Flood Adaptation and Resilience

Chapter 130 examines flood adaptation as the capacity to live with a dynamic river system without accepting avoidable death, repeated destitution or permanent service collapse. Mithila, Vajji and Anga occupy one of South Asia’s most intensively settled floodplains. In north Bihar, rivers descending from Nepal carry high sediment loads into a very low-gradient plain; the Bihar State Disaster Management Authority describes 73.63 per cent of North Bihar’s geographical area as flood-prone. Yet ‘flood’ is not one phenomenon. Flash floods, river floods, drainage congestion and chronic waterlogging create different lead times, depths, durations and recovery problems. Resilience must therefore be matched to flood type rather than reduced to a universal embankment solution. The institutional landscape has changed substantially since the 2008 Kosi disaster. Bihar’s Flood Management Improvement Support Centre now combines hydrometeorological data, inundation mapping, embankment asset management and forecasting. Its Kosi Flood Forecasting and Early Warning System is designed to provide at least 72 hours of lead time at multiple sites on the Kosi and tributaries. Across the border, Nepal’s Department of Hydrology and Meteorology issues daily 72-hour flood bulletins using telemetric rainfall and water-level networks together with basin models and regional forecast products. These capabilities are important, but resilience is created only when forecasts become trusted warnings, warnings become action, and infrastructure, households and public services can absorb the shock and recover. The chapter consequently treats flood management as a portfolio. Structural measures—embankments, spurs, revetments, drainage works, culverts and raised infrastructure—remain indispensable in many locations. Non-structural measures—forecasting, hazard maps, land-use rules, shelters, drills, insurance, health preparedness and community organization—address the residual risk that structures cannot eliminate. Wetlands and flood-storage areas add a third layer by working with floodplain hydrology. The goal is not a fantasy of a flood-free river basin. It is an adaptive system in which protection, preparedness, response, recovery and learning continually reduce risk. 130.1 Flood adaptation means living with water without normalising loss Adaptation begins by rejecting two extremes: the belief that every flood can be prevented and the belief that recurring loss is simply natural. Rivers need space, but people also need safe settlement, reliable transport, schools, health services and livelihoods. A resilient floodplain therefore accepts seasonal water where it can be accommodated while defending critical assets and giving households time and resources to act. The relevant outcome is not only peak water level. It is mortality, crop loss, days of school closure, isolation of health facilities, livestock survival, contamination of drinking water, debt after the event and the time needed to restore normal activity. This shifts evaluation from kilometres of embankment constructed to the performance of the whole social-ecological system. 130.2 Four flood regimes require four different responses Bihar’s disaster-management framework distinguishes flash floods associated with rainfall in Nepal, river floods with longer but still limited lead time, drainage-congestion floods near confluences, and permanently waterlogged areas. The distinction is operationally important. A flash-flood corridor needs fast upstream sensing, redundant warning and pre-identified evacuation routes. A river flood permits more staged relocation of livestock and movable assets. Drainage congestion requires outlet management, culvert capacity, pumping or channel restoration as well as flood protection. Chronic waterlogging is closer to a HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II land-and-water management problem than an emergency alone. Treating all four regimes as the same event produces poor investments because speed, duration and safe refuge differ radically. 130.3 The basin is transboundary even when administration is not The major north Bihar rivers cannot be understood inside a single administrative boundary. Rainfall, landslides, sediment and river discharge generated in Nepal move downstream into Bihar, while downstream embankments and drainage conditions influence where water spreads and how long it remains. The Koshi, Kamla, Bagmati and other systems are therefore hydrological continuities crossing political jurisdictions. Adaptation requires routine exchange of observations, forecasts and operational information without implying that the two countries share identical institutions or legal responsibilities. Cross-border cultural and market ties can accelerate communication, but informal personal networks are not a substitute for dependable public protocols. Basin geography makes cooperation a technical necessity rather than an optional diplomatic add-on. 130.4 The 2008 Kosi breach as an institutional turning point The breach of the eastern afflux bund near Kusaha in Nepal on 18 August 2008 redirected Kosi flows and caused catastrophic inundation across parts of Bihar. The disaster demonstrated how low-probability infrastructure failure can overwhelm assumptions based on normal monsoon operations. Recovery therefore had to address both destroyed assets and the institutions that manage risk. World Bank-supported reconstruction included more resilient housing and infrastructure, while subsequent basin-development work strengthened embankments, forecasting and flood-management capacity. The historical lesson is not that one project ‘solved’ the Kosi. It is that a breach converts maintenance, inspection, emergency communication and contingency planning from technical back-office functions into central components of public safety. 