‘Everybody loves a good flood’: the political and social transformation of the eastern Tarai (Nepal) through flood control infrastructures

l’eau


Introduction 1
Floods have always challenged human societies which have adapted to varying degrees. The southern Himalayan foothills, including the Tarai plain in Nepal, is certainly the most challenging region for humans due to the power and irregularity of the waterflow -which can multiply almost a hundredfold -, to the force of this sediment-laden water, but also to the high population density (500 inhabitants/km 2 ). The management of these foothill rivers, which has long been controversial due to the aforementioned factors (Mishra 2008a(Mishra , D'souza 2003, has become the dominant development model for the modern states of India and Nepal, intent on developing a plain which represents a significant economic potential, especially for Nepal. 2 After Independence, India embarked on a massive water infrastructure plan to control flooding and to manage water for irrigation and power generation. The plan was inspired by the success of the Tennessee Valley Authority model (Ekbladh 2010, D'Souza 2003 which is based on the control and regulation of hydrological fluctuations through large-scale structures (Scott 1998). The first major project involved the tumultuous Koshi, the main tributary of the Ganges and an emblematic river of Himalayan hydrological dynamics (Baghel 2014).

Fig 1. Koshi tributaries and study fields
Source: MA Candau, 2022. Flood control: a radical transformation of the Koshi plain 6 The damming of the Koshi in the second half of the 20th century has constrained the route of the river and has (theoretically) protected the land behind the dikes from the river's sudden movements. This has allowed permanent occupation of the plain by agriculturalists in particular, leading to a profound socio-territorial transformation (Candau et al 2015). Today, however, the development of the plain is being challenged by increased flooding.
Methodology and study area 7 This article is based on a period of 30 months of fieldwork in Nepal and India from 2012 to 2014. To study the relationship between water development and flooding, I first focused on the areas most exposed to flooding, eventually concentrating on two border villages, Tilathi in Saptari district and Narsingh in Sunsari district (fig 1), along tributaries of the Koshi River that remain largely affected by recurrent flooding. Village selection was limited to those inhabited since at least 1945 to ensure the availability of historical records and information about the consequences of the Koshi water development project launched in 1954. My main strategy was to base myself all year round as close as possible to the study sites and, from there, to explore the region. I chose to use a mainly qualitative method of enquiry inspired by grounded theory (Bryman et al 1999). One hundred and eighty non-directive and semi-directive individual and group interviews were conducted in nepali, in maithili and in english with the help of Mukesh Mehata, my interpreter and assistant, from the adjacent district of Morang. The majority of the interviews were recorded, then transcribed. The themes of the surveys were mainly related to human settlement, development policies -their consequences and the socio-environmental changes caused by them -, the flood regime and how it has evolved and its effects, vulnerable areas and local adaptation measures. I developed different survey grids according to the defined social strata: farmers, landless people, labourers, agricultural workers, city dwellers, industrialists, shopkeepers, elected officials, non governmental organizations (NGOs), staff managers, planners, representatives of central authorities and researchers. Satellite photos, the 1992 topographic map, simple sketches and photographic coverage complete this approach. 8 The Tarai/Madhesh is particularly unstable and naturally prone to flooding (Delcaillau 1992). At the Himalaya outlet, the deposits of these ever-changing rivers are made up of a series of coalescing cones built over time by the Himalayan rivers abandoning their loads of sediment as they leave the mountains (Zurich et al 2006). That is partly the reason why these rivers can abruptly change course in a process known as 'avulsion' (Fort 2011, Sinha 2009. Within this network of waterways, the Koshi is characterised by exceptional flow variations (Sinha 2009); in an annual flow of 52 billion m 3 of water, it moves an estimated 120 million m 3 of sediment (Mishra 2008b). Up until its damming in the late 1960s, records show that over the previous 200 years the Koshi's channel shifted 150km in a predominantly westward direction (Sinha 2009). 9 On a smaller scale, its tributaries, whose catchment area is limited to the Siwaliks, exhibit the same hydrological irregularity driven by the climatic cycle. The Sunsari is a perennial river originating from the Siwaliks. It flows along an ancient paleochannel of the Koshi. Originally a low-flowing river, its characteristics changed after its main tributaries were connected to the Sunsari Morang Irrigation Project (SMIP) network in the 1980s (fig 1). 3 This connection allows the SMIP to use these rivers as outlets in the event of water and/or sediment overflow in the irrigation canals. As a result, the inflow of water increased and the course of the Sunsari gradually became wider and deeper, resulting in more flooding in adjacent areas. To address this problem, dikes were built. However, as these are not continuous structures, they only check flooding in some areas while in others they actually increase it. 10 The Khado is a temporary river. During the monsoon, its flow can increase very rapidly and carry large quantities of sediment. The construction of a border embankment to divert the Khado's course near Tilathi and along the Indo-Nepalese border (fig 1) has led to very heavy sediment deposition and long-term flooding. The embankment, constructed perpendicular to the flow, acts as a dam. 11 Elderly people from the region explain that the flooding of these two rivers used to be less frequent and more predictable. Today, the consequences of these floods are catastrophic for people, their property, crops and infrastructure (Kumar 2016, Candau 2018).
