Lec 2 - Foundations for hazard and risk assessment

Question 1

Components of a risk assessment

Answer 1

1) Risk identification - Recognise and describe the risks (single/multi hazards) and the scope of the assessment
2) Risk analysis - understand the nature and sources of the risk and estimate the level of risk (HEV)
3) Risk evaluation - compare risk with risk criteria (acceptable, tolerable etc)

This lecture will focus on environmental/natural hazard assessment

Question 2

1) Risk identification and scope - features

Answer 2

• Identify the hazard scenario (or set of) of interest e.g. 1 in 100 yr flood
- note that hazards can be coupled, cause cascades or chains
• Select modelling approach depending on:
- hazard type
- Purpose of assessment
- Scale of analysis
- Data and model availability
• Main components of hazard analysis: frequency/magnitude relationship; initiation process; and travel or transmission of hazard

Risk assessment is often synonymous with risk mapping, given the importance of spatial location

Question 3

Now 2) of risk assessment - the Risk Analysis; i.e. the methods used to analyse the hazards and data needed, scale

Answer 3

Type of scale - possible objectives - model approach:

1) Global - Prioritisation of countries/regions; early warnings - RMA/IBA
2) Small: provincial to national scale - Prioritisation of regions; Analysis of triggering events; Implementation of national programs; Strategic environmental assessment; insurance - EVA/RMA/IBA
3) Medium: municipality to provincial level - Analysing the effects of changes; Analysis of triggering events; Regional development plans - RMA/IBA
4) Local: community to municipality - Land use zoning; Analysing the effects of changes; Environmental Impact Assessments; Design of risk reduction measures - QRA/EVA/RMA/IBA
5) Site-specific - Design of risk reduction measures; Early warning systems; detailed land use zoning - QRA/EVA/RMA

• What scale you should work at depends on the type of hazard! e.g. floods very important for comunity scales, less applciable at global level

Question 4

Carrying on for 2) Types of hazard model

Answer 4

1) Statistical hazard model AKA the black box model
• Method: Derive a statistical relationship between observed hazard drivers and resulting hazard event.
• Data: Using observed events and its probabiltiies along with system propertories (preparatory factors)
• Assumption: this model only good for similar settings; if climate changes occurs model breaks down etc

For example for floods:
Trigger = Rainfall
Preparatory factors = Steam gradients, hydrological system of shed

2) Physically-based AKA grey/white box models
• Method: The hazard processes are explicitly represented by physics-based equations solved using numerical modelling
• Data: Measured physical parameters of the system
• Assumptions: the model and input data describe the physical system completely!! - note the system can never be fully defined so many grey box models use a mix of measured and empirical inputs

3) Qual. or Quan. based - based on judgement/logic or analytical approach.
4) Heuristic models - using subjective expert knowledge to descrbie teh hazard system - can make use of semi-quan indices
5) Stochastic and probabilistic models - input values sampleted from their probabiltiity distributions - helps bridge gap between black and grey box models
6) Spatially distributed models e.g. GIS

Question 5

carrying on 2) Sources of model uncertainty

Answer 5

1) Aleatory = inherent randomness of the system e.g. natural variation in soil.
2) Epistemic (system dynamics) = arising from a lack of knoweldge about how to represent the catchment system in terms of both model structure and parameters
3) Epistemic (forcing and response data) = Uncertainty arising from lack of knowledge about the forcing data or the response data with which model outputs can be evaluated. Can be becasue of interpolation issues, limited gauging networks, extrapolation of rating curve etc
4) Epistemic (disinformation) = uncertainty in system representation or forcing data that are KNOWN to be wrong. . Includes false positives/negatives
5) Semantic/lingusitic = uncertainty about what statements or quantities in the relevant domain actually mean e.g. storm runoff, baseflow, hydraulic conductivity etc. Words can have different meaning
6) Ontological = uncertainty due to different belief systems, assumptions leading to different uncertainty estimates.

Question 6

Carrying on for 2) Typical risk modelling frameworks and outputs for end-users i.e. how the resultant outputs might be used by engineers

Answer 6

1) Quantitative Risk Analysis (QRA)
• Components of the risk eqn (HEV) are quantified for at least 3 hazard scenarios of diff frequencies.
• Vul. is on scale of 0-1
• Expected losses associated with each hazard scenario plotted to create a risk curve - area under curve gives total losses.
• log frequency on y-axis, loss on x axis

2) Event tree analysis
• usefull where one hazard triggers another (cascading multi-hazards)
• all possible chains of events are considered
• The resulting probabilities are calculated for each possible outcome (the end of each branch) e.g. P1P2P3 or P1* (1-P2)

3) Risk matrices (or Consequence-Frequency Matrices CFM)
• Not all risks can be numerically quantified due to uncertainties or lack of data
• Expert judgement used to estimate hazard frequency, mangitude or intensity and the consequences (E and V)

4) Indicator- based approach
• Used when not all components of risk eqn can be quantified due to lack of data.
• The area of interest is sub-divided into units that can be assessed against indices (e.g. building type and poverty indicators)
• Indices are standardised from 0 to 1 and weighted according to expert judgement before being combined to give hazard and consequence and final risk index

Question 7

Part 3) of the risk assessment - the Risk Evaluation - how safe is safe enough? comparing to criteria

Answer 7

• Many risk management decisions made on financial grounds, therefore requiring an economic value to be attributed to human life.

• Human capital method = based on lost future earnings from death/indury
• Willining to pay approach = how much people prepared to pay to avoid death, measures risk aversion like this
• Aim of risk management = bring ALARP. CBA used to prioritise risk reduction resoruces
• Set at national levels