130.5 Embankments protect, redistribute and concentrate risk Embankments can protect dense settlements, roads and fields from frequent inundation and remain a core part of Bihar’s flood-management system. Their benefits are real, but the risk they create must also be managed. When water is excluded from the floodplain, sediment may remain within the embanked channel, drainage from outside may be impeded, and a breach can release water rapidly into areas whose residents have come to depend on protection. Embankments therefore require inspection, maintenance, erosion control, access for repair, monitoring of river attack and explicit breach scenarios. The correct comparison is not ‘embankment versus no embankment’ in the abstract; it is the performance of a maintained embankment system combined with drainage, forecasting and emergency planning. 130.6 Embankment asset management turns maintenance into risk reduction A structure is not resilient merely because it was once built to a design standard. Its condition changes through erosion, animal burrows, unauthorized cuts, vegetation, seepage, settlement, river migration and repeated high-water seasons. FMISC’s embankment asset-management approach is therefore significant: it treats flood-protection works as an inventory of assets whose condition, vulnerability and maintenance needs can be recorded and prioritized. Risk-based maintenance is economically preferable to waiting for visible failure because the consequences of a breach can be orders of magnitude larger than routine repair. Asset registers also create institutional memory when staff rotate. The long-term challenge is to connect inspection data to budgets and timely works rather than leave it as a mapping exercise. 13331333 GAJENDRA THAKUR 130.7 Sediment and channel change are central, not secondary Himalayan rivers carry large sediment loads, and the transition from steep uplands to the Gangetic plain encourages deposition and channel instability. Flood resilience must therefore include sediment management and geomorphology. A river can become more hazardous even without a larger rainfall event if aggradation reduces conveyance, shoals redirect flow toward an embankment, or a channel migrates toward a settlement. Dredging is not a universal remedy because sediment volumes are immense and intervention can simply move the problem. Better practice combines surveys, remote sensing, cross-sections, local observation and basin modelling to understand where erosion and deposition are changing risk. This is why flood management cannot be reduced to rainfall prediction alone. 130.8 Drainage congestion can make protected land remain flooded A field or village outside an embankment may still flood because local rainfall and tributary water cannot escape when the receiving river is high. Roads without adequate cross-drainage, silted channels, blocked culverts and poorly maintained sluices can deepen this problem. The resulting waterlogging may last far longer than the river peak, damaging crops, contaminating water sources and delaying access to schools and markets. Drainage is therefore the less visible half of flood protection. Resilience planning should map natural drainage lines, protect outlets from encroachment, size culverts for present hydrology, maintain sluice gates and identify locations where temporary pumping is justified. A successful embankment that creates chronic interior waterlogging is not a successful flood-resilience system. Figure 516 — Flood resilience as a chain from monitoring and warning through protection, continuity, recovery and adaptation. 130.9 Forecast lead time is an economic asset Forecasting creates value by converting hydrological knowledge into time. A warning received hours or days before dangerous water arrives can move cattle, seed, documents, medicines, pumps and vehicles; pre- position boats; close unsafe roads; protect transformers; evacuate older people; and relocate hospital supplies. Lead time should therefore be valued in avoided loss, not only forecast accuracy. But a long nominal lead time is useless if the message arrives late or is not trusted. Operational performance must track the entire chain from gauge to model to district to village. False alarms, unexplained uncertainty and warnings that do not specify expected action can reduce compliance. Good systems communicate both hazard level and what recipients should do before the next update. 130.10 Kosi-FFEWS and the move toward 72-hour forecasting Bihar’s Kosi Flood Forecasting and Early Warning System was developed to provide a lead time of at least 72 hours using validated hydrological and hydraulic models. FMISC describes experimental forecasts for water level and discharge at seventeen sites on the Kosi and three tributaries—Kamla Balan, Bhutahi Balan HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II and Khando/Tiljuga—during the monsoon season. This is a major change from dependence on immediate gauge reading alone. Modelled forecasts can extend decision time, but they must be continuously tested against observed water levels and changing channel geometry. A 72-hour horizon should be treated probabilistically: uncertainty generally increases with lead time, so response plans need thresholds for watch, preparation and evacuation rather than a single deterministic number. 