The Tarai/Madhesh: a strategic area 12 While, today, the Tarai/Madhesh is covered in rice fields as far as the eye can see, this was not always the case and must be understood in the evolving light of socioterritorial political orientations (Candau et al 2015). Many historical accounts present the region as undeveloped, malaria-prone and covered in forest until 1951, yet 'the ancient and medieval history of the region is a cyclical one in which humans and the forest have alternately dominated' (Gaige 1975: 59). A very early human presence has been attested to. The region was also a historical passageway where powerful kingdoms prospered. Indeed, the characteristics of the foothills favoured a succession of flourishing kingdoms, such as the powerful ancient kingdom of Videha (900 to 500 BC) in the eastern Tarai (ibid, 58). Numerous conflicts and transfers of authority between the dominations of the Himalayan and Gangetic plain kingdoms marked these territories during the second millennium (Bernardo 2010), especially in the region west of the Koshi (including present-day Saptari district), a rich and dynamic plain territory and by far the most coveted between the 15th and 19th centuries (Regmi 1978a: 161-66, Warner 2014. Anthropological data reveals that this 'land between' (Krauskopff 2006) had long been used by cattle breeders from the Gangetic plain (Gaige 1975). 13 The economic potential of the Tarai plain was gradually exploited by the newly unified state of Nepal which, from the late 18th century, encouraged its colonisation by people from northern India (Regmi 1978b). At that time, the Tarai was occupied by populations from the plain, both the Tharus, an ethnic group , and caste populations. These caste populations, whose date of settlement is difficult to determine (Gaige 1975: 61), have descendants culturally very close to the recently settled Indian populations and are all now called Madheshis.
14 The eastern Tarai played a predominant economic role from the beginning of the 19th century (Regmi 1984). The farmers of the plain 'supplied the rest of the country with rice and other valuable commodities' (Meyer 2000: 16). However, the Nepalese state regarded the plain as a legally and culturally distinct and dominated territory (Whelpton 2006). For example, a passport was required by Tarai people who wished to travel to the mountains (Gaige 1975, Warner 2014. 15 In 1951, upon the fall of the autocratic Rana regime which had reigned for a century, the monarchy took over the reins of power and opened the borders. International aid flowed in accompanied by a political-scientific discourse that was to be embodied in development plans mainly focused on the Tarai plain. 16 Numerous programmes were then implemented: malaria eradication, projects to settle Nepalese migrants expelled from Assam and Burma (Dahal 2013), road and water infrastructure development and land reform (in 1964) (see Aubriot et al, as well as the maps designed by Picard, this issue). The subsequent massive transformations led to a wave of migration by mountain populations or Pahadi (Candau et al 2015). Although it represents only 17% of Nepal's territory, the Tarai is currently home to more than 53% of the country's population according to the 2022 Census and, within a few decades, has become the country's most important region from both a socio-economic and a (geo)political point of view. 17 The spectacular development of the Tarai plain in the second half of the twentieth century was made possible by the development of large-scale water projects. These, inspired by the Tennessee Valley Authority (TVA) model of total control over the natural dynamics of rivers, also allowed for the reinforcement of the state construction of a new political order while delivering the 'fruits of development' (Baghel 2014), objects of the legitimation of power. Flood regulation of the Koshi: a river under control? 18 In 1953, particularly disastrous floods in northern Bihar and the Tarai plain led to the formation by the government of India of an expert committee, the Central Water Power Commission (CWPC) for the Koshi (Mishra 2008b), to study flood control options. Despite a century-long controversy over the appropriateness of damming Himalayan rivers, especially the Koshi during the period of British colonisation in India (Mishra 2008b, Sinha 2009), the Koshi project was launched in newly independent India and in a Nepal liberated from Rana autocracy. The project included three major components: a dam on the Nepalese side of the river near the border, dikes on both banks in both countries, and extensive irrigation schemes. 19 The dam, completed in 1962, is located at Bhimnagar (Nepal) and is 1,140m long. Dikes were built in 1959. In Nepal, a 34km-long embankment was constructed on the eastern riverbank upstream of the dam and another, 12km long, on the western bank. In India, downstream of the dam, 125km and 126km embankments, respectively along the eastern and western banks of the Koshi, were constructed to protect 214,000ha of land from flooding. In addition, water is diverted from the dam to two canals on both sides of the river to irrigate 612,500ha of land mainly in India (Malla 1995). 