environmental risk A river system as a historical actor The Kosi is not a backdrop to the history of Mithila. It has shaped settlement, agrarian opportunity, transport, administrative boundaries, public works, migration, and political expectations. Yet a river is not an intentional political actor. Historical explanation must connect geomorphology and monsoon hydrology with institutions that decide where people may live, which structures are maintained, how warnings move, and whose losses are compensated. This chapter therefore treats environmental risk as produced by an interaction of hazard, exposure, infrastructure, and social capacity. Seasonal high water can be beneficial in one landscape and disastrous in another. A breach can convert a protected tract into a catastrophe zone. An avulsion can move the main channel far more abruptly than ordinary bank erosion. Using one word - flood - for all three processes hides crucial differences. The Sapt Kosi basin: one river assembled from many headwaters In Nepal the river system is commonly understood through the Sapta Kosi, the seven-river network that gathers Himalayan and trans-Himalayan waters before entering the eastern Tarai and then Bihar. The basin includes very high-relief terrain and a large sediment-producing mountain catchment before the river reaches the low-gradient Gangetic plain. This contrast between steep headwaters and a flat depositional foreland is central to Kosi behavior. Political boundaries divide the basin among China, Nepal, and India, while water and sediment do not stop at those boundaries. The most consequential flood-management relationship for the lower basin is bilateral India-Nepal cooperation because the barrage and important embankment reaches lie in Nepal while much of the downstream exposure lies in Bihar. Basin geography and state jurisdiction therefore operate at different scales. From mountain gradient to the Gangetic plain A river leaving a high-energy mountain environment loses gradient and transport capacity as it enters the plains. The Kosi carries water, gravel, sand, silt, and finer material generated across a tectonically active and monsoon-affected catchment. Deposition is therefore not an accidental by-product: it is part of the normal construction of the alluvial landscape. Infrastructure changes the places in which deposition can occur. When flows are constrained between embankments, channel-bed and bar evolution, bank attack, and local aggradation must be monitored together with discharge. The historical problem is consequently not 'too much water' alone. It is the combined management of water, sediment, drainage, and a densely inhabited floodplain. The Kosi megafan and the language of avulsion The Kosi has built a large megafan in north Bihar. A megafan is a broad depositional landform produced where a large river repeatedly distributes sediment across a foreland surface. Such terrain contains abandoned channels, active distributaries, low areas, natural levees, wetlands, and settlement mounds. Its history cannot be reconstructed by assuming that the modern channel has always occupied the same corridor. Avulsion is the rapid relocation of a river into a different course, usually over a much larger distance than ordinary lateral bank migration. The 2008 event is a particularly important example. Distinguishing avulsion from routine flooding matters because the probability, warning problem, infrastructure exposure, and recovery geography differ when the main river itself changes route. Historical maps and the caution against a simple westward- migration story A familiar account states that the Kosi migrated continuously westward by more than one hundred kilometres over roughly two centuries. Chakraborty, Kar, Ghosh, and Basu re-examined twenty-eight maps published between 1760 and 1960 and argued that the mapped channel positions instead show substantial oscillation, with the river often remaining in an east- central belt of the megafan rather than sweeping uniformly from east to west (Chakraborty et al. 2010). This does not make old maps useless. It makes them evidence that must be compared critically. Survey accuracy, map scale, the representation of multiple channels, publication date, and the distinction between a principal channel and a seasonal branch all affect interpretation. The book therefore avoids converting a useful regional image of mobility into an exact two- century trajectory that the cartographic record itself does not securely establish. Why 60 km and 120 km can both appear in accounts of 2008 Scientific discussions of the 2008 avulsion use different measures of displacement. Chakraborty and colleagues describe a shift of about sixty kilometres in the central part of the megafan, while Sinha and co-authors describe an approximately 120-kilometre avulsion in relation to the modern course and avulsion geometry. These figures refer to different ways of measuring a spatially complex channel relocation rather than a single survey line that all authors define identically. For historical writing, the safest point is qualitative but still precise: the 2008 change was exceptionally large, rapid, and eastward, and the floodwater occupied an older course outside the embanked channel. Where a numeric distance is used, the method and source should accompany it. Sediment is not a footnote to flood control The Kosi transports very large sediment loads from the Himalaya. As flow enters the plains, deposition changes bars, channel depth, local slopes, and the relation between the channel bed and surrounding land. Sinha et al. (2014) identify sediment flux, planform dynamics, and human intervention as important parts of avulsion susceptibility. Recent work also maps reaches of aggradation and evaluates silt-management options. No single sediment policy follows automatically from this diagnosis. Dredging, channel training, embankment modification, sediment extraction, and floodplain accommodation each redistribute costs and benefits. A source- controlled history therefore records sediment as a governing constraint and treats proposed engineering remedies as proposals whose ecological, financial, and social consequences require separate evidence. The colonial flood archive: a record of recurring administrative concern The Bihar State Archives' digitized collection on floods from 1871 to 1939 shows that colonial officials repeatedly documented crop damage, relief, river-course change, bund construction, drainage, embankment law, and district reporting. Entries include an 1895 record on bunds along the Kosi, a 1907 file on