Presentation of risk

According to the definitions previously given, risk or specific risk is defined as the average rate of loss or 'attrition rate'. While this is useful for estimating losses over a long period of time, it can give a misleading idea of the nature of the risk from natural hazards. Most of the losses from these events actually occur through infrequent large single events, rather than in the form of a slow continuous process of destruction. A variety of different methods have been developed for the presentation of risk to help overcome this difficulty.

One method is the use of f:N curves, such as those shown on page 20 (figure 4) which present the frequency of events with different numbers of casualties (or magnitude of losses expressed in some other way). Presenting risk in this way is thought to be closer to the way people actually perceive it. However, such relationships always show aggregated losses for a large region and period of time. They do not help to identify the geographical distribution of damage, for which risk mapping is needed.

Risk maps attempt to show the spatial or geographical distribution of expected losses from one or more natural hazards. Because of the way natural hazards occur, the presentation of annual risk, as defined above, is not necessarily the most useful, and several different ways of presenting losses are commonly used including:

a. Scenario Mapping: The presentation of the impact of a single hazard occurrence. Scenario mapping is often used to estimate the resources likely to be needed to handle an emergency. The number of people killed and injured, and the losses arising in other elements is estimated. From these can be estimated the resources needed for medical attention, to reduce disruption, accommodate homeless, and minimize the recovery period. See Example 1 (A scenario event).

b. Potential Loss Studies: Mapping the effect of expected hazard occurrence probability across a region or country shows the location of communities likely to suffer heavy losses. The effect of the hazard of each area is calculated for each of the communities within those areas to identify the 'Communities Most At Risk'. This shows, for example, which towns or villages are likely to suffer highest losses, which should be priorities for loss-reduction programs, and which are likely to need most aid or rescue assistance in the event of a major disaster. See Example 2 (Potential loss study).

c. Annualized Risk Mapping: Calculation of the probable levels of losses occurring from all levels of hazards over a period of time. The probability of each level of hazard occurring within that unit time period is combined with the consequences of that level of hazard to generate the expected loss within that time. Summing up the losses over all levels of hazard gives the total losses expected with time.

The map indicates expected losses over both time and space. With sufficient detail in the calculation, the likely effect of mitigation policies on reducing earthquake losses can be estimated, and costed. The relative effects of different policies to reduce losses can be compared or the change in risk over time can be examined. See Example 3 (Annualized risk).

Q. What is the main advantage of risk mapping over the plotting of risk curves (f:N curves) as previously discussed?

A._______________________________________________________________ _________________________________________________________________ _________________________________________________________________
_________________________________________________________________ _________________________________________________________________ _________________________________________________________________

ANSWER Risk mapping presents risk in a geographical way that shows the risk trends over an area and allows comparison of risk levels in different geographic areas. The f:N curves, on the other hand, show aggregated losses for a large region over a given time period.

Example 1: Scenario mapping

The map in figure 5 shows the expected consequences of an earthquake of a particular magnitude (Surface-wave magnitude, Ms = 7.2) occurring with its epicenter at a particular location in Bursa Province, Western Turkey. The magnitude and location are within the range of possible occurrences, i.e. they are consistent with seismological knowledge of the faulting and earthquake history of the region. The earthquake is thus a possible event, and not the largest or the most damaging which could occur. The probability of its occurrence has not been calculated. The damage distribution resulting from this event has been estimated from:

a) statistics on damage distributions caused by other earthquakes in this region for a range of building types, and

b) a knowledge of the actual composition of the present building stock in this area

Each settlement in the affected area is represented by a circle, the area of which represents the population of the settlement. The proportion of the circle which has been shaded indicates the expected extent of damage to the settlement (more precisely the proportion of the residential buildings which can be expected to suffer heavy or irreparable damage).

Table 2 accompanies the map and gives totals on the amount of damage to houses and numbers of people killed, injured or made homeless. It also gives a breakdown between the villages, towns, and the provincial capital city of Bursa.

Table 2
Summary of expected damage and losses caused by hypothetical 7.2 magnitude earthquake in Bursa Province, Turkey¹²

	VILLAGES	TOWNS	BURSA CITY	TOTAL
HOUSES LIGHTLY DAMAGED	34,000	21,000	50,000	105,000
HOUSES HEAVILY DAMAGED	15,000	9,000	30,000	54,000
HOUSES COLLAPSED	4,000	2,000	6,000	12,000
PEOPLE KILLED	2,000	800	1,500	4,300
PEOPLE INJURED	6,000	2,500	4,500	13,000
PEOPLE HOMELESS	73,000	36,000	130,000	239,000

Even though the map shown in figure 5 is an illustration of possible effects and not a prediction, it is able to play an important part in warning local officials, fire departments and the public at large of possible consequences of this particular hazard as an aid to mitigation planning.

Q. What information is best presented in the scenario map? What basic information is not provided in this type of map?

A._______________________________________________________________
_________________________________________________________________ _________________________________________________________________ _________________________________________________________________

ANSWER Probable property damage to the cities and towns shown due to a hypothetical M = 7.2 earthquake is clearly indicated. The probability of the occurrence of such an earthquake is not presented.

