present study, a reach of 620 km on the main stem of Brahmaputra River, i.e., its entire course in Assam from upstream of Dibrugarh up to the town Dhubri near Bangladesh border has been considered. The geographical bounds of the area are 25˚45' to 28˚N latitude and 90˚ to 96˚30'E longitude. Figure 1 shows the study reach of the river.
3. Material and Method
3.1. Data
The basic data used in this study are digital satellite images of Indian Remote Sensing (IRS) LISS-I (Path/Row 13/49; 14/47,48,49; 15/47,48; 16/47,48) and LISS-III (Path/Row 110/51,52; 111/51,52; 112/51,52; 113/51,52; 114/53) sensor, comprising of scenes for the years 1990 and 2008. The other collateral data used in the present study are Survey of India toposheets nos 78/J, K, N, O; 82/L, P; 83/B, E, F, I, J, M at 1:250,000 scale and Landsat ETM images (Path/Row 135/41, 42; 136/42; 137/42; 138/42;), which are freely downloadable from the website www.glcf.umiacs.umd.edu/data/landsat.
ERDAS Imagine 9.3 image processing software has been used for processing the satellite images. ArcMap 9.3 GIS software has been used for analysis of the river bank data.
3.2. Methodology
Satellite images for years 1990 and 2008 have been processed to obtain the required information about the study area. The following section describes the preparation of satellite images for information extraction.
3.2.1. Geo-Referencing and Digital Processing of Satellite Data
The satellite images procured from the NDC (NRSC Data Center), the official data supplier for IRS satellite images in India, do not have the global coordinate and projection systems. Therefore, the data need to be georeferenced before commencing the analysis. Since the topographic maps procured from Survey of India belong to year 1965, there are increased chances of error, if the toposheet data is taken as a reference for geo-referencing the satellite images. Therefore, Landsat ETM images have been taken as reference. Toposheets have been used for assistance in identification of features. The Landsat images are precisely ortho-rectified with the following parameters:
Projection Type: Universal Transverse Mercator (UTM)
Spheroid Name: WGS 84 Datum Name: WGS 84 Zone: 46-47 Landsat ETM images for the study area have been downloaded and a mosaic has been prepared. Then individual image of IRS has been registered with respect to the Landsat image. For each image an average of 18 GCPs (ground control points) have been taken and georeferencing has been carried out using second-order polynomial with nearest neighborhood resampling keeping grid size as 23 m. RMS (root mean square) error has been kept below 0.8 pixel. Survey of India toposheets have been utilized for accurately identifying the GCPs on IRS and Landsat images.
The geo-referenced images of the same year have been mosaiced together. Brahmaputra River is covered with clouds for most part of the year, therefore the satellite images of one year couldn’t be procured for same date or month. There is a time gap of about 2 to 4 months in the satellite images of one year. Therefore, radiometry of the images differ a lot. Making a mosaic without balancing the radiometry will not give good results. Therefore, radiometric normalization has been performed before mosaicing by histogram equalization and matching. This has given very good results, providing uniform brightness levels for similar features for the entire river basin. Also, feather option has been used for mosaicing to get the seamless boundaries between different images of same year. The mosaics for the years 1990 and 2008 have been prepared in this manner.
3.2.2. Preparation of NDWI Image
NDWI (Normalized Difference Water Index) shows the moisture content in soil and vegetation. NDWI image for IRS LISS-III data has been prepared using the following formula [17]
Since the IRS LISS-I doesn’t have SWIR band, therefore its NDWI image can be created using the following formula [18]
NDWI is useful to assess water content in a normalised way. This index increases with vegetation water content or from dry soil to free water. The SWIR reflectance indicates changes in soil and vegetation water content, while the NIR reflectance is affected by internal structure of leaf and soil but not by water content directly. The combination of the NIR with the SWIR removes variations induced by internal structure of leaf and soil, thus improving the accuracy in retrieving the vegetation and soil water content [19].
3.2.3. Delineation/Digitization of River Bank Line
The entire river from Dhubri to upper Assam beyond Dibrugarh has been divided into 120 strips at an equal spacing of approximately 5 km, reference cross sections have been drawn at the boundary of each strip. Each ten cross sections have been grouped as a reach with numbering from downstream to upstream of the river (of equal base length). Base line of latitude 25.966˚N and longitude 90˚E has been taken as permanent reference line, so that they maintain their identity when the morphology is changed. Table 1 shows some important locations within various reaches of the river.