river-course changes affecting the Nepal-British boundary, and a series of early-twentieth-century flood and engineering reports. The archive is valuable precisely because it predates the postcolonial Kosi Project. It demonstrates that the problem of governing mobile rivers was not invented in the 1950s. At the same time, colonial records were produced for revenue, administration, engineering, and order. They reveal official priorities more directly than they reveal the full experience of cultivators, fishers, boatmen, laborers, or people whose settlements shifted outside the documentary centre. Embankment law and revenue logic in colonial Bihar The Bengal Embankment Act of 1873 gave government a formal legal framework for embankment administration. Across north Bihar, colonial policy was repeatedly pulled between protecting taxable cultivation, preserving drainage, limiting unauthorized bunds, and avoiding expensive commitments to structures that could create new liabilities. Christopher Hill places these interventions within a longer history in which environmental management and social control were intertwined (Hill 1997). The lesson is methodological as well as historical. 'The colonial state opposed embankments' and 'the colonial state built embankments' are both too simple. Policy varied by river, decade, fiscal expectation, engineering opinion, and perceived downstream effect. The Kosi entered independence with a long archive of debate rather than a settled technical consensus. From colonial caution to postcolonial hydraulic developmentalism After independence, flood control became part of a wider developmental promise linking engineering, irrigation, agricultural productivity, transport, power, and state capacity. The Kosi project emerged within this setting. Its advocates did not present an embankment in isolation; they envisioned a multi-purpose system of barrage, headworks, flood banks, canals, roads, and power generation. This shift matters because infrastructure produces path dependence. Once villages, roads, canals, markets, and agricultural expectations adjust to a protected corridor, the consequences of failure change. A breach can then affect land that has been settled or intensified partly because routine flooding was expected to be reduced. The 1953 scheme and the promise of a multi-purpose project A contemporary Yojana account in September 1963 described a Kosi scheme drawn up in 1953 at an estimated cost of Rs 44.76 crore, combining flood protection with extensive irrigation and a barrage carrying road and rail links. Such official publicity captures the developmental expectations surrounding the project at the moment when its principal works were being completed. The figures in promotional sources should be read as planned benefits, not retrospective proof of delivered outcomes. Canal command, seasonal water availability, drainage, maintenance, land acquisition, and farm-level access determine whether nominal irrigation potential becomes reliable irrigation. The historical significance lies in the scale of the promise and the way it reorganized public policy around the river. 25 April 1954: the original India-Nepal Kosi Agreement The Governments of India and Nepal signed the Kosi Agreement at Kathmandu on 25 April 1954. The treaty text describes a barrage, headworks, afflux and flood banks, canals, protective works, irrigation, hydropower, and erosion protection. It authorized construction on Nepalese territory and specified land, compensation, communications, and related project arrangements. The agreement is primary legal evidence for what the two states authorized and allocated. It is not, by itself, evidence that every maintenance obligation was performed to the same standard in every year. Evaluating implementation requires engineering inspections, budget and work records, bilateral committee minutes, field evidence, and post-disaster investigation. Sovereignty, access, and infrastructure on Nepalese territory The Kosi Project is unusual because major structures serving downstream Bihar are located in Nepal. This creates a permanent relationship between sovereignty and operational access. Repair teams, materials, leased land, inspection, barrage operation, canal interests, and local use of project areas can all become bilateral matters even when the immediate risk appears local. The arrangement should not be described as if one country simply controls the other's river. Nepal retains sovereignty over its territory; India and Bihar agencies exercise project functions through treaty-based arrangements and bilateral mechanisms. Political criticism in both countries often reflects wider histories of asymmetry, sovereignty, and benefit-sharing, which should be attributed rather than presented as a single uncontested national view. Construction, embankments, and the remaking of the lower Kosi corridor Construction began in the mid-1950s. Official reporting in 1963 stated that embankments had been completed in 1959 while work on canals and associated facilities continued. The engineering intervention confined much of the lower river to a narrower corridor and connected the flood-control objective to irrigation infrastructure. Confinement did not erase the river's internal mobility. Braided channels, bars, erosion fronts, and sediment deposition continued between the embankments. The social effect was equally uneven: some land received reduced routine inundation, while settlements and fields within the embanked corridor remained exposed to the river's changing channels. The Kosi Barrage: completion and function The Kosi Barrage at Hanumannagar/Bhimnagar was completed in 1963. It regulates diversion into canals and provides a major transport crossing as part of the project system. Its gates and headworks are therefore important to irrigation as well as flood-season operations. A barrage is not a high storage dam. It does not create the same volume of flood-storage capacity as a large reservoir. Confusing barrage operation with reservoir flood control leads to exaggerated expectations about what gate opening alone can accomplish during a major basin-wide event. Canals, irrigation, and the development case for the project The Eastern and Western Kosi canal systems were intended to convert flood-control expenditure into agricultural benefit. Irrigation is therefore part of the project's historical ledger and should not disappear from an account focused on disasters. The official development narrative emphasized stabilized cultivation, multiple cropping, and wider economic integration. Canals also intersect with