Full chapter text
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.