Figure 5 - Example 1 - A scenario event

Example 2: Potential loss mapping

The potential loss map presents risk as the levels of losses that would occur if a certain level of hazard were to occur at all the locations simultaneously (see figure 6). In this case the type of loss plotted (Map 4) is urban earthquake casualties in Turkey. Casualties are defined as those people whose houses are liable to be totally destroyed by the largest expected earthquake - a measure used because it has been found in Turkey to correlate closely with the numbers of killed and injured. The potential loss plotted in each location is derived from three other types of geographically varying data, which are shown in Maps 1, 2 and 3 (see figure 6).

Figure

Map 1 shows the earthquake hazard in terms of the maximum intensity of earthquake which might possibly occur there - based largely on reinterpretation of historical records. This map is published by the Earthquake Research Institute of Turkey and is also used to define the level of earthquake which new buildings should be designed to resist.

Map 2 shows the elements at risk - in this case the total size of the urban population. Larger towns and cities (over 25,000 population) are plotted individually, and are identified by circles whose area represents the population - apart from the four largest cities whose population is specified. The population in the smaller towns of 2,000 to 25,000 population is shown in the form of a population density. This information is derived from national census data. Other elements at risk - bridges, schools or roads could be mapped in a similar way.

Map 3 shows one aspect of the vulnerability of those elements at risk. The casualties are caused by the collapse of buildings. The vulnerability of a building depends primarily on the type of construction. A useful approximate classification of the building types in Turkey divides them into just three types. For each of these building types damage statistics from past earthquakes have been used to derive vulnerability functions, showing expected proportions of the buildings of each type which may collapse at different intensities:

Type A: Rubble and adobe walls (1% collapse at intensity VII, 5% at VIII, 50% at IX)
Type B: Brick and timber walls (1% collapse at intensity VIII, 5% at intensity IX)
Type C: Reinforced concrete frame (5% collapse at intensity IX)

To assess total damage the distribution of the urban residential buildings between these three classes is needed. Such information is available from Turkish census data. Map 3 plots this data for the province center and other towns for each province of Turkey. The way in which the proportion of reinforced concrete buildings increases towards the richer, more affluent west is immediately apparent, as is the predominance of weaker rubble and adobe buildings in the south east.

Map 4 shows the analysis of the three preceding maps for each location. This is derived by estimating the numbers of people living in each building type, (from Maps 2 and 3) and then estimating the potential proportion of collapsed buildings of each type if the largest earthquake were to occur there. The total potential casualties are obtained by adding those from all three building types.

Figure 6 - Example 2 - Potential loss study

1 - HAZARD

2 - ELEMENTS AT RISK (population)

3 - VULNERABILITY

4 - CASUALTY RISK (potential loss of life)

The total potential loss plotted in this way helps to suggest priorities for what national planning should be. In this case the large cities in the west have the greater potential loss (because of larger population), though the potential loss in the larger eastern cities is also significant (because of weaker buildings). Few countries have census data giving as precise data on building types as Turkey, and the distribution of building types may have to be estimated in some other way.

Q. The potential loss map in figure 6 combines three types of maps (hazard, population, and vulnerability). What are some of the assumptions that have been made to produce the final map?

A._______________________________________________________________ _________________________________________________________________ _________________________________________________________________ _________________________________________________________________
_________________________________________________________________ _________________________________________________________________ _________________________________________________________________

ANSWER 1. It must be assumed that the present population estimates will still be valid at the future unspecified date of the hazard occurrence. 2. It is assumed that the building materials of the houses are the main factor contributing to vulnerability in this area, (as opposed to siting and configuration of the structures, for example).

Example 3: Annualized risk mapping

The annualized specific risk from any hazard at any location is the average expected total losses from all events over an extended time period divided by the number of years involved. It is expressed as a proportion of the total value (or number) of the total population of that element at risk. The annualized risk can be shown in the form of a contour map (see figure 7). This map plots the annual risk contours for village housing in a particularly high-risk part of Eastern Turkey. Loss is defined as heavy damage or collapse, measured by the proportion of all houses suffering this level of damage. The risk increases towards Karliova in the top right hand corner of the map and then begins to decline. At Karliova, with an annual risk of 2%, about 50% of houses would be expected to be lost within 25 years, whereas at Palu (bottom left) losses would be only half this. Such calculations make the perhaps unrealistic assumption that destroyed houses are replaced by new houses built in the same way.

One feature of all damage distributions is considerable variation between villages, and some indication of this variation can be obtained by plotting, instead of the average expected loss, the loss which can be expected to be exceeded by a given proportion of the locations. The specific risk exceeded by 75% of all villages is shown as a second set of contours.

Either of these plots can be used to measure the reduction in risk resulting from some change in the elements at risk such as strengthening the building stock or changing the settlement pattern. Such plots can therefore be very useful in mitigation planning.

Figure 7 - Example 3 - Annualized risk - % housing loss per year Bingrovince, Turkey

Remember... In reading risk or loss maps of any of these types it is important to realize that they do not offer predictions. Because of the uncertainty of the knowledge available about hazards, their recurrence patterns and their effects, all loss estimates are merely extrapolations into the future of the observed statistical distribution of occurrences of hazards and their effects in the past. Quite large-scale shifts in the pattern of occurrence of both geological and climatological hazards can and do occur, and development planning must consider this possibility.

Q. The annualized risk map presents most clearly: (check the appropriate answer)

A. the number of people exposed to the hazard
the degree of severity of the hazard expected the comparative probable losses between different sites the probability of the specific hazard occurring

ANSWER the number of people exposed to the hazard the degree of severity of the hazard expected the comparative probable losses between different sites the probability of the specific hazard occurring