drainage. Cross-drainage structures, seepage, road embankments, local channels, and maintenance determine whether irrigation infrastructure improves or worsens waterlogging in a particular tract. The same linear infrastructure can deliver water to one area while obstructing evacuation of rain or seepage water elsewhere. Hydropower and the meaning of a multi-purpose river project The 1954 agreement and early project descriptions included hydroelectric generation as one of several objectives. The postwar planning ideal was that a river-development project could combine flood control, irrigation, power, erosion protection, transport, and regional modernization. Multi-purpose language can conceal trade-offs. Water levels, sediment management, canal deliveries, ecological flows, flood- season safety, and maintenance budgets do not always point toward the same operational choice. Historical evaluation should therefore separate each promised benefit rather than treating the project as one indivisible success or failure. Rehabilitation entered after engineering had already begun Field studies and later Kosi literature emphasize that the problem of people living between the new embankments was not adequately incorporated into the initial project design. Rehabilitation became an issue after construction was underway. Housing sites and grants were offered, but the distance between relocation sites and agricultural land made some arrangements difficult to sustain. This is a fundamental political-ecology point: an embankment does not merely protect a pre-existing population map. It creates inside, outside, adjacent, and downstream categories. People whose homes and fields fall on different sides of a structure may face a choice between physical safety, access to land, schools, markets, and kin networks. Life between the embankments Dinesh Kumar Mishra and the ICIMOD case study by Singh, Ghose, Chaudhary, and Hansda document communities living within the Kosi embankments, where seasonal inundation, erosion, sand deposition, and isolation can coexist with continued cultivation and strong attachment to place. Later counts vary with date and definition, so this book avoids presenting one population figure as timeless. The category 'inside the embankments' is itself diverse. Some villages occupy higher mounds; others lie close to active channels. Some households have boats, land, remittances, or political connections; others depend on wage labor and marginal sites. Risk must therefore be analyzed at village and household scales, not assigned uniformly from a map. 19 December 1966: the revised Kosi Agreement India and Nepal signed a revised Kosi Agreement at Kathmandu on 19 December 1966, superseding the earlier arrangement while retaining the project framework. The Ministry of External Affairs preserves the revised text together with the 1954 agreement as an official bilateral record. Revision is evidence that infrastructure treaties are not frozen at construction. Land, communications, compensation, maintenance, and operating arrangements can be renegotiated as experience exposes ambiguities or political concerns. Later disputes must therefore be checked against the revised text rather than quoted from the 1954 agreement alone. A breach chronology is an infrastructure history, not a flood list World Bank and engineering sources record major post- project breaches at several locations, including Dalwa in 1963, Jamalpur in 1968, Bhatania in the early 1970s, Bahuarawa in 1980, the Nauhatta/Hempur area in 1984, Joginia in 1991, and Kusaha in 2008. Some secondary lists differ on the year or exact naming of particular sites, which is why precise local claims require the underlying engineering record. The analytical point does not depend on forcing every event into one table. Repeated breaches demonstrate that the embankment system requires continuous inspection, erosion protection, emergency materials, access roads, and institutional readiness. The risk is not exhausted when initial construction ends. 1968: exceptional flow, breach, and the importance of measurement sites Bihar Water Resources Department records a historical 1968 maximum of about 913,000 cusecs at Barahkshetra and about 788,200 cusecs at Birpur Barrage. These are different measurement sites, so the numbers should not be substituted for one another. Major breaches also occurred in 1968. The distinction illustrates a broader rule for river history: a discharge number is meaningful only with location, time, measurement method, and context. Basin rainfall, tributary timing, sediment, channel geometry, gate operation, and embankment condition can all affect what a given discharge means downstream. 1984: breach memory and the persistence of sand-cast land The 1984 eastern-embankment failure near Nauhatta/Hempur became a major regional disaster and remains an important reference point in local studies. The ICIMOD case study selected Chandrain partly because the village had experienced the direct consequences of this breach and continued to live with altered land and water conditions. Disaster duration therefore exceeds the hydrograph. Sand casting can reduce soil productivity; erosion can remove titled land; drainage changes can persist; debt and migration may continue long after relief camps close. A history limited to fatalities and inundated hectares misses these slower forms of loss. 1991: Joginia and the politics of maintenance A breach near Joginia in Nepal in 1991 forms another important pre-2008 warning in the infrastructure record. The event reinforced the fact that critical reaches lie across the international border and that maintenance must be organized through access and cooperation rather than through a purely domestic chain of command. Maintenance politics is often less visible than construction politics. New barrages and bridges produce ceremonies; annual inspections, stone supply, spur repair, encroachment removal, and drainage clearance rarely do. Yet the latter determine whether protective infrastructure retains its function over decades. Diagram 50: selected legal, engineering, breach, and recovery anchors Diagram 50. Kosi infrastructure and risk: selected anchors, 1954-2026. The sequence distinguishes legal agreements, construction, breaches, recovery programmes, and continuing governance rather than treating them as one undifferentiated flood chronology. Original analytical diagram prepared for this book, 2026. © Gajendra Thakur, 2026. Sources: India-Nepal Kosi Agreements (1954, 1966); Yojana (1963); World Bank project records; Bihar WRD flood bulletins; Lok Sabha records; India and Nepal water- resource agencies. 