20 In order to compensate Nepal for the loss of land caused by the construction of the dam on its territory but also by the use of its natural resources for the construction of the infrastructure, the SMIP, Nepal's largest irrigation scheme, was built in several stages . This system was meant to cover 64,000ha in Sunsari and Morang districts. As per a bilateral agreement, India put up the first constructions, which quickly became inoperative for various reasons: displacement of the Koshi, silting of the canal and poor quality of construction. Between 1978 and 2001, the project advanced on a more solid footing thanks to financing from the World Bank. This irrigation network (fig 1), covering 84% of the plain in Sunsari district, facilitated rapid cultivation of the area which, previous to the construction of the dam, was the Koshi's flood plain. 21 Despite the excitement of bringing under control a river with a maximum flow of 24,210m 3 /s over a width of about 1km -the first project of its kind in India and the world (Mishra 2008a) -the Koshi project has failed to deliver. Today, the exceptional flooding of the Koshi due to breach is largely the result of schemes meant to tame the river but which have failed to do so. The cost of the Koshi project has been very high (Malla 1995) due to the frequency of dam repairs caused by breaches (Mishra 2008b). Furthermore, only two years after the implementation of these schemes, issues concerning land waterlogging due to seepage through the dikes had to be dealt with. While the project was originally supposed to protect more than 2,000km² in India and Nepal, by 1990 more than 3,000km² of land was waterlogged (Gill et al 2018) on a recurrent basis for periods of up to 8 months.
Tilathi: a commune in the shadow of hydraulic developments 22 Tilathi VDC ( fig 2) is located in Saptari district on the banks of the Khado River and on the Indian border. It is an ancient urban centre with a rich intellectual and political history (Regmi 1978b, Candau 2018. 24 Despite having one of the highest literacy rates in the country according to the 2011 census, the socio-economic situation has deteriorated in the VDC due to increased flooding. 4 What are the implications, with regards this situation, of the Indian-built border embankment that is used to divert the Khado River? The Khado: the 'sorrow' of Tilathi alluvium which become trapped by the dam. The riverbed then tends to rise on top of the deposits, forming an unstable network of flows that dominates the surrounding land. According to our surveys, the resulting intensification of flooding means that the Khado is perceived locally, especially in Tilathi, as the 'sorrow' of Saptari, as is the case of the Koshi in Bihar.

Flooding and development: causes and consequences Fig 3. Diversion of the Khado River through the border dike near Tilathi village
The diversion is located in the no-man's land area between Nepal and India Picture: MA Candau, dry season, May 2014. 30 Moreover, the numerous infrastructures (roads, buildings etc) built in India, just downstream of the border along a west-east axis, reinforce this damming effect on the natural drainage of the rivers. 8 According to our interlocutors, the frequency of flooding, which was two to three times per year before 1974, has tended to increase to nine times per year during recent monsoon seasons. The multiplication of infrastructures on the Indian side amplifies the phenomenon and large areas with stagnant surface water problems arising in the dry season have been observed ( fig 4).   5) as a result of an exceptional sedimentation of 8m at this location within 30 years. The productivity of riparian land has been seriously affected by sedimentation. According to the different protagonists, the situation is gradually deteriorating and becoming intolerable. 10 Every year, floods cover an estimated 16 to 20 hectares of agricultural land which is mostly ruined by alluvium deposits. A process of desertification is therefore taking place due to the gradual sterilisation of agricultural lands affected by these massive floods and the sandy sedimentation. 33 The first difficulty in living with floods is the struggle at home with rising water levels (up to 3m). Humidity scours the earthen walls causing collapse. The second difficulty is of a sanitary nature: stagnant water in a tropical environment generates diseases (diarrhoea, cholera, typhus, polio etc) which can readily turn into epidemics, especially when groundwater is affected. About 65 people die every year in Saptari district due to water-related diseases. 11 The third difficulty is economic and therefore social. Floods destroy crops and grasslands, sterilise land with sedimentation and close communication routes, thus blocking trade and access to supplies. This all contributes to the impoverishment of the region.

