A landslide is the movement of a mass of rock or debris or earth down a slope, under the influence of gravity (Cruden and Varnes, 1996), and it is one of the major geo-hazards in the hilly terrains of North Eastern Region (NER) of India. The NER, which is buttressed between the Himalayan arc collision zone to the north and the Indo-Burma arc subduction zone to the east, experiences large/great earthquakes and high rain fall. Thus, landslides and other slope failures occur in these mountainous terrains due to combined effect of several geological factors aided/triggered by rainfall, earthquakes as well as by anthropogenic activities. It causes loss of lives and properties almost every year apart from disruption of communication links.

The NER consists of eight states, namely, Arunachal Pradesh, Assam, Manipur, Meghalaya, Mizoram, Nagaland, Sikkim and Tripura. Almost entire region is susceptible to landslides of various degrees due to terrain condition. The plain areas of the Brahmaputra and Barak Valleys of Assam and valley portions of Manipur and Tripura are also susceptible for landslides at places. The region has an area of 262,179sq km (NEC, 2015), from valley to difficult mountainous terrain with elevation as high as 6000m above mean sea level (MSL) in Arunachal Pradesh. The Patkai-Naga-Lusai range of Nagaland and Manipur extends down to the Arakan-Yoma belt of Myanmar. The Assam state has lower hill cover area about 24 per cent of the landmass is hilly (GoI, 2013). Mizoram and Tripura states com-prise alternate ridge and valley topography, and Meghalaya, on the other hand, is a plateau region of an average elevation 1000m. The whole region has dis-tinctive rocks of different geological ages form Archaean to Quaternary (Bagchi et al., 2020).

An attempt is made through this study, to examine and assess the landslide scenario of NER and to understand the challenges it poses to the society and on developmental activities. NER is affected by all types of landslides in varying size and magnitude, and is at the top with respect to Developmental Disability Index (DDI) derived by Government of India (GoI, 2013). More than 50% of the national highways in the region are affected by frequent landslides, especially during monsoon season. We studied different aspects of landslides in NER and highlight the results here. 

Various organizations and academic institutions carry out landslide investigation and susceptibility-hazard-risk mapping by conventional and or by innovative techniques for awareness to minimize its affect to the society. The Earth Observation (EO) data of various spatial, spectral and temporal resolutions for different aspects of landslides are studied by many researchers worldwide (Guzzetti et al., 2009; Gorum et al., 2011 and many more). High resolution imagery (QuickBird, IKONOS, CARTOSAT-1 & 2) is one of the best options now for landslide mapping, and various operational sensors are also developed (Westen et al., 2007). The importance and limitations of optical (both mono/ stereoscopic and PAN/multispectral), SAR, LiDAR remote sensing data with varied spatio-temporal resolutions for landslide mapping are elaborately explained by Guzzetti et al. (2012).  In India use of geo-informatics is evaluated by many experts (Lakhera, 1982; Champati ray, 1996, 2004; Sarkar et. al., 1995, 2004; Martha et al., 2010). In this study, we mostly analyzed the EO data coupled with some field checks either on operational mode or on user demand. The EO data with varying spatio-temporal resolutions, e.g. Cartosat, LISS IV MX, LISS III MX, IKONOS, QuickBird and Google Earth images, are used. The EO data have an advantage due to its synoptic coverage which provides regional appreciation and mapping, monitoring of dynamic changes of the terrain, survey and its evaluation especially in an inaccessible terrain. The mountainous terrain of the NER being inaccessible, it is difficult to survey by conventional method alone; remote sensing data, the only means of survey, are critically analyzed in this study.

2. 1 Geo-environmental set up of NER

2.1.1 Physiography

A major part of the NER belongs to Extra-Peninsula of the mighty Himalayan ranges and its extensions with folded and faulted mountain chains (Fig. 1). Broadly the region can be divided into following four groups:

2.1.1.1 Mountainous areas

Arunachal and Sikkim are part of Himalayan ranges with an average elevation of 6400 m in the greater Himalayas. The ranges follow an E-W trend in the western part and gradually swings to NNE-SSW in the northeast till the syntaxial bend, where Arunachal trends NW-SE (Fig. 1).

2.2 Valley areas

The Brahmaputra (Assam) valley is narrow and elongated along the gigantic Bramhapura River, about 660 km long and 90 km wide zone (Fig. 1). In the western part, the valley is dotted with numerous inselbergs. As the river Brahmaputra is braided in nature, associated geomorphic features are present all along the valley. 

The Barak valley in Assam is basically triangular-shaped along the meandering river Barak which has originated from Manipur. The valley is transected by north-south trending low-altitude linear hillocks as well as with residual hills at places. The Imphal valley in Manipur is lacustrine in nature, while the plain areas in Tripura show both erosional and depositional features.