The river bank-line has been identified and delineated for all the satellite image mosaics of 1990 and 2008. NDWI image proved very useful in identifying the river bank line. While the shallow water channels have been considered the part of river, old and new soil/sand deposits at the river banks posed some ambiguity of interpretation as the river. A few of these soil/sand patches are at considerable distance from active water channel, but have dark tone on satellite image, indicating higher moisture. On the other hand, several soil patches very close to the active water channel bear bright signature, an indication of low moisture. NDWI has been used in delineating the soil/sand areas which have higher moisture content. It has been observed that the areas with recent soil deposits have higher moisture in comparison to other areas adjacent to river bank, due to the river related activities associated with them. Either the river was flowing through that area in the recent past, or that area was submerged in water, when there was high flow in the river. These areas have been marked as a part of river. Thus the use of NDWI image helped in delineating/demarcating the river bank line.
The identified river bank lines for the left (south) and right (north) banks of the river, have been digitized using ArcMap software. Two river bank lines have been prepared for the years 1990 and 2008. The length of arcs of both the left and right banks for all the above years have been calculated using GIS software. Erosion and deposition area has been estimated through area estimation using GIS software tools for polygon areas with the shifting bank-lines in study period.
4. Results and Discussion
Mosaic images of year 1990 and 2008 with digitised bank lines and reference cross-sections drawn are presented in Figures 2(a) and (b). Table 2 accounts for the reach wise erosion and deposition area for the period 1990-
Table 1. Identification of reaches in respect of the location in the vicinity.
(a) (b)
Figure 2. Satellite image of entire Brahmaputra river from U/s Dibrugarh to Dhubri (a) 1990; (b) 2008.
2008.
It is seen from Table 2 that there has been significant erosion in the river during a period of 18 years as compared to deposition. Erosion is more pronounced in the southern bank and overall rate of erosion and deposition work out to be 62.7 km2/year and 12.6 km2/year respecttively. The erosion rate has been calculated after excludeing forest area of southern bank in Reach 12.
Figure 3 depicts the river bank lines in the year 1990 and 2008. The integration of river bank lines of year 1990 and 2008 revealed that the Brahmaputra River has shifted its bank line drastically causing severe damage to the good agriculture as well as habitat area on its both sides. The erosion sites are located throughout its reach between U/s of Dibrugarh to Dhubri within India. Figure 4 depicts the erosion and deposition in study reach of the river for a period of 18 years. Figure 5 gives a graphical plot of reach wise total erosion/deposition in both banks of the river. It is evident from this graph that three to four major geological control points are present in the Brahamputra in Assam flood plains which are holding the river. These geological control points are in the vicinity of Jogighopa near Goalpara, Pandu near Guwahati, Tezpur and Bessemora near Majuli. Each control point usually has well defined hydrographic profile. Within geologically controlled reaches, intermittent fanning out and fanning in are exhibited which are being temporally severed in the geological increasing time scale.
Table 2. Erosion/deposition area along Brahmaputra for the period 1990 to 2008.
Figure 3. River bank lines in discrete years.
Figure 4. Erosion & deposition (1990-2008).
Figure 5. Total Erosion & Deposition over eighteen years (1990-2008) in various reaches of the Brahmaputra River.
Figure 6 depicts erosion and deposition in all the twelve reaches of the river. In reach-1 (in the vicinity of Dhubri) the river takes a southwesterly turn which is mainly due to the existence of a highland in the northern part of the turn which has prevented further westward flow of the river. This reach exhibits huge erosion in both the banks with more north-westward expansion. In reach- 2 and reach-3 near Goalpara and Palasbari respectively, the Brahmaputra River flows along the northern edge of the Shillong Plateau and through numerous inselbergs (granitic and gneissic hills) belonging to an exposed part of the Shillong Massif [16]. In these two reaches, right bank (northern) shows considerable shifting by following a preferred course closer to the northern edge of the Shillong Plateau with narrowing down to incised channel at Pandu near Guwahati. This change has probably occurred