Flood, Drought and Water Security Risk Management

Introduction

Floods and droughts are hydrological extremes whose impacts arise not only from physical processes but also from the location, condition and adaptive capacity of exposed systems. A flood becomes disastrous when people, infrastructure, ecosystems or economic activities occupy hazardous areas and cannot withstand inundation. Similarly, meteorological drought becomes a water-security crisis when storage, allocation rules, institutions and demand management cannot absorb prolonged deficits.

Risk management therefore extends beyond structural engineering. It integrates hydrology, hydraulics, statistics, spatial analysis, asset management, social vulnerability, forecasting, emergency planning and governance. The central relationship is commonly expressed as:

Risk = f(hazard, exposure, vulnerability, consequences).

This is a conceptual relationship rather than a universally valid multiplication formula. Quantitative risk should normally be estimated by integrating conditional consequences over the probability distribution of hazardous events.

Learning Objectives

After studying this chapter, readers should be able to:

1. distinguish hazard, exposure, vulnerability, consequence and risk;

2. undertake quantitative flood and drought risk assessments;

3. apply hydrological routing, including the Muskingum method;

4. interpret drought indices and water-supply performance metrics;

5. evaluate deterministic, probabilistic and ensemble forecasts;

6. design emergency plans and impact-based early-warning systems;

7. compare structural, non-structural and nature-based risk-reduction measures; and

8. address uncertainty, equity, governance and climate non-stationarity.

1. Risk Concepts and Assessment Architecture

1.1 Hazard

A hazard is a potentially damaging physical event or condition. Flood hazards may be described by discharge, water depth, velocity, duration, rate of rise, debris load or contamination. Drought hazards may be described by precipitation deficit, soil-moisture deficit, low streamflow, groundwater decline or reservoir depletion.

A hazard assessment should specify:

• spatial extent;
• magnitude and intensity;
• duration and timing;
• probability or frequency;
• onset speed; and
• interactions with other hazards.

A return period of T years corresponds to an annual exceedance probability p = 1/T under an assumed stationary annual-maximum model. It does not imply that events occur regularly every T years. The probability of at least one exceedance in n independent years is:

P = 1 − (1 − 1/T)^n.

For a 100-year event over 30 years, P = 1 − 0.99³⁰ ≈ 0.260, or 26.0%.

1.2 Exposure

Exposure represents people, assets, services, economic activities or ecosystems located where a hazard may occur. It is quantified by intersecting georeferenced hazard intensity with receptor inventories.

For hazard scenario s, exposed asset value may be estimated as:

E_s = Σᵢ Vᵢ Aᵢs,

where Vᵢ is the replacement value or relevant quantity of receptor i and Aᵢs is an indicator or fractional overlap showing whether it is exposed under scenario s. Fractional overlap is useful for agricultural land, large facilities and population grids.

Exposure databases may contain:

• building footprints, occupancy and floor elevations;
• population by age, disability, income or mobility;
• roads, bridges, power systems and water facilities;
• crops, livestock and industrial production;
• hospitals, schools and emergency facilities; and
• wetlands, habitats and cultural assets.

Asset values should be defined consistently—for example, replacement cost rather than market value—and adjusted to a common price year. Day–night population variation, seasonal tourism and mobile populations require time-dependent exposure models.

1.3 Vulnerability

Vulnerability is the propensity of an exposed receptor to suffer harm. It has physical, social, economic, institutional and environmental dimensions. Quantitative physical vulnerability is often represented by a mean damage ratio:

MDR(h) = expected repair or replacement loss / exposed value,

where h is hazard intensity, such as flood depth. MDR lies between zero and one when only direct asset loss is considered.

Damage-state fragility may be represented by a lognormal model:

P(DS ≥ d | h) = Φ[(ln h − ln θ_d)/β_d],

where θ_d is the median intensity causing exceedance of damage state d, β_d is logarithmic dispersion and Φ is the standard normal cumulative distribution. This model applies only when positive intensity and a lognormal fragility assumption are defensible.

Empirical vulnerability curves can be fitted using generalized linear models, beta regression for fractional damage, survival models or Bayesian hierarchical methods. Relevant explanatory variables include construction type, building age, warning time, duration, contamination and preparedness. Calibration data should distinguish structural damage, contents loss and business interruption.

Social vulnerability indices combine standardized variables such as poverty, age, disability and access to transport. If zᵢk is the standardized value of variable k for area i, an index may be written:

SVIᵢ = Σₖ wₖzᵢk.