130.11 Data integration: rainfall, river level, inundation and asset condition FMISC’s current public flood-information architecture brings together weather reports, Nepal rainfall and water-level information, flood bulletins, 72-hour water-level forecasts, barrage status, inundation maps and Central Water Commission products. The analytical importance of this integration is that no single dataset is sufficient. Rainfall indicates potential runoff; river levels show the current hydrological state; forecasts estimate future conditions; satellite imagery shows where water actually spread; and asset inspections reveal whether a protection line is physically vulnerable. Resilience increases when these datasets are linked spatially and temporally so that a district can ask not only ‘how high will the river rise?’ but ‘which road, embankment reach, settlement and service facility is exposed if that forecast materialises?’ 130.12 Nepal’s telemetric system expands the upstream window Nepal’s Department of Hydrology and Meteorology has developed a national flood-forecasting system that issues daily forecasts for the following three days during the monsoon. A July 2025 bulletin described real-time observation from roughly 260 rainfall stations and about 180 water-level stations, supplemented by basin models, numerical weather prediction, regional outlooks and global flood products. For Bihar, upstream observations are especially valuable because intense rain in Nepal can precede downstream flooding. The resilience benefit depends on interoperability: station identifiers, thresholds, timestamps and warning categories must be understandable across agencies. Data sharing becomes most valuable when it is routine before a crisis, because emergency improvisation is vulnerable to communication failure precisely when conditions are most severe. 130.13 The last mile is where forecast skill becomes social protection A technically accurate forecast can fail socially if it stops at a website or district office. Last-mile warning requires a chain that works during power cuts, weak mobile coverage, night-time events and language diversity. Sirens, public-address systems, mobile messages, local radio, ward representatives, school networks and trained volunteers should therefore be redundant rather than mutually exclusive. Messages need location, expected timing, likely severity and an action instruction. Communities also need to know the safe route and the destination before the warning arrives. The strongest warning system is one rehearsed through drills, because a familiar procedure reduces hesitation and makes responsibilities visible. Trust is built by consistent updates, including when a forecast is revised downward. 130.14 Community-based early warning adds local sensing and ownership The Ratu River experience across Nepal and Bihar illustrates the value of community-based flood early warning. ICIMOD and government partners established systems in upstream Nepali locations and downstream Bihar communities so that rising water could be communicated before arrival. Community systems do more than install instruments: local caretakers observe gauges, relay information, maintain equipment and know which households require assistance. This creates ownership and can provide warning where formal networks are sparse. The limitation is sustainability. Volunteer turnover, equipment failure and 13351335 GAJENDRA THAKUR uncertain maintenance finance can erode performance. Community-based systems should therefore complement—not replace—public forecasting, with clear responsibility for calibration, communication and integration into district disaster plans. Figure 518 — Transboundary flood warning becomes useful only when upstream observation is converted into a trusted last-mile message and timely household action. 130.15 Evacuation is a logistics system, not a final instruction ‘Move to higher ground’ is only the first line of an evacuation plan. Households need routes that remain passable as water rises, transport for people who cannot walk, arrangements for livestock, secure shelter space, drinking water, toilets, lighting, privacy, cooking and medicines. Schools and public buildings often become shelters, so their structural safety and continued educational function must be planned together. Boats require trained operators and fuel; rescue teams require local maps and communication. Evacuation planning should also account for people who refuse to leave because they fear theft or loss of cattle. Resilience improves when shelters protect both life and the assets on which recovery depends, reducing the perceived cost of complying with a warning. 