18 August 2008: the Kusaha breach On 18 August 2008 the eastern afflux embankment failed near Kusaha in Sunsari district, Nepal, roughly twelve kilometres upstream of the Kosi Barrage. The river escaped the engineered corridor and spread into an older eastern course. The resulting inundation affected Nepal and large areas of Bihar, especially Supaul, Madhepura, Saharsa, Purnia, and Araria in the principal World Bank recovery documentation. The date and location are secure. Causal narratives require greater care. Scientific and administrative analyses discuss channel geometry, embankment erosion, sediment, maintenance, inspection, and the evolving relation between the river and protective works. Describing the disaster as an unpredictable act of nature alone is therefore inadequate. The 2008 event was an avulsion, not merely a very large seasonal flood Rajiv Sinha emphasized that the 2008 disaster was distinctive because the river avulsed into a different course. Water occupied an old channel belt outside the embanked river, producing a geography of exposure that routine flood-frequency language could not describe well. Later GIS work identifies the Kusaha reach as avulsion-prone in a broader assessment of channel geometry and planform dynamics (Sinha et al. 2014). This distinction also explains why historical channel maps matter to modern risk. A channel abandoned for decades can remain topographically relevant. Roads, settlements, and administrative boundaries may have developed across an older corridor whose hydraulic connectivity is reactivated when an avulsion occurs. Why discharge alone does not explain Kusaha The breach occurred at a discharge well below the Kosi system's highest recorded 1968 flows. That fact is important because it shows that structural failure cannot be inferred from discharge magnitude alone. Local bank attack, seepage, embankment condition, channel position, sedimentation, access for repair, and the timing of maintenance all matter. It is equally unsafe to reverse the argument and claim that one administrative lapse alone explains the whole avulsion. The disaster emerged from a coupled river-infrastructure system. Accountability questions can be investigated without reducing a complex geomorphic event to a single technical variable. The emergency geography of 2008 World Bank project documentation reports about 3.3 million people affected in Bihar, roughly one million evacuated, and hundreds of thousands sheltered in relief camps. Such figures are administrative estimates compiled for response and recovery, not a census of every individual movement. They are nevertheless strong evidence of the exceptional scale of disruption. Evacuation was not simply movement away from water. It required boats, roads, bridges, dry land, information, family reunification, livestock decisions, medical support, and camp management. The spatial arrangement of infrastructure determined who could leave quickly and who remained isolated. Relief camps, documentation, and unequal recovery Large camps can deliver food, water, health services, and temporary shelter at scale, but they also produce new problems of privacy, sanitation, gendered safety, disease exposure, and access to official registration. Households separated during evacuation may have different evidence of loss or land rights. Recovery programmes transform these documentary questions into eligibility decisions. A destroyed house on disputed land, a tenancy without formal papers, or a household split across migration routes may fit badly into standardized categories. Disaster governance therefore includes record- keeping as much as rescue. From emergency response to the Bihar Kosi Flood Recovery Project The World Bank approved a US$220 million credit in 2010 for the Bihar Kosi Flood Recovery Project; it became effective in March 2011. The programme combined housing reconstruction, roads and bridges, flood-management capacity, livelihoods, and emergency-response improvement rather than treating reconstruction as a housing programme alone. The project created a large administrative archive of beneficiaries, works, contracts, and institutional learning. Those records are valuable to historians, but project indicators should not be mistaken for the whole social outcome. Completion of a road or house is a measurable output; restored livelihood security, social inclusion, and long-term maintenance require additional evidence. Owner-driven housing reconstruction The recovery project used an owner-driven housing model in which eligible households were expected to manage reconstruction with financial and technical support. World Bank documentation presents this as a means of increasing household choice, ownership, and accountability compared with contractor-built standardized settlements. Owner-driven reconstruction still depends on access to land, materials, labor, banking, technical advice, and documentation. It can empower households with these resources while burdening those who face labor shortages, disability, migration, debt, or disputed tenure. The model should therefore be evaluated through differentiated beneficiary experience. Roads and bridges as disaster infrastructure By project completion, the Bihar Kosi Flood Recovery Project had reconstructed hundreds of kilometres of roads and dozens of bridges. Connectivity is not an ancillary benefit in a floodplain. Roads and bridges affect evacuation, market access, health referrals, school attendance, construction costs, and the speed at which emergency materials reach an embankment. Linear infrastructure can also obstruct drainage if cross- drainage is inadequate. The correct question is therefore not whether roads are good or bad for flood resilience, but how their elevation, culverts, bridges, maintenance, and alignment interact with local water movement. FMISC: institutionalizing flood information Bihar's Flood Management Improvement Support Centre was developed to strengthen hydrological observation, remote sensing, modelling, inundation mapping, embankment asset management, and public information. Its role expanded through the post-2008 recovery and basin-development programmes. This marks an important change in flood governance. Earlier systems depended heavily on gauge readings, field patrols, and administrative telegrams or phone chains. Contemporary management increasingly combines those practices with digital elevation models, satellite-derived inundation, telemetry, modelling, GIS, and public dashboards. Technology improves the information environment