2.3 Plateau areas 

Almost the entire state of Meghalaya is a plateau with an average elevation of 1000 m, except the foothills of Garo and Jaintia Hills. The highest point (1961 m) of the plateau is called Shillong peak. The Mikir hills in Assam is a fragmented part of Meghalaya (also called Shillong plateau). The Shillong-Mikir plateau is a fragmented part of Indian shield (Evans, 1964; Nandy, 2001).

2.4 Indo-Burma Ranges

The Indo-Burma Ranges (IBR), which extends from Eastern Himalayan Syntaxis (EHS) southwards along the eastern side of the Bay of Bengal to the Andaman Sea, embrace the states of Nagaland, Manipur, Mizo-ram, Tripura and southern Assam within the Indian territory (Fig. 1). In the southern part of NER, the range is N-S trending while gradually swings to NNE-SSW in the northern part in Nagaland. The average height of hill ranges in the southern part is about 900 m above MSL, whereas the general elevation increases towards east with the highest peak Saramati (3841m), located at the eastern margin of Nagaland.

2.5 Geology

The NER represents complete stratigraphic sequences ranging from pre-Cambrian to Recent showing conformity or unconformity relationships between the formations. The oldest geological formation in the reg-ion is pre-Cambrian gneissic complex of the Shillong- Mikir plateau. On the other hand, the Himalayas, con-sist of formations ranging from Proterozoic to early Paleozoic, and vary from low-grade metamorphics in the southern sections to high-grade schists to the crest. The Himalayan foothills zone contains Tertiary rocks of Mio-Pliocene deposits of post-orogenic phase. The rest of the NER is formed by Tertiary rocks with diff-erent marine facies, ranging from Eocene to Pliocene deposited either on shelf or geosynclinal basin-system.

2.6 Seismotectonics

Seismically the NER is one of the most active regions in the world due to collision tectonics in the Himalayas to the north and seductions tectonics in the Indo-Burma zone to the east. The Indian plate gently dips below the Himalayas, and atypically subducts below the Indo-Burma Ranges that causes high seismicity forming several seismotectonic domains, namely the Eastern Himalaya collision zone, Eastern Himalayan Syntaxis (EHS) and the Indo-Burma zones at the plate boundaries, and the Meghalaya, Assam valley and Bengal basin in the intra-plate region (Fig. 1). The EHS is a much complex zone where the Himalayan arc and the Indo-Burma arc meet (Fig. 1). The entire NER is marked as the highest risk zone (Zone V) in the seismic zoning map of India (BIS, 2002). In the collision zone, the Eastern Himalayas produced the 1934 great Bihar-Nepal earthquake (revised Mw 8.2) and several large and strong earthquakes including the recent 2009 Bhutan (Mw 6.3) and the 2011 Sikkim (Mw 6.9) earthquakes (Fig. 1). The 1934 great earthquake, however, falls farther west, out-side of our study region. The EHS produced the 1950 Assam-Tibet great earthquake (Ms 8.7) that caused some 15,200 casualties and huge damages to properties (Poddar, 1950). Magnitude of this earthquake is, however, revised to Mw 8.4 (Ambraseys and Douglas, 2004).

The Indo-Burma subduction zone is seismically most active and marked with several large earthquakes (M 7.0-7.6) (Fig. 1), (Kayal, 2008). In the intra-plate zone, the Meghalaya (Shillong) plateau produced the 1897 great earthquake Ms 8.6 with estimated ground acceleration of 1g and some 1500 casualties  and huge property destruction (Oldham, 1899). Bilham and England, (2001) revised its magnitude to Mw 8.1 and proposed a pop-up tectonics model of the Shillong plateau between the Oldham fault and Dauki fault for this great earthquake. The Oldham fault is proposed based on GPS data, that separates the Brahmaputra valley and Shillong plateau, and the gigantic E-W trending Dauki fault separates the Shillong plateau and the Bengal basin (Bangladesh) (Fig. 1). The intra-plate Assam valley, on the other hand, produced two large earthquakes, the 1869 (revised Mw 7.4) and the 1943 (revised Mw 7.3), by transverse tectonics along the Kopili fault (Kayal, 2008).

The revised magnitudes of the earthquakes mentioned here are reported by Ambraseys and Douglas, (2004). In addition to these, there are frequent felt earthquakes (Mw 5.0-6.0) with severe shaking in the region (Kayal et al., 2012; Hazarika and Kayal, 2021). These earthquakes play a major role for landslides in the NER.