Weights wₖ may be assigned by expert judgement, analytic hierarchy process, principal-component analysis or outcome-based statistical calibration. Such indices are sensitive to variable selection, scale, normalization and weighting. They should not be treated as direct probabilities of death or loss without empirical validation.

1.4 Consequence and risk

For mutually exclusive event states s, expected annual loss is:

EAL = Σ_s p_s L_s,

where p_s is the annual probability of state s and L_s is its conditional loss. When an exceedance-probability damage curve is available, expected annual damage is the area under that curve and can be evaluated numerically. Care is required not to sum exceedance probabilities as if they were mutually exclusive probabilities.

Losses include:

• direct tangible damage;
• indirect production and network disruption;
• intangible health, mortality and cultural impacts;
• environmental damage; and
• long-term distributional effects.

1.5 Quantitative exposure and vulnerability example

A flood map identifies 240 exposed houses, each with a replacement value of $120,000. Depth–damage analysis gives:

• 100 houses at 0.5 m depth, MDR = 0.10;
• 90 houses at 1.2 m depth, MDR = 0.35;
• 50 houses at 2.0 m depth, MDR = 0.65.

The exposed value is 240 × $120,000 = $28.8 million. Expected direct damage is:

L = 100(120,000)(0.10) + 90(120,000)(0.35) + 50(120,000)(0.65)

= $1.20 million + $3.78 million + $3.90 million

= $8.88 million.

The portfolio-level damage ratio is 8.88/28.8 = 0.3083, or 30.8%. This estimate remains conditional on the selected flood scenario and excludes casualties, infrastructure interdependence and indirect economic losses.

1.6 Uncertainty analysis

Important uncertainties include hydrological sampling error, terrain error, climate projections, hydraulic roughness, asset location, vulnerability functions and human response. Monte Carlo simulation can sample uncertain inputs and generate a loss distribution rather than a single estimate. Sensitivity analysis identifies influential parameters, while scenario analysis explores deep uncertainties that cannot be assigned reliable probabilities.

2. Flood Hazard Analysis and Routing

2.1 Flood frequency and design events

Annual maxima may be modelled using distributions such as the generalized extreme-value distribution. Peaks over a threshold may be fitted with a generalized Pareto distribution, provided threshold selection and event independence are addressed. Model choice should be supported by diagnostics, regional information and uncertainty intervals.

Stationarity assumes that distributional parameters remain constant. Land-use change, river regulation and climate change may violate this assumption. Non-stationary models can relate parameters to time or climate covariates, but their additional complexity and extrapolation uncertainty must be justified.

Design floods may be estimated through frequency analysis, rainfall–runoff modelling, regional regression or, for high-consequence structures, probable maximum precipitation and probable maximum flood procedures. The method must match catchment data, decision consequences and regulatory requirements.

2.2 Hydrologic and hydraulic routing

Hydrologic routing relates inflow, storage and outflow without explicitly solving spatial momentum equations. Hydraulic routing solves forms of the Saint-Venant equations and can represent backwater, floodplain storage, controls and flow reversals.

The continuity equation for a routing reach is:

dS/dt = I − O,

where S is reach storage, I is inflow and O is outflow. Flow units must be consistent, and the time unit used for storage parameters must match the routing interval.

2.3 Muskingum method

The linear Muskingum storage model is:

S = K[XI + (1 − X)O],

where K is a travel-time-like storage parameter and X is a dimensionless weighting factor. For conventional linear channel routing, 0 ≤ X ≤ 0.5.

Applying the trapezoidal rule to continuity between time steps j and j + 1 gives:

Sⱼ₊₁ − Sⱼ = (Δt/2)(Iⱼ + Iⱼ₊₁ − Oⱼ − Oⱼ₊₁).

Substitution of the storage relation and rearrangement yield:

Oⱼ₊₁ = C₀Iⱼ₊₁ + C₁Iⱼ + C₂Oⱼ,

where

C₀ = (Δt − 2KX)/[2K(1 − X) + Δt],

C₁ = (Δt + 2KX)/[2K(1 − X) + Δt],

C₂ = [2K(1 − X) − Δt]/[2K(1 − X) + Δt].

The coefficients sum to one. For all three coefficients to be non-negative:

2KX ≤ Δt ≤ 2K(1 − X).

Thus, if Δt < 2KX, reducing the time step further does not correct a negative C₀; it makes the violation worse. A routing interval at least equal to 2KX is required, subject also to the upper bound. Parameter estimation, hydrograph resolution and numerical suitability must be considered together rather than adjusting the interval mechanically.