but does not remove the need for local interpretation and action. Kosi-FFEWS and the value of lead time The Kosi Flood Forecasting and Early Warning System operated by FMISC provides experimental forecasts of water level and discharge at seventeen sites on the Kosi and three tributaries, with a stated aim of at least seventy-two hours of lead time. It operates during the monsoon season and uses numerical weather prediction together with hydrological and hydraulic models. A forecast is useful only if uncertainty and action thresholds are understood. Lead time must be translated into messages that reach district offices, panchayats, embankment communities, boat operators, schools, health facilities, and households. Warning science and warning governance are therefore inseparable. Community participation in embankment surveillance FMISC project documents include community participation in embankment surveillance as a formal activity. The logic is straightforward: residents near an embankment may notice seepage, bank attack, animal burrows, unauthorized cutting, or rapid local change before a distant office receives a technical report. Community surveillance should complement, not replace, professional responsibility. Asking residents to report danger without providing communication channels, repair capacity, safe access, or feedback can transfer responsibility downward without transferring power. Effective participation requires a defined response chain. Sediment-management plans and the limits of a single engineering cure Post-2008 programmes commissioned flood-and-sediment studies, river-behavior analysis, surveys, and master planning. Recent scholarship has explored commercial and engineered uses of deposited sediment as well as targeted removal. These efforts reflect a recognition that embankment safety cannot be separated from channel morphology. Large-scale sediment extraction, however, can alter habitats, bank stability, groundwater interaction, and local livelihoods. Any proposal to 'desilt the Kosi' must specify where, how much, at what cost, with what disposal or market, and with what downstream consequence. A basin producing continual sediment cannot be managed as if one clearance operation permanently solves deposition. Drainage congestion: the flood that protection can create behind itself Communities outside an embankment may be protected from direct river inundation and still experience severe waterlogging. Rainwater, seepage, tributary flows, blocked channels, roads, canals, and insufficient sluices can trap water behind protective structures. ICIMOD fieldwork describes villages where chronic standing water limited cultivation and mobility despite their nominally protected position. This is why flood-prone area and protected area statistics must be interpreted carefully. Protection from a design river flood is not the same as freedom from water-related hazard. Drainage is a separate infrastructure system whose failure may be slower, less dramatic, and economically persistent. Sand casting, silt, and the uneven agronomy of floods Floodwater does not leave one uniform sediment. Fine silt can replenish soil fertility in some settings, while thick sand deposition can bury fields and make cultivation difficult for years. Erosion can remove land completely. The agrarian meaning of a flood therefore depends on sediment texture, depth, timing, crop stage, drainage, and access to labor and credit. This helps explain why memories of pre-embankment flooding can be ambivalent. Some cultivators remember annual inundation as part of a productive cycle, while others experienced destructive erosion or crop loss. Historical testimony should preserve that variation instead of turning 'living with floods' into either nostalgia or proof that structural protection is unnecessary. Livestock, fisheries, and floodplain livelihoods The Kosi floodplain supports more than crop agriculture. Livestock require dry standing space, fodder, veterinary access, and routes to market. ICIMOD fieldwork reported monsoon stress on cattle in waterlogged areas and the decline of some common grazing resources. Fishers and boat-dependent households may gain seasonal opportunities while facing different forms of insecurity in channel and wetland access. Flood policy that measures only cropped hectares therefore misses important livelihood systems. Fisheries, grazing, fuel, reeds, sand, ferrying, and seasonal labor connect households to changing river and wetland environments in ways that do not fit a simple protected-versus-flooded classification. Koshi Tappu: floodplain ecology across the Nepalese project landscape The Koshi Tappu Wildlife Reserve in Nepal was established in 1976 and became Nepal's first Ramsar-listed wetland in 1987. The reserve lies within the Saptakoshi floodplain in Sunsari and Saptari and protects riverine grassland, wetlands, wild water buffalo habitat, migratory birds, and other aquatic and terrestrial species. Its presence is a reminder that the Kosi is also an ecological corridor. Barrage operation, channel change, siltation, pollution, invasive species, flood pulses, and human use affect habitat. Flood management that seeks total hydraulic stability can conflict with ecological processes that depend on disturbance and changing water levels. Caste, class, and the micro-geography of vulnerability A 2023 Frontiers in Water study of Kosi villages in Darbhanga and Saharsa found that vulnerability was shaped not only by location relative to embankments but also by caste and economic class. In several study villages, better-off groups occupied higher central sites while Musahar, Mallah, and other marginalized households were more likely to occupy peripheral, low-lying, or embankment-edge locations (Sahani et al. 2023). The study is geographically specific and should not be universalized to every Kosi settlement. Its wider methodological lesson is robust: a village-level flood map can conceal inequality within the village. Elevation, landholding, housing quality, transport assets, political access, and social discrimination can place households exposed to the same river at very different levels of risk. Gendered care work and disaster risk Floods redistribute unpaid work. Securing drinking water, cooking with wet fuel, caring for children and older people, protecting documents, finding privacy and sanitation, managing menstruation, and maintaining household food supplies often fall disproportionately on women. Evacuation can also separate women from land titles, wage earners, or transport assets controlled by other household members. Gender