Management and solutions: questions of scale 34
In the Tilathi region, a significant part of the population manifests acute political awareness due not only to its rich history of political engagement but also to its high standard of education (Candau 2018). This dimension explains the development of local popular movements to safeguard and defend the living environment, going as far as opposing the decisions of the central Nepalese state and even those of the Indian authorities across the border. In 2012, following a breach in the Kunauli embankment resulting from the river's floods, India undertook reinforcement work (raising the 'Everybody loves a good flood': the political and social transformation of th... European Bulletin of Himalayan Research, 60 | 2023 embankment) and repairs. As a result of this work, flooding along the Nepalese part of the river increased in severity. A spontaneous uprising broke out, with people on both sides of the border engaging in violent fights on the dam. Similar events occur with each new Indian intervention on the infrastructure. 35 In both government thinking and local debates, however, the construction of dikes seems to be the one-size-fits-all solution to the ills of flooding. In 2012, at the request of local people, the Ministry of Irrigation released funding for a protective dam and entrusted its construction to the Department of Water Induced Disaster Prevention (DWIDP), according to a survey I conducted in February 2013. Unfortunately, work did not start until the end of the fiscal year which coincided with the start of the monsoon in July 2013. According to the local population, 'the DWIDP wanted this expenditure to be done before the budget was closed'. 12 Construction work started in July. When the first flood came in August, the work was destroyed. For the local population, local journalists and NGOs, it seems obvious that the construction of a protective dam should be carried out well before the monsoon since, during the rainy season, the only possible measures are emergency measures. 13 Faced with this terrible waste, people demonstrated for three days to express their discontent. The inhabitants of Tilathi then rallied together to deal with the emergency situation and built a protective dam with their own means, using rice bags filled with sand and bamboo (fig 6). 'The use of their knowledge of the river and its dynamics allows them to correctly position the dam, which then plays its protective role'. 14 This dike held throughout the monsoon and is a perfect example of the value of taking local knowledge into account alongside expert knowledge. illustrates the lack of responsiveness of the Nepalese central government and its miscalculated encouragement of unilateral Indian actions. 37 The people of Tilathi and the flood-hit VDCs are aware that the problem of flooding cannot be solved without real government support. The inhabitants denounce Nepal's ambiguous position vis-à-vis its powerful neighbour. In the recurrent context of Nepal's ethnic conflict (Lawoti et al 2012), the demands of the Madheshi movements are particularly virulent in the eastern Terai, especially in Saptari district. Despite this, the Nepalese state is relatively unreactive -even passive in the eyes of inhabitantsleading them to believe the state has an unconfessed desire to see the Madheshi cultural centre disappear. Such a situation illustrates the changes in hydrological dynamics and their consequences on the population and activities of a region previously known for its rich political, cultural and socio-economic past that is emblematic of Madheshi culture in this part of Tarai. Let us now look at the Sunsari River, a left-bank tributary of the Koshi River.
Narsingh: an area experiencing increased flooding 38 Narsingh Tappu is a VDC bordering India, located approximately 10km from the Koshi in Sunsari district whose capital, Inurawa, is about 5km away (fig 7). The VDC is bounded in the north by the Mahendra National Highway and in the west and south by the Indian border. The landscape consists of a generally flat plain with an altitude ranging from 80m in the north to 70m in the south. The soil in this territory, the former bed of the Koshi river, is made up of ancient deposits, including a large proportion of sand which is very prone to erosion. 39 The population, of about 25,000 inhabitants, is essentially made up of Madheshis who settled here in successive waves due to (1) the clearing of territories suitable for human settlement following the westwards displacement of the Koshi at the end of the 19th century, (2) land policy incentives implemented from the beginning of the 20th century, and (3) management of the Koshi.
40 Narsingh was a transhumance territory for herders from the Indian state of Bihar and Saptari and represents a source of poverty-stricken manpower (Candau 2018) which partly explains the territory's configuration: fields, grasslands and industries including three factories: two paper mills and a slaughterhouse. The majority of the population is made up of Muslims, Yadavs (herder caste), Telis (business men, intermediate caste), Musahars (Dalit-Madheshis) and of some marginalised janajatis from the plain such as Jhangads and Dhanuks. The populations' income comes mainly from agriculture, animal husbandry and, to a lesser extent, local industries, trade and government jobs.