130.16 Livestock survival is central to rural recovery Cattle, buffalo, goats and poultry are productive assets, savings and sources of food. Flood plans that evacuate people but ignore animals can produce large post-disaster losses or cause households to delay evacuation. Safe raised sites, fodder reserves, veterinary care, carcass disposal and access to clean water should therefore be part of preparedness. Fodder shortages often appear after the flood because stored straw is soaked and grazing land remains submerged. Livestock disease risk also rises when animals crowd on embankments or raised roads. The economic value of animal protection is particularly high for smallholders because a lost dairy animal may represent years of accumulated household capital. Flood-resilient agriculture thus includes animal logistics as well as crop recovery. 130.17 Housing resilience depends on site, plinth, materials and recoverability Rebuilding after the Kosi disaster demonstrated the importance of housing that can resist or recover from flooding. A stronger wall alone is insufficient if the house is in a high-velocity channel or the plinth lies below recurrent water levels. Site selection, raised plinths where appropriate, drainage around the dwelling, safe electrical placement, water-resistant lower-wall materials and secure storage for documents and grain can reduce damage. Owner-driven reconstruction can incorporate household preferences and local knowledge, but it needs technical assistance and transparent finance. Because many families modify houses over time, resilience guidance must also apply to incremental construction. The useful metric is not whether a house HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II remains completely dry; it is whether occupants survive, critical contents are protected and the building can be cleaned and reoccupied quickly. 130.18 Roads and culverts determine whether a flood becomes isolation Transport infrastructure can be both protection and obstruction. Raised roads provide evacuation routes and temporary refuge, but they can act as unintended embankments if cross-drainage is inadequate. Washed- out approaches, undersized culverts and overtopped bridges can isolate villages even after rainfall stops. Resilient road design therefore needs hydrological sizing, erosion protection, multiple drainage openings and maintenance of approaches. Network redundancy matters: a district should know which alternative route reaches a health centre if the primary road is cut. Post-flood repair priorities should restore connectivity based on service dependence, not only traffic volume. In a dense rural economy, a short broken link can interrupt milk collection, school attendance, market access and emergency care across a much larger area. 130.19 Flood-resilient health and WASH prevent the second disaster Deaths and illness after inundation can come from contaminated water, interrupted treatment, snakebite, diarrhoeal disease, skin infections and loss of access to maternal care rather than from drowning alone. Flood- resilient water and sanitation therefore require protected or elevated hand-pumps where feasible, chlorination supplies, safe temporary toilets, waste management and rapid testing of drinking-water sources. Health facilities need backup power, medicine stocks, cold-chain continuity and referral routes. Pregnant women, dialysis patients, people using regular medicines and persons with mobility limitations should be identified before the monsoon, with privacy safeguards. Public-health resilience is measured by continuity: how quickly safe water, routine treatment and surveillance continue when the normal physical network is disrupted. 130.20 Crop calendars can convert warning into agronomic action Flood forecasts have agricultural value when they connect to crop stage. A warning before nursery establishment may change sowing location; before transplanting it may delay labour hiring; before harvest it may justify early cutting and rapid removal; before a forecasted prolonged inundation it may trigger relocation of pumps, fertilizer and machinery. Farmers also need contingency seed for re-sowing after water recedes. The same water depth can be tolerable for one crop stage and destructive for another, so district advisories should combine hydrological forecasts with agronomic calendars. This is a practical bridge between Chapter 129’s climate-risk analysis and flood resilience: resilience is not a separate disaster sector but a set of adjustments embedded in farm decisions. 130.21 Submergence-tolerant crops and recovery packages reduce residual loss No protection system can prevent all field inundation. Crop resilience therefore includes varieties and management practices that reduce loss when water enters the farm. Submergence-tolerant rice can survive specific periods of complete inundation better than susceptible varieties, while short-duration crops can help recover a delayed season after floodwater retreats. Seed reserves, rapid soil assessment, drainage restoration, livestock feed and access to small credit are part of the same recovery package. The emphasis should remain crop- and location-specific: a variety tolerant of submergence may not solve prolonged waterlogging, late- season lodging or drought after flood recession. Resilience comes from matching biological traits to the actual flood regime and keeping replacement inputs available when commercial supply chains are disrupted. 13371337 GAJENDRA THAKUR Figure 517 — Flood resilience depends on a portfolio of structural protection, forecasting, drainage, emergency services, flood-storage landscapes and household finance. 