should not be reduced to a statement that women are always more vulnerable. Women also organize savings groups, relief distribution, boats, household adaptation, and local communication. The correct historical question is how property rights, mobility, labor division, public authority, and relief design shape capacity at a particular time and place. Migration as adaptation, necessity, and structural outcome Seasonal and long-distance labor migration from the Kosi region predates any single flood. Repeated crop loss, uncertain land, weak local employment, and poor connectivity can nevertheless intensify the economic logic of migration. Remittances may finance raised plinths, repairs, education, food, and debt repayment, making migration part of household risk management. Migration also creates new vulnerabilities. A household may be divided when warning arrives; documents and compensation claims may be handled by people who remain behind; migrant workers can lose both destination employment and home assets during wider crises. Flood history therefore connects rural environmental change to urban and inter-state labor markets. Health, schooling, and the long tail of seasonal isolation Flood exposure is measured inadequately if analysis ends when water recedes. Roads may remain damaged, wells contaminated, clinics inaccessible, and schools used as shelters. Interrupted schooling can become permanent dropout where families move repeatedly or need children's labor. Maternal and emergency health risks increase when transport is by boat or foot over long distances. These effects are especially important for settlements between embankments or in chronically waterlogged tracts. A resilient infrastructure programme must therefore treat schools, health centres, drinking-water systems, raised access routes, and communications as part of flood risk management rather than as separate social sectors. Diagram 51: protection, chronic water, and residual breach risk Diagram 51. Embankment protection redistributes risk. This schematic separates chronic drainage and waterlogging, exposure inside the river corridor, and residual breach risk in protected tracts; it is not a surveyed cross- section and is not to scale. Original analytical diagram prepared for this book, 2026. © Gajendra Thakur, 2026. Analytical synthesis based on Mishra (2003); Singh et al. (2009); Sahani et al. (2023); FMISC and World Bank flood-management documentation. The Bihar Kosi Basin Development Project, 2015-2025 The World Bank-supported Bihar Kosi Basin Development Project followed the recovery programme with a broader emphasis on flood-risk management, connectivity, and livelihood development. By 2024 reporting recorded 28.14 kilometres of embankment work, more than 300 kilometres of rural roads, and more than fifty new bridges, together with agricultural and institutional components. A 2025 implementation-completion record documents final costs and results. The project illustrates a shift from post-disaster reconstruction toward basin resilience. It also shows the limits of large programme indicators: some agricultural and livelihood targets were revised or under-achieved even while infrastructure components advanced. A balanced history must preserve both completion claims and implementation shortfalls. September 2024: a high-flow stress test Heavy rain in Nepal and the region produced very high Kosi flows in late September 2024. Bihar Water Resources Department records a Birpur Barrage discharge of about 6.61 lakh cusecs on 29 September, the highest there since 1968, while Barahkshetra peaked still higher upstream during the episode. All barrage gates were opened and extensive monitoring and alerts followed. The event is useful because it demonstrates that extreme discharge and infrastructure failure are related but not identical. The barrage continued to function, yet sections of embankment elsewhere were under severe stress. Site-specific geometry and structure condition remain decisive. The 2024 Bhubhaul breach on the Western Kosi Embankment A Lok Sabha answer in 2025 records that the Western Kosi Embankment breached near Bhubhaul village in Kiratpur block at kilometre 38.210, damaging about 220 metres of embankment. The same answer states that overtopping associated with increased discharge caused the failure and gives a repair cost reported by the Government of Bihar. This episode belongs in the long breach history without being conflated with Kusaha. It occurred on a different reach, under a different hydraulic situation, and did not reproduce the 2008 avulsion geography. Comparing breaches requires location- specific evidence rather than a single narrative of institutional failure. 2025-26: maintenance remains an annual governance cycle Government of India reporting for 2025-26 describes annual post-monsoon inspection and recommended flood-protection works for the Kosi, including works on Nepalese territory reimbursed through the border-area flood-management programme. The Kosi High Level Committee continues to inspect and recommend anti-erosion and maintenance works before each flood season. This recurrent cycle is historically significant. The Kosi Project is not a completed object from 1963; it is an infrastructure system that must be reproduced every year through inspection, procurement, stone and earth supply, spur repair, access, monitoring, and emergency readiness. India-Nepal joint mechanisms after the original treaty Bilateral water governance now includes the Joint Committee on Water Resources, the Joint Standing Technical Committee, the Joint Committee on Inundation and Flood Management, and the Joint Committee on Kosi and Gandak Projects, alongside specialized technical and project mechanisms. These institutions supplement rather than replace the 1954/1966 legal framework. Their existence also corrects the impression that transboundary management occurs only during crisis diplomacy. Routine meetings address maintenance, leased land, encroachment, canal and drainage issues, flood forecasting, and project operations. The difficulty is institutional continuity: agreements at committee level must still be translated into timely field action. The 2026 joint committee and operational details In 2026 Nepal's Department of Water Resources and Irrigation publicized the eleventh meeting of the Nepal-India Joint Committee on Kosi and Gandak Projects. Public reporting from the meeting addressed pre-flood erosion-control materials, encroachments around project infrastructure, leased-land demarcation, drainage and waterlogging, information exchange, and