The Sunsari river: Koshi paleochannel 41 The meeting of the Thalaha and Belaha Khola rivers at the entrance to the Ramdhuni forest forms the Sunsari (fig 7). This perennial river follows an ancient course of the Koshi River. Its main tributaries, all on its left bank, run from north to south: the Sehra Khola, the Khadara Khola, the Kakro Khola and the Kuluwa Khola. These essentially drain the eastern part of the plain and all have their source in the Siwaliks. The Sunsari River, generally remaining between its banks during the low-water period, usually overflows downstream from the village of Bhutaha in Nepal during the monsoon  1990,1999,2008,2010,2013 and 2014 -with a marked acceleration over the last ten years. Increasingly complex floods 43 The Sunsari has grown from a modest river to a large river within just a few decades, causing destructive floods with major socio-environmental consequences, especially in the south of the district. Like the consequences of mining in the Siwalik, and just as in Tilathi, channelisation and diking have caused notable flooding here.
Channelling the Sunsari tributaries: hydraulic restructuring of the SMIS 44 In the Sunsari basin, between Mahendra Road and the Chatara Canal (the main irrigation canal), the hydrological dynamics of the Sunsari have changed considerably since the establishment of the SMIP. The current tributaries of the Sunsari, small streams that run along a north-south axis perpendicular to the Chatara Canal, were over-dammed and diverted downstream to be connected to the Sunsari for two main reasons between 1983 and 1988: (1) to protect the Chatara canal from run-off water that can create flooding and damage it, and (2) to dump excess sand during maintenance of the canal. The region's riparian land suffers badly from erosion and sand deposits: During the monsoon, the DWIDP opens the canal weirs to release their overflow into the rivers. Their flows increase as well as the level of their bed with the high sediment load associated with the Koshi, especially sand. This leads to increased flooding, bank erosion and sand deposition on the surrounding land. 15 46 This hydraulic restructuring has significantly altered the hydrological regime of the Sunsari basin which is characterised by a sharp increase in flow and sedimentation in all rivers. 'In my youth we could jump from one bank of the Sehra to the other, but after the construction of the SMIP, all the rivers have become wider and deeper leading to wider and more frequent divagations'. 16 47 The Sunsari River, the outlet for these artificial rivers, rapidly deepened and widened, causing increasingly intense flooding over increasingly large areas in the southern part of the district. The area most vulnerable to flooding is ward 6 of Narsingh VDC downstream of the Sunsari (fig 7): 'the Sunsari has become four times bigger than before and today the flood level reaches 1m in the village which was once preserved'. 17 During the monsoon season this area also suffers from waterlogging caused by the Suksena canal, a secondary canal of the Chatara canal and oriented north-south. 'The drainage system is not maintained in the southern part of the district due to a lack of resources'. 18 48 Throughout the monsoon, rainwater is trapped along the edge of the canal where the land is lower. Flood waters from the Sunsari River mix with the rainwater near the canal, which prevents crops from being grown on this agricultural land or destroys the crops. The resulting stagnant water leads to problems such as epidemics and hinders the movement of people and goods. During these episodes, people are forced to open gaps in the canal banks to let the water flow out (fig 8). After the monsoon, they plug them again for irrigation purposes. The vicious circle of containment: downstream of the Sunsari 49 Hydrological changes have led to more pronounced flooding downstream. As a result, in 2010, a protective embankment was built in Narsingh VDC. The structure extends from Narsingh's ward 1 bridge at Mahendra Road to Narsingh ward 5 (see fig 9). The DWIDP in Biratnagar coordinated the project, which was funded by the Indian Embassy under the Indo-Nepalese Economic Cooperation Agreement (2003) to allocate funds for river embankment. 50 A limited budget of Rs 3 million determined the quality of the structure, which is made of earth, and did not allow any compensation for requisitioned private land. The work was laborious because it met with strong opposition, including from political leaders who owned some of the land and from residents of Babiya VDC on the other side of the embankment. RM, one of the landowners concerned gave three main reasons for the opposition movement: (1) the contractor refused to give compensation, (2) the dam built only on the western bank would protect Narsingh but would affect Babiya on the eastern bank, and (3) the most vulnerable people, the homeless (sukumbasi) living on government land without a title, would be trapped between the river and the dam. As a result, landowners have prevented the construction of the dam on their farms, causing discontinuity in the course of the dam: it extends from the road to the bridge for 3km, then only 1.3km downstream and continues for 1.6km to the end of the landless hamlet (khatwe) in ward 5 of Narsingh (fig 9). been dammed to make the course of the river straighter: 'We tried to change the course of the river but it didn't work'. 19 The water then flowed into these spaces with greater force.