130.22 Credit, insurance and social protection shape the speed of recovery A household that survives physically can still enter a debt spiral if crops, livestock, tools and stored grain are lost together. Recovery finance therefore determines whether a flood becomes a temporary shock or long- term impoverishment. Crop insurance, emergency grants, concessional credit, public employment and portable food or cash benefits can stabilize consumption while production restarts. Design matters: delayed compensation loses much of its value; documentation requirements can exclude people whose papers were damaged; tenants and sharecroppers may not appear in land records; migrants may be absent when local lists are prepared. Resilient social protection should therefore be pre-arranged, portable and linked to transparent damage assessment rather than invented after each event. 130.23 Vulnerability is differentiated by gender, age, disability and caste-class position Flood depth alone does not determine who is harmed. People who cannot swim, older adults, pregnant women, small children, persons with disabilities, landless households and families living on low-cost marginal sites may face greater evacuation and recovery burdens. Women often carry responsibility for children, food, water and livestock while having less control over transport or cash. Social hierarchy can influence access to boats, shelters, information and compensation. Inclusive preparedness therefore requires disaggregated risk mapping, accessible shelters, targeted evacuation assistance and representation of vulnerable groups in local planning. The objective is not to label communities as helpless; it is to identify where unequal resources convert the same hazard into unequal loss. 130.24 Migration is both an adaptation mechanism and a recovery constraint Migration can diversify household income before a flood and provide remittances for rebuilding afterward. Migrants may also return with cash, contacts or transport. But absence can leave older adults and women managing evacuation and farm recovery with fewer working-age members. A flood at origin can coincide with precarious employment at destination, limiting the remittance buffer precisely when it is needed. Recovery systems should therefore recognise multi-local households. Portable identity, bank access, digital transfers and communication with migrants can speed assistance. Long-term displacement after erosion differs from circular labour migration and requires land, housing and service solutions. Flood HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II resilience should not treat all mobility as either failure or success; its role depends on choice, safety and the household resources that make movement possible. 130.25 Wetlands, chaurs and flood-storage landscapes are infrastructure Floodplains contain natural and semi-natural depressions, wetlands, oxbows and chaurs that temporarily store water, support fisheries and moderate local drainage. Filling or disconnecting these spaces can shift water elsewhere and increase waterlogging even when the development itself remains dry. Treating wetlands as infrastructure does not mean freezing all land use. It means mapping storage function before conversion, protecting key drainage corridors and designing roads and settlements so that water can move through the landscape. Where possible, seasonal storage can be combined with fisheries, biodiversity and groundwater recharge. The economic comparison should include avoided flood damage and ecosystem services, not only the immediate market value of filled land. 130.26 Floodplain zoning is a risk-allocation policy Land-use regulation determines who will be exposed to future floods. Hospitals, power substations, schools used as shelters, hazardous-material stores and dense housing should not be sited using the same risk tolerance as seasonal agriculture. Floodplain zoning can distinguish areas where building is prohibited, restricted, elevated or required to meet special standards. Implementation is difficult in densely settled regions with insecure tenure and strong land-market pressure, so zoning must be accompanied by transparent maps, compensation or relocation mechanisms where rights are affected, and regular updates as rivers change course. Otherwise zoning becomes either unenforced paper or a tool applied selectively against weaker residents. Resilience requires both technical hazard information and procedural fairness. 130.27 Nature-based solutions complement engineered protection Nature-based approaches use ecological processes to reduce risk or the consequences of risk: restoring flood-storage areas, reconnecting selected wetlands, vegetating erosion-prone surfaces, maintaining riparian buffers and designing space for water where settlement density allows. The World Bank’s recent description of Kosi-basin work explicitly includes nature-based solutions alongside strengthened embankments. Such measures are not substitutes for critical engineered works at dense settlements or infrastructure. Their value is complementary: they can slow local runoff, store water, reduce erosion and provide livelihood or biodiversity benefits. Performance should be monitored with the same discipline applied to concrete structures. ‘Nature- based’ should not become a label for projects whose hydrological effect is assumed rather than measured. 