2.7 Climate

The NER displays the character of tropical climate, especially in the valley areas. The region experiences a prolonged monsoon with heavy to very heavy rains, from June to September with light to heavy pre-monsoon shower in April and May. The southwest monsoon is the main source of rain; about 90 % of the total rain is received during this period and June is the rainiest month (Dikshit and Dikshit, 2014).There is a climatic contrast between the valley and the mountainous areas. The mean temperature in January in Assam valley is around 16°C, the Arunachal and Nagaland mountainous areas fluctuates between 14°C and a sub-zero temperature. During summer, in April-May temperaturesin the plains vary between 30°and 33°C, while in hills it is of around 20°C with a mean minimum of 15 °C. The NER is the rainiest part of the country and receives much higher rain than the average of 1,000 mm for the whole country (Dikshit and Dikshit, 2014). The average rainfall of Brahmaputra valley is around 2,000 mm with local variations in the rain shadow zones, while Cherrapunji in Meghalaya receives a mean annual rainfall of 11,465 mm.On an average, the hilly areas of the region receive 2,000–3,000 mm of rain yearly.The regional climate in NER is grouped into 3 Koeppen classes, i.e. ‘A, ‘C and ‘D.

Fig. 1: Geo-environmental set up of the NER, India.

The SRTM elevation data 3 Arc Second (90 m spatial resolution). Base layers taken from NESDR, NESAC database. Major Rivers of the World taken from https://datacatalog.worldbank.org/search/dataset/0042032. Epicentres of the great, large and strong earthquakes (ISC Catalog) are shown in the map. Two great earthquakes (Mw > 8.0) are shown by largest red stars, large earthquakes (Mw > 7.0) by red circles and strong earthquakes (Mw 6.0 - < 7.0) by smaller red stars. Major faults and tectonic features are drawn from EO data taking reference from the published Seismo-tectonic Atlas (GSI. 2000). These are: MCT: Main Central Thrust, MBT: Main Boundary Thrust, HFT: Himalayan Frontal Thrust, ITSZ: Indus Tsangpo Suture Zone, MT: Mishmi Trust, LT: Lohit Thrust, EHS: Eastern Himalaya Syntaxis, JF: Jiali Fault NT, YF: Yemla Fault, CT: Canyon Thrust, NT: Naga Thrust, Dis T: Disang Thrust, CMF : Churachandpur Mao Fault, MF : Mat Fault, Syl F : Sylhet Fault, Kop F: Kopili Fault, OlF: Oldham Fault, BSZ : Barapani Shear Zone, DT : Dapsi Thrust, DF: Dauki Fault, Dhb FDhubri Fault etc.  Inset: Key map, rectangle indicates the study region; prominent reference features are:  BR: Brahmaputra River, BkR: Barak River, JH: Jaintia Hills, GH: Garo Hills etc. 

2.8 Landslides in NER

As described above, based on the geo-environment of different areas in NER, the landslide problems are analyzed area wise to assess the risk zones. These assessments, based on the EO, field surveys and available published data, are presented below.  

2.8.1 Mountainous areas  

Landslides are major problems for two Himalayan states, Arunachal Pradesh and Sikkim. The 2,500 km long Himalayan collision zone is not uniform in seismic pattern or in tectonic fabrics (Kayal, 2001 and 2010), so with the geology and slope conditions (Val-diya, 1987). A good correlation of landslide distribution with slope is observed in Arunachal.  Majority of landslides occur below the altitude of 2000 m; maximum landslides occur at altitudes between 500 and 1000 m. This concentration of landslides in lower altitude belt may be attributed to geologically younger and geomorphologically unstable state of the frontal Arunachal Himalayas. Additionally, frequency of landslides peaks at slope value of 40° and above.  This may be due to the presence of hard rocks at slope <40°areas which remain stable if undisturbed by catastrophic events like earthquakes, rain etc. In analysis of pre- and postmonsoon data of LISS IV MX, image of 5.8 m resolution, it is observed that about 3542 landslides occurred in Arunachal in the year 2014 after monsoon. While a single monsoonal episode in 2014 indicated 1750.8mm rainfall with -1% DEP (Srivastava and Guhathakurta, 2004). Most of these landslides were triggered by heavy rainfall in Arunachal. The state has a population of more than one million with a decadal growth of about 26% (Census, 2011). The 1950 great earthquake (Mw 8.4) recorded the first seismic induced landslides in Arunachal Himalaya. A detailed macro-seismic survey was done by Poddar (1950); he reported several landslides and subsequent failures of engineering constructions in Subansiri district area due to severe ground shaking by this great earthquake. Arunachal Pradesh being much rugged and inacc-essible, we examined the temporal satellite imagery of Landsat TM and ETM (1987, 1990, 1995, 1996 and 2000), IRS P5 (Cartosat 1stereo data: 2007 to 2011) and IRS P6 (Resource: LISS IV MX of 2008, 2009 and 2012) along with Survey of India (SOI) topographical maps (1960-62 and 1963-64) to assess the landslide risk conditions. A 3D visualization by draping multispectral image over DEM provides a realistic portrayal of the terrain (Fig. 2).  It helps us to recognize prominent locations with concave slope pattern and shapes indicating an old landslide of larger dimension in the entire northeast Himalayas (Fig. 2). 