52 Locals question the appropriateness of this intervention: 'They built a diversion channel to divert the river's course but it is higher than the bed and its width is too small. So the water had no possibility of going along this new path!'. 20 Villagers point to the lack of seriousness and quality in the development of the project. As in Tilathi, the dike was built in a hurry at the beginning of the monsoon season, just before the end of the fiscal year. Moreover, the actual design of the dam by DWIDP engineers was based solely on engineering formulas: 'We have no hydrological data on the dynamics of the river and we don't have a map of the areas at risk; that's our shortcoming. We also have to deal with large landowners who want the river to flow in a small area'. 21 This discourse is countered by local people: 'The DWIDP and the contractor want above all to maximise their profits'. The interpretations, when combined, give an account of a situation that is both absurd from a management point of view and catastrophic from an economic and social point of view. 53 In the end, the only territory in Narsingh that is constantly protected is ward 1 where factories are located, whereas there is an intensification and displacement of flooding in the neighbouring territories. In Narsingh, flooding has increased by 50% in the affected area, in its southern part: wards 4, 5 and 7 (fig 9). In addition, Babiya VDC on the opposite bank has experienced flooding in previously unaffected areas, as has Bokraha VDC to the north (see fig 9). According to a local journalist, a Sunsari specialist: The river started its movement about 20 years ago and, in the process, eroded land in Narsingh 1. The embankment was decided under the influence of a big landowner, former chairman of the municipality and co-chairman of the district, in order to protect his land and his sister's land in Narsingh 1. 22 54 Curiously, the same phenomenon occurred only a year after the dike had been built. Thus, in 2011, after exceptional floods in Bokraha VDC, following intense flooding caused by the 2010 Narsingh dike upstream, the latter was provided with a small 500mlong dike in record time at the request of MY, a politician and representative of this VDC in the DWIDP. The same mechanisms were repeated: the sukumbasi remain trapped between the river and the dam, and devastating floods moved to land upstream of the dam.
55 It is clear from these examples that dikes are built in places requested by influential people, 23 without following a comprehensive flood management plan or consulting different types of local inhabitants. In effect, a powerful constituency (large landowners and influential people), with a keen interest in embankments, has emerged. Therefore, it is no longer simply a technical question or a political choice by the government but a question of the vested interests of influential people. As a result, exposure to flooding of certain areas and populations is on the increase.

Flood-adaptation methods have become insufficient
56 While people over 80 describe a highly developed flood-coping system which prevailed before the damming of the Koshi river, the situation has changed significantly. They used to know how to recognise the occurrence of floods by observing, for example, the colour and the level of the water, the behaviour of animals, and precipitations. This knowledge enabled them to prepare themselves if they anticipated heavy floodingmoving the elderly, women, children, livestock and their belongings to higher ground, for example. In addition, life went on during floods thanks to the many boats that enabled them to move around. Today, the situation is quite different because the hydraulic restructuring work with the building of dikes has given the populations a false sense of security. In order to enjoy a feeling of safety, most people came to settle behind the dikes: immigration therefore exploded in the region, erasing ancient knowledge and practices of coping with flood conditions. Flooding in Narsingh creates total isolation because the roads are cut off, limiting people's movements. In the event of illness, people rally round to carry patients on stretchers through the water, sometimes travelling for kilometres before reaching a passable road. They no longer have boats to get from one place to another during these periods and they told us that they remained in their flooded houses for fear of thieves coming from neighbouring Bihar. In 2013, the agricultural losses due to floods were very high, as much as 50% according to villagers. In anticipation of the floods, inhabitants prepare satwa (a kind of flour made from a mixture of corn, chickpeas and wheat which is then roasted) which they then mix with water, and they stock up on dried food (buja or puffed rice, churya or beaten rice), although the poorest among them cannot afford to. The latter are then forced to take out loans (60% minimum interest rate per year) from local notables to buy grain or medicine. Every year, entire families fall into this 'poverty vicious circle', are forced to sell their possessions and migrate to India to work in an attempt to repay, if not their debts, at least the interest on them.
57 The containment policy is showing its limits. Water is not drained properly because of the obstacles made up of dikes and other infrastructures. A new type of flooding has emerged, with water that stagnates and can remain trapped for up to eight months, making the land unusable. According to our observations and the testimonies we collected, in order to help water escape, the populations concerned breach the dikes and canals, preferably at night to avoid conflict with neighbouring communities, the local administration and the police.