2.4 Corrected Muskingum calculation

Let K = 10 h, X = 0.20 and Δt = 4 h. The admissible interval is:

4 h ≤ Δt ≤ 16 h.

The coefficients are:

C₀ = (4 − 4)/(16 + 4) = 0,

C₁ = (4 + 4)/20 = 0.40,

C₂ = (16 − 4)/20 = 0.60.

Suppose I₁ = 105 m³/s, I₂ = 130 m³/s and O₁ = 80 m³/s. Then:

O₂ = 0(130) + 0.40(105) + 0.60(80)

= 42 + 48

= 90.0 m³/s.

The current inflow I₂ has zero weight because Δt = 2KX in this particular calculation. This is a boundary case, not a general property of Muskingum routing.

2.5 Hydraulic inundation modelling

One-dimensional models are suitable where flow is predominantly along a defined channel. Two-dimensional depth-averaged models are preferred for complex floodplains, urban flow paths and levee breaches. Model development includes terrain conditioning, roughness assignment, boundary conditions, representation of structures and calibration against water levels, discharges or observed flood extents.

Hazard classifications may combine depth and velocity because shallow fast flow can be more dangerous than deeper stagnant water. Duration and rate of rise are also critical for crop damage, evacuation and structural stability.

3. Drought Assessment

3.1 Types of drought

Drought propagates through connected systems:

• Meteorological drought: precipitation below a climatic reference.
• Agricultural drought: insufficient soil moisture for vegetation or crops.
• Hydrological drought: reduced streamflow, groundwater or storage.
• Socioeconomic drought: water shortage relative to human and environmental demand.
• Ecological drought: water deficit causing ecosystem stress or transformation.

These forms do not begin or end simultaneously. Catchment storage may delay hydrological drought, while high demand can create socioeconomic drought without an exceptionally rare rainfall deficit.

3.2 Standardized Precipitation Index

The Standardized Precipitation Index (SPI) compares accumulated precipitation with a fitted probability distribution for the same location, calendar period and accumulation scale. The fitted cumulative probability is transformed to the standard normal distribution:

SPI = Φ⁻¹(H),

where H is the precipitation cumulative probability, including an adjustment for zero precipitation where relevant.

If H = 0.1587, SPI ≈ −1.00. Common descriptive thresholds classify SPI between −1.0 and −1.49 as moderately dry, although operational categories may differ.

Short accumulation periods reflect rapidly responding moisture conditions; 6–12-month periods are more relevant to reservoirs and streamflow. SPI measures precipitation anomaly only and does not directly represent temperature-driven evaporative demand.

3.3 Other drought indices

The Standardized Precipitation Evapotranspiration Index applies a similar standardization to climatic water balance, commonly precipitation minus potential evapotranspiration. The Standardized Streamflow Index and standardized groundwater indices address hydrological responses. The Palmer Drought Severity Index uses a soil-water accounting framework but has assumptions that may not transfer well across climates.

No single index is universally sufficient. Operational drought monitoring should combine rainfall, soil moisture, vegetation condition, streamflow, groundwater, reservoir storage and demand indicators.

3.4 Severity, duration and frequency

A drought event can be defined when an index remains below a selected threshold. Its duration is the number of consecutive deficit periods, while severity is the accumulated deficit below the threshold. Because duration and severity are correlated, copula models or multivariate frequency analysis may be used to estimate joint drought risk. Results are sensitive to threshold choice and serial dependence.

4. Water Security and System Performance

Water security is the capacity to provide acceptable quantities and qualities of water for people, economies and ecosystems while managing water-related risks. It includes availability, access, affordability, quality, ecological integrity and institutional capacity.

For a simulated sequence of water-system states, three useful performance metrics are:

• Reliability: fraction of periods in which demand or another target is met.
• Resilience: conditional probability of returning to satisfactory operation after a failure period.
• Vulnerability: magnitude of deficit when failure occurs.

Consider six periods with deficits [0, 10, 20, 0, 5, 0]. Three periods meet the target, so reliability is 3/6 = 0.50. Of the three failure periods with a following period, two are followed by success; resilience is therefore 2/3. Mean conditional deficit is (10 + 20 + 5)/3 = 11.67 units. Maximum deficit, cumulative episode deficit and population affected may also be reported because an average alone can conceal severe episodes.