satellite imagery. These details are more revealing than a generic statement of bilateral friendship or dispute. They show the mundane administrative work required to keep a transboundary hydraulic system operating: who can enter a site, where materials come from, how land boundaries are marked, and which agency answers when water cannot drain. Climate change: necessary context, dangerous shortcut Climate change is altering temperature, precipitation extremes, snow and glacier processes, and the probability of compound hazards across the Himalaya. These changes can modify the future operating environment of the Kosi system. They strengthen the case for flexible design standards, better forecasts, updated hydrology, and adaptation planning. They do not justify retroactively labeling every historical breach a climate-change disaster. The 2008 Kusaha event requires specific analysis of channel and embankment conditions; the 2024 event requires its own rainfall and hydraulic evidence. Climate attribution should be event-specific where possible and contextual where it is not. Glaciers, monsoon rainfall, and the problem of one-source explanations Because parts of the Kosi basin reach the high Himalaya, glacier and snow processes matter to long-term hydrology. Yet the lower-basin flood season is strongly shaped by monsoon rainfall and tributary synchronization. Public narratives sometimes leap from 'Himalayan river' to 'glacier flood' without showing the actual mechanism of a particular event. A disciplined basin history asks which part of the hydrograph came from rainfall, snowmelt, glacier melt, or an exceptional high-mountain event, and over what timescale. It also asks how that water interacted with sediment, floodplain storage, embankments, and drainage. Hazard sources are multiple and their relative importance changes by event. The proposed Sapta Kosi High Dam: proposal is not infrastructure history A Sapta Kosi High Dam Multipurpose Project and Sun Kosi storage-cum-diversion concept have remained on the India- Nepal bilateral agenda, with a joint project office established for surveys and preparation of a detailed project report. The proposal is often invoked in debates over flood control, power, irrigation, displacement, seismic risk, and benefit sharing. This book treats it as a proposal and negotiation history, not as an existing flood-control structure. Statements about the amount of flooding it would prevent or the social cost it would impose depend on design, operating rules, sedimentation, reservoir extent, seismic assessment, and final agreement. Those cannot be inferred from the existence of a study office. Nature-based approaches and floodplain room Recent resilience programmes increasingly discuss nature- based approaches alongside embankment strengthening. In a Kosi context, such approaches can include protecting wetlands, preserving drainage corridors, maintaining floodplain storage, stabilizing selected banks with vegetation, and designing infrastructure to work with rather than against local water pathways. 'Nature-based' should not become another slogan. Wetland restoration cannot substitute for repairing a dangerous breach; an embankment cannot substitute for drainage; relocation cannot be ordered without land and livelihood rights. A layered strategy is more realistic than replacing one universal solution with another. Remote sensing, inundation maps, and a new documentary archive Modern flood history is increasingly documented from above. FMISC uses remote-sensing products and has developed inundation and forecasting systems; satellite images of the 2008 avulsion became central to scientific explanation because they showed the new flow path across the megafan. Digital elevation data help identify low corridors that may not be obvious from administrative maps. Remote sensing has limits. A pixel cannot reveal who owns a field, whether a household has a boat, whether a road culvert is blocked, or whether a relief message reached a Dalit hamlet. The strongest risk archive combines satellite observation with gauges, engineering surveys, cadastral and census data, and field testimony. Public dashboards and the politics of open hydrological data By 2026 Bihar's Water Resources Department and FMISC were publishing current barrage discharges, flood bulletins, and forecasting products online. Open data allows journalists, researchers, local administrations, and residents to compare current readings with historical peaks and to verify some official claims quickly. Data publication is not the same as risk communication. Technical units, update times, site names, trend labels, forecast uncertainty, and network connectivity all affect usability. A public dashboard becomes effective disaster infrastructure only when its information can be understood and acted upon beyond specialist offices. Protected-area statistics require interpretive caution Bihar Water Resources Department publishes basin-wise figures for flood-prone and protected areas; for the Kosi/Adhwara grouping it reports extensive embankment length and a large protected area. Such tables are useful for comparing the administrative scale of flood-control works across basins. They should not be read as a map of household safety. 'Protected' generally refers to the intended effect of flood-control works, not a guarantee against drainage congestion, erosion, embankment failure, or exposure within an embanked corridor. Statistical categories need to be reconciled with local topography and actual hazard pathways. Against the protection-versus-river binary Kosi politics is sometimes written as a conflict between engineers who want embankments and activists who want an unconstrained river. Local experience is less binary. A household may demand urgent repair of a threatened embankment while also criticizing waterlogging produced by the same system. Farmers may value canal water and resent blocked drainage. Fishers may depend on seasonal connectivity that a protection work interrupts. Policy should therefore be evaluated reach by reach and function by function. Maintenance, drainage, warning, evacuation, insurance, land rights, wetland protection, controlled inundation, channel training, and relocation can be combined differently in different places. Environmental justice and the distribution of residual risk Infrastructure always leaves residual risk. The central political question is who carries it. Landowners on higher ground, households beside a road, residents within the embankments, people in waterlogged exterior tracts, and those below a breach-prone reach do not