Fig. 2: 3D visualization of an old landslide, different features of landslide is demarcated.

1: Zone of Depletion; 2: Zone of Accumulation; 3: Displaced material; 4: Crown; 5:  Radial Cracks; 6:  Toe of surface of rupture; 7:  Toe. Image: Resourcesat LISS IV MX image of 5.8m spatial resolution; Backgournd: SRTM DEM of 30 m spatial resolution. Source: NESAC-SR-96-2013. 

The area mainly falls in Siwalik Group, Gondwana Super Group and quartzite of Miri Fr. Presently, the landslide affected area of the 1950 great earthquake is covered with vegetation with a few smaller landslides along the radial cracks and with almost no human influence. These slides are also responsible for sedimentation in River Brahmaputra and its north bank tributaries.

Contrasting bright tones indicating exposed surfaces and shapes help in interpretation of active landslides from EO data (Fig. 3), either though visual interpretation or through semi automated object based classification technique. However, nowadays, AI based classification techniques are also used for delineation of landslides from the EO data.

Fig. 3: 3D view of a rock slide (due to wedge failure) in quartzite terrain. Source: NESAC-SR-96-2013. Image: Resourcesat LISS IV MX image of 5.8m spatial resolution; Backgournd: SRTM DEM of 30 m spatial resolution. Source: NESAC-SR-96-2013.

In the year 2020, some 12 casualties were reported from various parts of Arunachal Pradesh, landslides along the road corridors forced commuters to be stran-ded for few hours. Further, formation of landslide-induced natural lakes in the Arunachal–EHS zone is found, which is a well studied phenomenon due to severe earthquakes and geologic conditions. A detailed investigation was done for a natural dam created in July 2020 by landslide(s) blocking the Yigong river, which is found to be a persistent problem in this part. The lake-reservoir and landslides are more or less permanent feature near the clustered earthquake swarm at the intersection of two major strike slip faults, the right lateral Jiali Fault (JF) and left lateral Yigonge Lulang Fault (YLF) (Fig. 1 & 4). 

Different episodes of moderate to strong magnitude earthquakes (Mw 5.0-6.0) trigger landslides in this zone leading to formation of natural dam & impoundment of water, while intricate fabrics of various fault and thrust systems as well as high altitude mountainous range make this area relatively fragile and susceptible to landslides. Sudden breaches of landslide dams are also responsible for massive floods along downstream of the Siang River in Arunachal Pradesh. At places, river Yiong Tsangpo exhibits braded pattern indicating higher rate of sedimentation resulted from landslide debris.

Fig. 4: Location of landslides and natural dam in EHS the WNW-ESE right lateral Jiali Fault.  Location of clustered earthquake epicenters and geological structures are after B. Mukhopadhyay, and S. Dasgupta, 2015. YLF: YigongeLulang strike slip fault; YF: Yemla Fault; CT: CanyonThrust.

In Sikkim, adverse geology exhibited by Daling Group of rocks, along with steep slopes mainly make the ter-rain most susceptible to landslide. Moreover, the Sik-kim Himalaya is seismically much active, but earthquakes are much deeper (40-50 km) in comparison to the western Himalayan shallower (depth 10-25 km) seismicity (Kayal, 2001, 2010). Study of temporal EO data shows that many landslides are active since decades (Fig. 5).

Fig. 5: An active landslide from Sikkim. a). Cartosat-1 image of 2011, and b). Airbus image of 03 Nov 2019.

It is observed that reported landslide density in Sikkim is 0.6 slide/km2 and the state has population of 610,577 (2011, census). Earliest available documentation of landslides in Darjeeling-Sikkim Himalayas was in October 1968 (Rao, 2019). In addition to rainfall triggered landslides, earthquake-triggered landslides are also common in Sikkim. Total 1,196 landslides are mapped within an area of 4,105 km2 using EO data; these were triggered by the September 18, 2011 earthquake Mw 6.9 (Martha et al., 2014). Like Arunachal Pradesh, formation of landslide induced dam and impoundment of water as well as threat related to breaching of the dam is also common in Sikkim (Fig. 6).