Water-security assessment should examine multiple sectors and distributional effects. High aggregate reliability can coexist with chronic insecurity among informal settlements, downstream users or ecosystems.

5. Forecasting and Early Warning

5.1 Forecasting systems

Flood forecasts combine observed rainfall and river levels, radar or satellite precipitation, numerical weather predictions, hydrological models and hydraulic models. Drought outlooks use climate forecasts, land-surface conditions, storage trajectories and demand scenarios.

Deterministic forecasts provide one future trajectory. Ensemble forecasts provide multiple plausible trajectories and support probability statements, such as a 35% probability that a warning threshold will be exceeded within 48 hours.

Forecast evaluation should include:

• bias and root-mean-square error for continuous variables;
• probability of detection and false-alarm ratio for events;
• Brier score for binary probabilistic forecasts;
• reliability diagrams and receiver-operating-characteristic analysis; and
• lead-time-dependent performance.

For observed outcome o equal to zero or one and forecast probability p, the Brier score is (p − o)², averaged across forecasts. Lower scores indicate better probability accuracy, but comparison with climatology or another reference forecast is necessary to establish skill.

5.2 End-to-end early warning

An effective warning system contains four linked elements:

1. risk knowledge;

2. monitoring and forecasting;

3. authoritative communication; and

4. preparedness and response capability.

Warnings should be impact based rather than limited to physical variables. “River level may reach 6.2 m” is less actionable than “ground-floor flooding is likely in specified neighbourhoods; evacuate before the eastern road closes.” Messages should identify the hazard, location, timing, uncertainty, expected impacts and required action.

Warnings must reach people through redundant channels, including sirens, mobile alerts, radio, community networks and accessible formats. Performance should be tested through drills and post-event reviews.

6. Emergency Planning and Risk Reduction

6.1 Emergency planning

A flood or drought emergency plan should define:

• legal authority and incident-command structure;
• thresholds and escalation levels;
• evacuation, shelter or water-rationing procedures;
• protection of hospitals and critical infrastructure;
• logistics, backup communications and mutual aid;
• arrangements for vulnerable populations;
• continuity of essential services; and
• recovery, damage assessment and learning procedures.

Drought plans should use staged triggers linked to progressively stronger actions. Early stages may emphasize communication and voluntary conservation; later stages may activate abstraction restrictions, emergency transfers or priority allocation. Trigger indicators should be transparent and robust to forecast uncertainty.

6.2 Risk-reduction measures

Flood measures include reservoirs, levees, diversion channels, drainage upgrades, floodproofing, retention basins, restored wetlands, floodplain zoning, building controls, warning systems and insurance. Drought measures include demand management, leakage reduction, conjunctive use, aquifer recharge, reuse, transfers, drought-tolerant crops, storage reoperation and diversified supply portfolios.

Structural measures may reduce frequent losses but can create residual-risk complacency or encourage development behind defences. Nature-based measures can provide habitat, water-quality and recreation benefits, but their effectiveness depends on location, scale, maintenance and event magnitude.

Options should be assessed using expected-loss reduction, lifecycle cost, robustness, environmental effects, equity and flexibility. Adaptive pathways are useful where future climate and demand are deeply uncertain: low-regret actions are implemented first, while predefined signals trigger later investments.

7. Applied Basin Case

Illustrative case: the Nira Basin

Background. The fictional Nira Basin contains an upstream reservoir, rapidly urbanizing floodplain communities and an irrigation district. Recent years have included damaging flash floods and two prolonged dry seasons.

Problem. Flood maps were outdated, warning lead time was under three hours, and reservoir rules prioritized average water supply without explicit drought or environmental-flow triggers. Low-income households occupied the most hazardous land.

Analysis. A two-dimensional hydraulic model produced depth–velocity maps for several probabilities. Building-level exposure was combined with occupancy-specific vulnerability curves. Three mutually exclusive annual flood-event classes were estimated:

• moderate: probability 0.080, loss $20 million;
• major: probability 0.015, loss $90 million;
• extreme: probability 0.004, loss $250 million.

Baseline expected annual loss was:

EAL = 0.080(20) + 0.015(90) + 0.004(250) = $3.95 million per year.

Drought analysis combined 6- and 12-month SPI, reservoir storage, seasonal streamflow ensembles and projected demand. Reliability analysis revealed acceptable basin-wide performance but frequent local supply failures in two settlements.

