Journal of Applied Sciences & Environmental Management, Vol. 9, No. 1, 2005, pp. 57-63

Flood Risk Assessment: A Review

*OLOGUNORISA, T E; ABAWUA, M.   J.

Department of Geography, Benue State University, Makurdi, Nigeria.

*E-mail: ologunorisatemi@yahoo.com

Code Number: ja05010

ABSTRACT: The paper reviews some of the techniques of flood risk assessment using case studies from different countries in the world.  These techniques are meteorological, hydrological, hydrometeorological, socio-economic and those based on Geographic Information System (GIS).  The paper concludes that GIS technique appears to be most promising as it is capable of integrating all the other techniques of flood risk assessment.@ JASEM

Flooding is the most common of all environmental hazards and it regularly claims over 20,000 lives per year and adversely affects  around 75 million people world-wide (Smith, 1996). The reason lies in the widespread geographical distribution of river floodplains and low-lying coasts, together with their long standing attractions for human settlement. Death and destruction due to flooding continue to be all too common phenomena throughout the world today, affecting millions of people annually. Arising from this, The International Decade for Natural Disaster Reduction (IDNDR) was launched by the General Assembly of the United Nations in 1987 to run  from 1990 to 2000. It’s aim is to reduce the loss of life, property damage and social and economic disruption caused by natural disasters (Askew, 1999). The resolution proclaiming the Decade makes specific reference to flood, tsunamis, drought and desertification among the principal disasters to be considered. Floods cause about one third of all deaths, one third of all injuries and one third of all damage from natural disasters (Akew, 1999).

The year 1994 marked the mid-point in the Decade and as a consequence, the United Nations conveyed  a World Conference  on Natural Disaster Reduction that was held in Yokohama in May of that year. One of the  10 “principles” underlying the Yokohama strategy is that “risk assessment is a required step for the  adoption of adequate and successful disaster reduction policies and measures”.

The need for a review of some of the techniques of flood risk in order to reduce large-scale losses  to lives and property has been  stressed  in the literature (Askew, 1999; Smith, 1999; Ologunorisa, 2001). The  objective of this paper  therefore is to review and synthesize  some of the techniques  of flood risk assessment into a coherent piece. The  paper undertakes first a theoretical clarifications of the concept of risk before reviewing flood risk assessment techniques.

Theoretical Clarifications

Risk is an integral part of life. Indeed, the Chinese word for risk “weji-ji” combines the characters meaning “opportunity/chance” and “danger” to  imply that uncertainty  always involved some  balance  between profit and loss (Smith, 1996). Since  risk cannot be completely  eliminated, the only option is to manage it. Risk assessment  is the  first step in risk management. Risk  assessment  according to  Kates and Kasprson (1983) comprises of  three distinct  steps.

a) An identification of hazards likely to result in disasters, e.g. what hazards events may recur?

b) An estimation of the risks of such  event, e.g. what is the possibility of such  event?

c)  An evaluation of the social consequences of the  derived risk, e.g. what is the loss created by each event?

However, for sound risk management to occur, there  should be  a fourth (d) step which addresses the  need to take post-audits of all risk assessment exercises. When  risk analysis is undertaken, risk (P) is taken as some product of probability (P) and loss (L).

R =    P x  L ……(1)

Flood  risk  involves both the  statistical  probability of  an event  occurring and the scale  of the potential consequences (Smith, 1996). All development of land within the  floodplain of a watercourse is at some risk of flooding, however, small. The  degree of flood risk is calculated from historical data and expressed in terms  of the expected frequency 10 year, 50 year or 100 year flood.

Flood  risk  is a function and a  product of hazard  and vulnerability  (Ologunorisa, 2001). That  is, Risk = Hazard x  Vulnerability. A real flood risk level requires a certain level of hazard, and for the same location, a certain level of vulnerability. A situation of risk is due to the incompatibility between hazard and vulnerability  levels on the  same land plot. The United Nations  Commission for Human Settlements (UNCHS – HABITAT) (1981) has defined the  three terms in the  following way:

a)       Hazard:  is the  probability  that in a given period  in a given area, an extreme potentially damaging natural phenomena occurs  that induce  air, earth movements, which  affect a given zone. The magnitude  of the phenomenon, the probability of its occurrence  and the extent of its impact can vary and, in some  cases, be determined.

2. Vulnerability – of  any physical, structural or socio-economic element  to a natural hazard is its  probability  of being damaged, destroyed or lost. Vulnerability is not static but  must be considered as a dynamic  process, integrating changes  and developments that alter and affect the probability of loss and  damage of all  exposed elements.