130.28 India–Nepal warning cooperation should be institutional and redundant The transboundary flood-warning chain currently draws on national hydrological agencies, district administrations, basin projects and community networks. ICIMOD’s 2026 field reporting from North Bihar noted that informal WhatsApp communication connects upstream Nepali and downstream Indian stakeholders, but also that informal mechanisms are not a reliable substitute for a formal cross-border system. Communities reported that longer lead times—around a day where feasible—would materially improve preparation. Institutional resilience therefore means agreed contacts, routine data exchange, shared threshold understanding, multilingual message formats and fallback channels when one system fails. Cooperation should be practiced during ordinary monsoon days, not activated only at extreme water levels. 13391339 GAJENDRA THAKUR 130.29 Resilience must be measured by outcomes across the full flood cycle A useful resilience dashboard should track more than structures completed. Before the event it can measure forecast lead time, warning reach, shelter readiness, embankment condition and the share of vulnerable households with an evacuation plan. During the event it can record deaths, rescue time, road isolation, service interruption and livestock loss. Afterward it can measure days to restore drinking water, school, health care, electricity and market access; time to compensation; crop re-sowing; household debt; and permanent displacement. These metrics expose trade-offs. A project may reduce inundated area but worsen drainage duration elsewhere; a warning may be accurate but reach only connected households. Outcome measurement turns resilience from rhetoric into a falsifiable institutional claim. 130.30 From flood control to adaptive basin governance The strongest conclusion of this chapter is that flood resilience is a governance process. Rivers change, sediment moves, climate alters rainfall extremes, embankments age, settlements expand and communication technology evolves. No master plan can therefore remain final. Institutions need an annual cycle of observation, forecast verification, asset inspection, community drills, monsoon response, post-event review and budgeted correction. Bihar’s FMISC, Nepal’s DHM network, community early-warning experience and basin projects show the components of such a system, but their value lies in integration and maintenance. The next chapter shifts from hazard management to the economy of the India–Nepal border after federalisation, where the same roads, markets, financial networks and administrative boundaries that shape resilience also shape everyday exchange. Figure 519 — Adaptive basin governance treats every monsoon as new evidence for observing, anticipating, protecting, responding, recovering and revising the next plan. Table 130.1 — Evidence architecture for analysing flood adaptation and resilience Evidence source What it establishes Decision / historical Main limitation use Hydrometeorological rainfall, river level, forecast initiation and station gaps, telemetry gauges discharge and timing threshold monitoring failure and local representativeness Hydrological / future water level, lead-time decisions and uncertainty rises with hydraulic models discharge and contingency planning lead time and changing inundation scenarios channel geometry Satellite inundation actual water spread and event reconstruction, cloud, revisit and maps persistence access and crop-loss water-under-vegetation assessment limitations Embankment asset location, condition, risk-based inspection condition data matter register erosion, seepage and and maintenance only if linked to timely maintenance need prioritisation works District flood hazard spatial exposure by evacuation, siting and historical maps require atlas settlement and land-use planning updating as rivers and HISTORY OF MITHILA, VAJJI & ANGA — VOLUME II Evidence source What it establishes Decision / historical Main limitation use infrastructure settlement change Community warning who received what test last-mile reach and informal logs message, when and trust communication may through which channel be incomplete or undocumented Shelter / evacuation occupancy, route evaluate preparedness attendance does not records failure, rescue and and operational capture households vulnerable-person capacity that stayed behind support Agriculture damage & crop area, stage, link hydrology to compensation surveys crop-stage data duration of livelihood loss may omit tenants or submergence and mixed livelihood loss recovery Health / WASH water contamination, measure secondary under-reporting is surveillance disease, facility public-health impacts likely where access is interruption and access most disrupted Household recovery debt, asset replacement, measure resilience attrition and recall bias surveys migration, return and beyond immediate increase with time service restoration survival Cross-border data timing and reliability of assess institutional formal access may exchange records upstream-downstream interoperability differ from actual information emergency use Post-monsoon review lessons, repairs, revised tests whether plans without budget & budgets thresholds and funded institutions learn execution do not create actions between events resilience 13411341