Inadequacy or lack of management measures? 58 In the Nepalese Tarai, these two sites caught my attention because of the recurrent flooding that occurs there and the resulting hardship. Our analysis shows that these floods are linked to the installation of protective dams and that they are distinct from other floods.
Floods that become more complex over time 59 In Tilathi, the dam built on the Indo-Nepalese border causes flooding due to a head dam. The dike, perpendicular to the river, acts as a storage dam blocking waterflow, which then accumulates, stagnates and sediments, leading to the sterilisation of land, the destruction of infrastructures and epidemics.
Koshi into a river that is over 200m wide in places, and, despite being partially dammed, it is the source of increasingly frequent and severe flooding which devastates the landscape. There is also flooding due to drainage (rainwater and land being trapped behind the infrastructure). This type of flooding, which is rarely taken into account at the development stage, is the result of several factors: (1) the construction along watercourses of protective dikes with undersized and unmaintained drainage gates which quickly become obsolete due to sedimentation; (2) SMIP irrigation canals with the same deficiencies act as a barrier to the flow. These floods have the particularity of stagnating over long periods, becoming the source of epidemics (malaria, fever etc), of land sterilisation and of the destruction of infrastructures.
61 These examples underline the complexity of the physical phenomena, while often indicating a combination of several criteria, which can therefore be called 'combined flooding'. All this shows the intertwining of the phenomenon and its causes, especially when combined floods become recurrent.

Inadequacy of infrastructures: indicative of power relations?
62 In our two case studies, we observe an increase in the complexity of the floods, which testifies to the inadequacy of the measures taken. According to a former minister of water resources and a water expert, the technological choice is inappropriate for Himalayan rivers that are particularly rich in sediments, with the result that 'matter is a co-equal passenger with water'. 24 Hydraulic projects are based on models developed for Western rivers (Mishra 2008, Dixit 2003. This inadequacy is typical of the scientific equations of so-called orthodox science which aims to be universal (Forsyth 2003). However, the problem also comes from the approach used by engineers. They do not consider the specificity of each river in their management plan. The DWIDP does not have a map of the areas at risk in Narsingh even though this department is committed to reducing them. 25 It is obvious that the only local inhabitants who have a say in these projects are political leaders, large landowners or affluent people. It is easy to see how planning without an integrative, cooperative and representative approach leads to approximate and unbalanced planning, often with disastrous consequences.
63 In Narsingh, the construction of a dam to protect certain areas has caused the flooding problem to shift to other areas. As a result, the populations of these newly affected areas need to build a protective dam to protect themselves, if they are able to obtain adequate political support. The consecutive series of construction projects simply moves the problem further along.
64 Pressure from political leaders, often also large landowners, influences the layout of dikes which may end up near the river or even in its bed. Engineers recommend or reject the installation of dykes and their positioning depends on the social or political pressure exerted on them (Mishra 2008). As engineers abandon their calculations, dikes are used in unexpected ways. They become emblems of the struggle for land, between the land gained through a river's divagation and the land protected by dikes. These examples reveal that power relations lie at the heart of the positioning of dikes (Candau 2018). 65 The water management sector generates the handling of large financial funds. The misappropriation of development project funds is widespread in the Global South (Molle et al 2009) and it seems that this is also the case in Nepal. The increasing cost of building and repairing dikes has forged a very strong link between politicians, engineers and contractors. 26 These groups of actors have a mutual vested interest in 'development' projects, especially because dikes can be regularly washed away by floods and therefore need to be constantly rebuilt (Candau 2018, Gill 2018. The most recent actor to play an integral role in the appropriation mechanism in both Nepal and Bihar is the mafia; a mafia made up of political party middlemen, but also powerful entrepreneurs who wear the two hats. These mafia groups have gained control of natural resources, such as timber in the Siwaliks. This exploitation also applies to sand and stone, which has led to an inexorable degradation of environmental conditions, resulting in even worse flooding, as in the case of Narsingh and Tilathi. Social fragmentation and inequalities due to planning 66 The development of the Koshi basin through the construction of dikes, dams and irrigation networks is a good example for examining how floods structurally generate disasters (White 1945, Wisner et al 2004 and for analysing the social dimensions of the decision process and 'who wins and who loses' (Peet et al 2011). As villagers struggle to influence the specific location of a dike, local inequalities hinder their efforts as they battle against those with greater financial means and social standing who have ready access to and influence over engineers and their supervisors. 67 The alignment of dikes in Narsingh is said to have 'solved' the 'flood problem' by protecting the land of political leaders, their supporters and influential people but by sacrificing marginal people and sukumbasi families living on the riverbanks. This highlights the fact that, even when a problem is supposedly 'solved', the question remains: solved for whom?