Approach. The proposed portfolio included upstream retention, wetland restoration, targeted drainage works, property-level floodproofing, revised zoning, impact-based warnings and assisted evacuation. Expected losses for the three flood classes were reduced by 35%, 45% and 25%, respectively. Residual expected annual loss became:

0.080(13) + 0.015(49.5) + 0.004(187.5) = $2.5325 million per year.

Expected annual benefit was therefore $1.4175 million. With an annualized cost of $1.10 million, the expected-loss benefit–cost ratio was approximately 1.29, excluding ecological and social co-benefits.

Reservoir operations were revised using staged drought triggers. Leakage control and irrigation scheduling were activated before emergency restrictions, while minimum environmental flows and basic domestic needs were protected.

Outcome. Because the case is illustrative, results are projected rather than observed. The integrated portfolio reduced modelled annual flood loss, improved warning lead time and made drought allocation rules more transparent. Residual extreme-event risk remained and required evacuation planning and periodic review.

Learning points. Basin risk cannot be managed through a single structure or index. Quantified exposure and vulnerability alter investment priorities; average system performance can conceal unequal outcomes; and adaptive operating rules are essential under non-stationary conditions.

8. Technology and Contemporary Developments

Remote sensing supports rainfall estimation, soil-moisture monitoring, flood-extent mapping and vegetation-stress detection. Internet-connected gauges improve real-time situational awareness. Digital twins can link observations, models and operational decisions.

Machine learning is increasingly used for rainfall–runoff prediction, inundation emulation, anomaly detection and damage estimation. It can be valuable when training data are representative and physical constraints are respected. Limitations include extrapolation failure, biased data, poor interpretability and false confidence. AI outputs should therefore be benchmarked, uncertainty-calibrated and subject to human oversight.

Climate services increasingly use storyline approaches, ensemble projections and stress testing rather than relying on one predicted future. Compound events—including coastal surge with river flooding, or drought followed by wildfire and sediment-laden floods—are receiving greater attention.

9. Governance, Ethics and Professional Practice

Risk decisions distribute protection, costs and residual harm. Engineers and authorities should disclose modelling assumptions, uncertainty and the limitations of design standards. Risk maps must not stigmatize communities or become a mechanism for withdrawing services without supporting relocation or adaptation.

Data governance is important where household vulnerability, mobile-phone movement or health information is used. Privacy, informed use, cybersecurity and appropriate data retention are professional obligations.

Transboundary drought management requires agreed data-sharing, allocation priorities and dispute-resolution mechanisms. Flood defences should also be assessed for transferred risk to downstream or opposite-bank communities. Regulatory compliance provides a minimum requirement; professional responsibility additionally requires attention to foreseeable residual risk, environmental justice and long-term system sustainability.

Chapter Summary

Flood, drought and water-security risk result from interactions among physical hazards, exposed receptors, vulnerability and institutional capacity. Advanced assessment combines probabilistic hazard analysis with geospatial exposure inventories, fragility or damage functions and consequence modelling. The Muskingum method provides a hydrological routing procedure when its storage assumptions and coefficient conditions are satisfied. Drought monitoring requires multiple indices and recognition of propagation from climate anomalies to social and ecological impacts. Reliability, resilience and vulnerability describe complementary aspects of water-system performance.

Forecasts create value only when linked to trusted communication and feasible action. Effective management combines structural, non-structural and nature-based measures, supported by emergency plans and adaptive pathways. Uncertainty, climate non-stationarity, equity, data ethics and residual risk must remain explicit throughout professional decision-making.

Key Terms

Hazard: A potentially damaging physical event or condition.

Exposure: People, assets, services or ecosystems located within a hazard zone.

Vulnerability: The propensity of an exposed receptor to suffer harm.

Fragility curve: A function giving the probability of exceeding a damage state at a specified hazard intensity.

Mean damage ratio: Expected direct loss expressed as a fraction of exposed asset value.

Expected annual loss: Probability-weighted mean loss per year.

Annual exceedance probability: Probability that a specified magnitude will be exceeded in one year.

Return period: Reciprocal of annual exceedance probability under a stationary model.

Muskingum routing: A hydrological channel-routing method based on weighted inflow and outflow storage.

Inundation model: A hydraulic model used to estimate flood depth, velocity or extent.

SPI: A standardized index describing precipitation anomaly over a selected accumulation period.

Drought severity: Accumulated deficit during a drought event.

Reliability: Fraction of periods in which a system target is satisfied.

Resilience: Ability or probability of recovery after system failure.

Water-system vulnerability: Magnitude of adverse consequences when system failure occurs.