3.    Risk – can  be related directly to the  concept of disaster, given that it includes the total  losses  and damages that can be suffered after a natural hazard: death and injured people, damage to property and  interruption  of activities. Risk implies a future potential condition, a function of the  magnitude of the  natural hazard and of the  vulnerability of all the  exposed  elements in a  determined moment.              

Risk Assessment Techniques

(i)            Meteorological Parameters

Various definitions have been used in defining  floods (that is large  rainfall surpluses) over a region depending on the  purpose in view. This  simple parameter, rainfall, which  can reflect many aspects of flood is used in  partially all definitions of flood. Meteorological  flood can be defined as a situation over a  region where  that rainfall  is  mostly higher  than the climatological  mean value because the natural  vegetation and economic activities  of the region have been adjusted  to the long-term average  rainfall of that region (Parthasarathy, et al 1987). Therefore, the conditions  which  lead to flood occur  when the rainfall amount over a particular  region is more than a certain amount, normal for  that region (Friedman, 1957, WMO, 1975).

Many countries, notably Costa Rica, Israel, Islands of Aruba, Jamaica, Botswana, Ethiopia, Malaysia, Mauritius, Korea, Pakistan, USA and Australia, use in some  form or other the  criterion of a  given percentage departure form the normal. When the seasonal rainfall is in an excess of 26-50 percent of normal over a meteorological subdivision is regarded as a moderate flood, and an excess of more than 50 percent of the  normal as a severe  flood; while  the rainfall between 0-26 percent of normal is regarded as a less severe  flood. This is the  definition that has been adopted by the  Indian Meteorological  Department (1971). See  also Ramon (1975) Government  of India (1976), and which  many countries in Southeast  Asia, and even U.S.A and Australia (Parthasarathy et al, 1987) have  adopted.

Laughlin and Kalma (1990) developed  a methodology for frost risk mapping based on regional weather data and local  terrain  analysis. Minimum air temperatures were  measured  during three winters with a network of stations in open, undulating  terrain. It was  observed that the change in minimum air temperature  with elevation could be  predicted from mean night time wind speed, total nighttime net radiation loss and a hill-top reference minimum  temperature. It  was also found that the deviation of temperatures at individual sites  could be predicted from a local terrain parameter which  reflects the  extent of cold air  accumulations. Finally, the  study describes the  model and illustrates the regional weather and terrain effect with three-dimensional  block diagrams.

Single et al (1990) after considering the total seasonal rainfall of June through September as well as  its time, developed an index  that has been evolved for  identifying a year as hydrological  flood/drought in  different parts of India. After giving a margin of 25% to the  mean index value for normal years, frequencies of flood/drought years have been calculated. In general,  frequencies of both hydrological  flood and drought years are more in  low rainfall  areas as compared to  high rainfall areas. They  compared quite closely with the  frequencies of meteorological excessive and  deficient  rainfall years respectively. But the  categorization  of some individual years is found to differ  between meteorological and hydrological  points of view. The  nature of these differences is however, not uniform at all the  stations.          

Hayden (1988) remarked that an  examination of the  literature  reveals neither  a global classification of flooding nor a regionalization or map of flood climates types on a global basis. His  major objectives  was to generate such a classification and a map of the resulting flood climate  types and to detail the basis for its constructed. A  coherent global climatology of flooding  is not  easily constructed from stream  discharge  frequency and magnitude statistics. The  difficultly arises, in part, because the frequency and magnitude of floods vary between and within drainage networks  because of variability in  basin characteristics. This variability is further  complicated by the  diversity of weather systems that may  give rise to flooding, each  with its own characteristics return interval  spectrum. Although the  variability in  basin characteristics prohibits regionalization on a global scale, the  meteorological causes and potential for  flooding can be classified and  regionalized. In the  study, flood climate regions were delineated on the basis of meteorological causation.

Olaniran (1983) examined flood generating mechanisms  at Ilorin. He  noted that in the  decade 1971-1980, the town experienced rains greater than 25.4 mm/day induce floods when  they occur in a  month about three or more times during the period of moisture surplus at Ilorin. The  study shows further that the  construction of Aso dam has not prevented the occurrence of flood at Ilorin which  is situated downstream of the dam on account of the  network characteristics and channel  slopes of the tributary  stream and  the increasing  rate of urban development downstream  of the dam.