68 Injustice also extends well beyond the construction of the dam, those who live closest to the infrastructure also suffer the most from its failures, side effects and state of disrepair, while rarely enjoying the benefits of its construction (Cannon 2000). In effect, social groups become fractured and unequal in the face of risk (Peet et al 2011). The provision of infrastructures divides these groups both physically and metaphorically (Wisner et al 2004). Mishra (1999) describes the violence that erupts within communities when villages, and sometimes even extended families, are separated by the construction of dams. In Tilathi and Narsingh, dike construction or breaches can lead to violent conflicts between neighbouring communities of the same culture and often of the same family.
69 In a group interview focusing on a satellite photograph of Narsingh ward 6, the people present, together with their political leader, became very angry when I mentioned the flooding problems on the opposite side of the Sunsari river in Jalpapur VDC. They saw the possibility of joint management as an affront. The embankments, erected at the behest of influential people, reflect the clash between political leaders, who orchestrate the diktat of development projects to ensure the security of their constituents and their villages even at the expense of those of their neighbours. Yet it is precisely this exclusion of others that allows local political leaders to maintain their power by deciding on whom to protect and who can be excluded to facilitate solutions. Indian political leaders to seek to control the Koshi river and its tributaries using dikes in order to limit flooding and the natural movement of the rivers. This process has protected the settled zone in Saptari district, while ensuring the development of the floodplain, especially in Sunsari district. 71 The two sites studied in this paper, located on tributary rivers of the Koshi, face several economic and ecological challenges besides threats to lives due to the construction of protective embankments. In Narsingh, protective dams follow a political logic that is tinged with a neo-feudal social arrangement that is further widening the gap, in terms of vulnerability, between the populations faced with the risk of flooding. In Tilathi, the retaining dam is leading to the inexorable destruction of areas once renowned for their socio-economic dynamism and for the wealth of their Madheshi culture. 'Nepal's passivity in the face of the unilateral actions on the part of its powerful neighbour' 27 is leading to a feeling of abandonment and exclusion among the local population. The distribution of flood hazards in Sunsari district highlights an inequality between the northern and southern parts, which in fact overlaps with a social division. Indeed, Madheshis mainly occupy these southern lands and are affected by the floods (Candau 2018). The global situation reignites Nepal's ethnic conflict that pits the population of Madheshi affiliation against the Pahadis (Lawoti et al 2012).
72 Each of the sites reveal unequal and partly ineffective flood protection measures. Worse still, they reveal a multiplication, diversification and amplification of flooding, including in areas not previously affected. The unsuitability of the technical choices is obvious and it is therefore necessary to modify them. 28 Nevertheless, people's views on dikes are more favourable to these infrastructures than those of many researchers (Das et al 2013, Gill 2018. They see these structures as an absolute necessity for their safety and survival even though they have suffered enormously from dike breaches. They blame government and governance for these structural failures. Tilathi, an ancient and culturally rich settlement, is a marginal case which stands apart in that its residents demand that a real technical protection plan be undertaken in consultation with the local population. 73 The official development policy is based on the dominant paradigm of 'building dikes to eliminate floods'. This is a high-visibility policy that officials can use to justify their actions to the population. The Director of the Intelligence Bureau (DIB) in Patna, recalls that 'building dikes is a sure way to get into corruption'. 29 The erection of the embankment at Tilathi and Narsingh by the DWIDP is a good example of this phenomenon. Hastily built at the beginning of the monsoon and at the end of the fiscal year and of low quality, the embankment broke at the first flood with devastating consequences for local people but delivered the 'benefits' of development to politicians, engineers and contractors. According to GB, a water expert, there is a complete dislocation of the management paradigm, with the water resources department focused solely on 'land and cement contracts', not on understanding hydrological dynamics.
There is no long-term vision that unfolds into a comprehensive prevention plan: 'no one cares about floods, what matters is how much money you can make in one season'. 30 Sainath (2002) has described why 'everybody loves a good drought'; several actors, including political and development actors, profit from droughts to the detriment of 'Everybody loves a good flood': the political and social transformation of th...
European Bulletin of Himalayan Research, 60 | 2023 causes of floods are multiplying. These two case studies are emblematic of a demonstration of how floods are transformed into socio-environmental disasters and are vital for studying the structural dimensions of who wins and who loses in this new water-balance configuration where areas are exposed or protected.