Ensemble forecast: A set of simulations representing plausible future conditions or uncertainties.

Impact-based warning: A warning communicating expected consequences and actions, not only physical measurements.

Adaptive pathway: A sequence of actions with future decisions triggered by observed change.

Review Questions

Short Answer Questions

1. Distinguish hazard, exposure and vulnerability.

2. Why does a 100-year flood not occur regularly once every 100 years?

3. State the assumptions represented by the linear Muskingum storage equation.

4. What information does SPI provide, and what does it omit?

5. Define reliability, resilience and vulnerability in water-supply analysis.

Long Answer Questions

1. Explain how flood hazard, exposure and vulnerability data are integrated to estimate expected annual loss.

2. Derive the Muskingum routing equation and discuss the conditions for non-negative coefficients.

3. Compare SPI, SPEI and hydrological drought indices for operational drought management.

4. Develop the main components of an end-to-end, impact-based early-warning system.

5. Critically evaluate structural, non-structural and nature-based approaches to flood-risk reduction.

Analytical / Application Questions

1. Construct an exposure and vulnerability model for a mixed residential–industrial floodplain, including uncertainty and indirect loss.

2. Given a reservoir inflow ensemble, sectoral demands and environmental-flow requirements, formulate drought triggers and performance metrics.

3. Evaluate a proposed levee using expected-loss reduction, residual risk, transferred risk, equity and climate non-stationarity.

Multiple Choice Questions with Answers

Question 1

Which expression gives the probability of at least one T-year event exceedance in n independent years under stationarity?

A. 1 − (1 − 1/T)^n

B. n/T in all circumstances

C. (1/T)^n

D. 1 − (1/T)^n

Correct Answer: A. 1 − (1 − 1/T)^n

Question 2

For K = 10 h, X = 0.20 and Δt = 4 h, C₀ = 0, C₁ = 0.40 and C₂ = 0.60. If I₁ = 105 m³/s, I₂ = 130 m³/s and O₁ = 80 m³/s, what is O₂?

A. 80.0 m³/s

B. 90.0 m³/s

C. 104.5 m³/s

D. 130.0 m³/s

Correct Answer: B. 90.0 m³/s

Question 3

For K = 10 h and X = 0.20, what is the minimum routing interval that permits a non-negative Muskingum C₀ coefficient?

A. 2 h

B. 3 h

C. 4 h

D. 16 h

Correct Answer: C. 4 h

Question 4

An asset portfolio has an exposed value of $28.8 million and expected direct damage of $8.88 million. What is its scenario-specific portfolio damage ratio?

A. 19.2%

B. 25.0%

C. 30.8%

D. 38.5%

Correct Answer: C. 30.8%

Question 5

Three mutually exclusive annual flood classes have probability–loss pairs (0.080, $20 million), (0.015, $90 million) and (0.004, $250 million). What is expected annual loss?

A. $2.35 million

B. $3.95 million

C. $7.20 million

D. $14.40 million

Correct Answer: B. $3.95 million

Question 6

If the fitted cumulative probability used in SPI calculation is 0.1587, the SPI is approximately:

A. −1.00

B. −0.16

C. 0.00

D. 1.00

Correct Answer: A. −1.00

Question 7

For deficits [0, 10, 20, 0, 5, 0], what is the period-based reliability when zero indicates that the target is met?

A. 0.33

B. 0.50

C. 0.67

D. 0.83

Correct Answer: B. 0.50

Question 8

Which quantity is most directly represented by a flood fragility curve?

A. Probability of exceeding a damage state conditional on hazard intensity

B. Market value of every exposed building

C. Annual probability of rainfall occurrence

D. Time required to route a hydrograph

Correct Answer: A. Probability of exceeding a damage state conditional on hazard intensity

Question 9

Which statement about the Brier score is correct?

A. A higher score always indicates a better forecast

B. It is used only for deterministic discharge forecasts

C. A lower score indicates smaller probabilistic forecast error

D. It measures exposed asset value

Correct Answer: C. A lower score indicates smaller probabilistic forecast error

Question 10

Which warning is most consistent with impact-based early-warning practice?

A. River stage is forecast to be 6.2 m

B. Heavy rainfall is possible somewhere in the basin

C. Model simulation 17 exceeds its median value

D. Ground-floor flooding is likely in the eastern district; evacuate before the access road closes

Correct Answer: D. Ground-floor flooding is likely in the eastern district; evacuate before the access road closes