receive the same combination of benefit and danger. Caste, class, gender, and political voice shape both exposure and recovery. Environmental justice does not require the claim that every unequal outcome was deliberately designed. It asks whether the distribution of protection, damage, compensation, participation, and long-term opportunity is unequal, and whether institutions recognize those inequalities when allocating resources. A basin-scale history without erasing local responsibility Because the Kosi crosses borders, basin-scale analysis is essential. Upstream rainfall, sediment, barrage operation, embankments in Nepal, canals, Bihar floodplains, and the Ganga confluence are physically connected. Yet 'the basin' should not become a device for dissolving responsibility into complexity. Specific agencies still have specific duties. A treaty assigns functions; a department owns an asset; a contractor completes a work; a district issues an evacuation order; a panchayat identifies a shelter. Basin thinking is most useful when it clarifies how these responsibilities connect rather than when it makes accountability impossible. What the Kosi teaches the wider history of Mithila The history of Mithila cannot be narrated only through dynasties, languages, courts, and literary canons. River management has changed the geography within which citizenship, caste hierarchy, agriculture, migration, and state authority operate. Embankments and canals are therefore part of social history as much as engineering history. The Kosi also resists narratives of fixed regional space. Villages move, channels change, land appears and disappears, and infrastructure redraws practical boundaries inside a formally stable district map. Environmental history is not an appendix to regional history; it helps explain how the region itself is continually materialized. Chapter conclusion: from flood control to risk governance The Kosi Project reduced some forms of routine inundation and created irrigation and transport infrastructure, but it also generated new dependencies on maintenance, drainage, sediment management, warning, and transboundary coordination. Its history is therefore neither a simple engineering triumph nor a simple proof that engineering fails. The strongest contemporary framework is risk governance: maintain structures that communities depend upon, identify where protection transfers water and sediment problems, improve drainage and forecasts, preserve ecological functions, document unequal vulnerability, and keep recovery rights visible before the next emergency. The 2008 and 2024 episodes show why the system must be understood as living infrastructure rather than a project completed in the twentieth century. Chapter-specific bibliography Bihar State Archives. 'Flood in Colonial Bihar (1871-1939): Selected Digitized Collection.' Government of Bihar. Chakraborty, Tapan, Rimpal Kar, Parthasarathi Ghosh, and Sounak Basu. 'Kosi Megafan: Historical Records, Geomorphology and the Recent Avulsion of the Kosi River.' Quaternary International 227, no. 2 (2010): 143-160. https://doi.org/10.1016/j.quaint.2009.12.002. Government of India and Government of Nepal. Agreement on the Kosi Project. Kathmandu, 25 April 1954. Reproduced in India Bilateral Treaties and Agreements, Ministry of External Affairs, Government of India. Government of India and Government of Nepal. Revised Agreement concerning the Kosi Project. Kathmandu, 19 December 1966. Ministry of External Affairs, Government of India. https://www.mea.gov.in/Portal/LegalTreatiesDoc/NP66B1487.pdf. Government of India, Ministry of Jal Shakti. 'Bilateral Mechanism for Flood Mitigation with Nepal.' Parliamentary information/Press Information Bureau, 19 December 2022. Government of India, Ministry of Jal Shakti. Standing Committee and departmental records on India-Nepal flood-management mechanisms, Kosi High Level Committee, and border-area maintenance, 2024-26. Government of India, Lok Sabha. Answer to Unstarred Question 3061 on the Western Kosi Embankment breach and Kosi Barrage discharge, 2025. Government of Bihar, Water Resources Department. Flood bulletins and River Discharge Dashboard, including Kosi/Barahkshetra and Birpur Barrage observations for 2024-26. Government of Bihar, Water Resources Department, Flood Management Improvement Support Centre. Kosi Flood Forecasting and Early Warning System; Flood Management Information System documentation and technical programme records. Hill, Christopher V. River of Sorrow: Environment and Social Control in Riparian North India, 1770-1994. Association for Asian Studies, 1997. Mishra, Dinesh Kumar. 'Life Within the Kosi Embankments.' Water Nepal 10, no. 1 (2003): 277-301. National Disaster Management Authority. National Disaster Management Guidelines: Management of Floods. Government of India, 2008. Nepal, Department of National Parks and Wildlife Conservation. Koshi Tappu Wildlife Reserve official profile and management documentation. Nepal, Department of Water Resources and Irrigation. Records of the Nepal-India Joint Committee on Kosi and Gandak Projects, including the eleventh meeting, 2026. Sahani, Ranjeet Kumar, Shrinivas Badiger, Abhishek Samrat, and Siddhartha Krishnan. 'Flood Frequency and Flood Intensity Changes in the Post Embankment Period in the Kosi Sub-basin India: Impact of Location, Caste, and Class on the Flood Vulnerability of the Marginal Communities.' Frontiers in Water 5 (2023): 1017945. https://doi.org/10.3389/frwa.2023.1017945. Singh, Praveen, Nilanjan Ghose, Nitin Chaudhary, and Regina Hansda. Life in the Shadow of Embankments: Turning Lost Lands into Assets in the Koshi Basin of Bihar, India. International Centre for Integrated Mountain Development, 2009. https://doi.org/10.53055/ICIMOD.507. Sinha, Rajiv. 'The Great Avulsion of Kosi on 18 August 2008.' Current Science 97, no. 3 (2009): 429-433. Sinha, R., K. Sripriyanka, Vikrant Jain, and Malay Mukul. 'Avulsion Threshold and Planform Dynamics of the Kosi River in North Bihar (India) and Nepal: A GIS Framework.' Geomorphology 216 (2014): 157- 170. https://doi.org/10.1016/j.geomorph.2014.03.035. World Bank. Bihar Kosi Flood Recovery Project: Project Appraisal, Needs Assessment, Owner-Driven Housing Reconstruction, and Implementation Completion and Results documentation, 2010-2018. World Bank. Bihar Kosi Basin Development Project: Environmental and Social Management Framework, Implementation Status and Results Reports, and Implementation Completion and Results documentation, 2015-2025. Yojana. 'Kosi Progress.' September 29, 1963, p. 27. Publications Division, Government of India.