McEwen  (1989) assessed  whether there have been changes in  the  magnitude, frequency, duration and seasonality of extreme  rainfall over the last 100 years within the middle River Tweed basin, Berwickshire, Scotland. Rainfall patterns were compared with  published analysis of other long-term rainfall records to evaluate regional  variations in the characteristics of extreme rainfall.  Information on the magnitude  and frequency of rainfall extremes was  used to substantiate and augment a previously  established  extended flood record of the  White adder  Water catchment  within the middle  River Tweed basin. He reported a coincidence  between increased frequency of  heavy rainfall  and increased frequency of moderate to  extreme floods during the 1870-1880s and a reduced flood incidence  in the 1970s which  was substantiated by an absence  of extreme rainfall  peaks in the middle River Tweed basin, Scotland.

Durotoye (2000) elaborated on the four major sources of water that largely contribute  to the  inundation of the deltaic  plains. These  include:

(a)   The  annual river peak discharge of River Niger especially  the “White Flood” in October and the “Black Flood” between December and March. The “Black Flood” discharge, which originates from the  Niger’s headwaters in Guinea generates  freak episodes  of high  discharge  during exceptionally high rain in wet years. The “White Flood” originates mostly from within  the Nigerian catchment  areas. It has a high suspended  load of  fine sand/silt and clay which are deposited to built the deltaic  alluvial  flood plains and  levees. She observed that high peak discharges cause bank and levees failures as the stream burst  their banks  seasonally to  create disastrous floods.

(b)         The  role  of heavy rainfall peaks;

(c)   Man induced floods through oil exploitations and explorations, and human interference  with the  courses of stream  channels  during  various constructional works;

(d)   Tidal floods especially in the  coastal areas and mangrove swamps. The  tidal flooding by ocean  water createds the characteristics  brackish water environment. Also  ocean surges also generate stormy high  tides resulting in  destructive flood.

Abams (1995) reported in his study  of floods in Kiama (in Bayelsa State), Nigeria located in the lower  deltaic  flood plain on River Nun  that the annual flooding experienced are a consequences of the  combined effect  of the “White Flood” discharges and the  heavy rains  from tropical storms. The  rainfall  is up to 3250 mm/yr, about 80% of  which is received between June and  October. Heavy torrential rains are known to fall continuously  for several days  (up to ten  days a times). The rain water collects over impermeable silty clay plains. Consequently, meandering  rivers not only over flow their  banks but are actively  eroding them and  causing bank  failures. It is also important to note that the  ground level of water  table is virtually several metres above  surface (up to 9m) rains, leaving little  or no room for percolation into the  ground.

Durotoye (1999) highlighted the  increasing tendency of human  occupation  of potential hazard area in Nigeria, and Niger Delta in particular. She observes that the physical conditions and natual processes prevailing  in the Niger Delta environment that  impinge  on the safety  of the people  living there are  numerous. She  identified geo-environmental problems such as  flooding, marine incursion, coastal recession  and subsidence.

ii.             Hydrological  Parameters

Trinic  (1997) did an hydrological analysis of high flows and floods of the  Sava River near Zargred (Croatia) in the period from 1926 to 1992. Particular attention was paid to the causes of flood  wave volumes from direct inflows above reference  discharged  and of constant duration were  analyzed. The  results of hydrological analysis  of the flood wave hydrographic can be used to improve manipulation  with waters using weirs, flood diversion canals and retentions. The article  analysis  flood waves of the Sava River near Zagreb for the  period 1926-1992, highlighting particularly the  disastrous flood of October, 1964. 

Kattelmann (1997) analyzed flooding  from rain-on-snow events is Sierra Nevada. He  observed that the most damaging floods in  rivers of the  Sierra Nevada of California’s have occurred during  warm storms when  rain fell in snow covered  catchments. These  large floods have inundated communities  and farms in California’s prime agricultural region. Forecasting of runoff  from rain-on-snow events have been  difficult for managers of dams and power  plants within the  Sierra Nevada and for  downstream  flood control agencies because of  uncertainties about  runoff  production at high altitudes  and snowmelt  contributions at low altitudes. The  results of the  analysis show that the high  potential for flood  generation from rain-on-snow events is related to their  large contributing area, intensity and duration  rainfall, opportunity for  snowmelt  contributions, and the timing of water released fro the now pack.

Nobilis and Lorenz (1997) analyzed flood  trends in Australia. The study deals with  the analysis  of previous floods, the  assessment of damage, and the  evaluation of possible changes in the flood behaviour due to natural  or artificial  influences. The results of the  analysis  show areas with  predominantly linear trend and areas with  predominant positive significant  (P=0.05) linear trend. The  authors observed that based on  such investigations, realistic  design values  may be calculated taking into  account the end of stationary due to  climate change.

Bogdani and Selenica (1997) analyzed  catastrophic floods and  their “risk” in the  rivers of Albania. The  study described the  main characteristics of floods in Albanian  rivers including information on the highest floods observed during  the last 150 years. In  addition, a brief overview of  flood regions in Albania, based on specific discharge is given, which  forms an index of flood “risk”.

Richard et al (1997) describes a two-dimensional  mathematical model and the  determination  of inundation  risk maps for two rivers in the  Rosario region of Argentina. The  mapping was made over  both Saladillo and Luduena rivers, for floods of return periods of 50,100 and 500 years. The  studied zones embraced an areas of 7000ha, with  a population of 500,000 inhabitants. Based on the results, state and  local government are planning  non-structural rules with the  associated  legislation.  The paper concludes  that it is important to use  association simulation models for urban planning  strategies and water resources management.

Kuchment  (1997) estimating  the risk of rainfall and snowmelt  disastrous  floods using physically-based  model of river runoff generation. Disastrous  floods can be  caused  by unusual  combinations  of hydrometeorological factors and river basin conditions that have not been observed  during a long observation period. Physically-based models of runoff  generation enable one to find dangerous possible combinations of  hydrometeorological  factors and to estimate  the risk of extreme  floods. Analysis  of runoff  generation on a number of the  Russian  rivers have shown that  although the probable maximum precipitation rate is usually  larger than the probable maximum snowmelt  rate, the maximum floods of the medium and large  rivers of Russia  are of snowmelt origin. A comparison of the  calculated maximum snow belt  and rainfall  discharge  has been carried out. It has been revealed that in the same region, the probable  maximum  discharge  may be of snowmelt of rainfall  origin depending on the  river basin area and runoff  generation mechanism. The  problem of the assignment of the meteorological inputs was discussed. To  demonstrate the  approaches  used, the  results of investigation of extreme floods generation  based on numerical simulation of the River Sosna, the  River Seim and the River Uda were shown.

(iii)       Socio-Economic Factors

Oriola (1994) observed that  various socio-cultural activities have promoted flooding in many of the Nigerian  urban environments. These  activities  are characterized  by stream or river  channel encroachment and abuse, increased paved surface and poor solid waste disposal  techniques, due to a high level of illiteracy, a low degree of community  awareness, poor environmental education, ineffective town  planning  laws and poor environmental management. He argued  that  government, at various levels needs to address these  issues. He concluded that flood risk  in the Ondo urban environment was a  function  of the following factors: Land-use pattern, refuse disposal  habits, the  nature of the  surrounding  buildings, distance of  building  from the course of the  streams, rainfall amount and duration, the relief  or the terrain, slope, gradient, and  other stream basin parameters.

(iv)          Combination of Hydrometeorological and Socio-Economic Factors.

Hogue et al  (1997) undertook an assessment of the  risks  involved with cyclones and  storm surges in Chitagong, the  second largest city in Bangladesh. The  study finds the extent of storm surge  flooding  and the  related risk in the  metropolitan area. To  identify the  risk, the  depth and extent of storm surge  flooding for different  probability of occurrence have  been predicted and were  expressed as  a hazard  index. The city area was divided into five categories of land-use: industrial area, commercial areas, planned housing  areas, unplanned housing areas and mixed areas. For each, population density and economic importance  of the areas have been considered and were  expressed as an  importance index. Using  the hazard index and importance  index, the risk for each area was  calculated. On the  analysis, the whole  city area was classified  into four categories: the  low risk area, the risk  area, the  high risk area, and the  severe risk area.               

Ologunorisa (2004) undertook an assessment of flood risk in the Niger Delta, Nigeria using a combination of an hydrological techniques  based on some measurable physical  characteristics  of flooding, and social-economic techniques based on vulnerability  factors. Some of the  physical characteristics of flooding  selected  include depth of flooding  (metres), duration  of flood (hours/weeks),  perceived frequency of flood occurrence, and relief or elevation (m) while the  vulnerability factors selected include proximity to hazard source, land use or dominant economic activity and adequacy of  flood alleviation schemes and perceived extent of flood damage. He  derived  rating  scale for the nine  parameters selected, and 18  settlements randomly selected across  the three ecological zones in  the region were  rated on the basis of the parameters. Three flood risk  zones emerged from the analysis. These are the  severe  flood risk zones, moderate flood risk zones and low flood risk zones. Some  strategies for  mitigating the hazard of flooding in the  region were identified.

Georgakos et al (1997) undertook an estimation of flash flood potential for  large areas in United States of America. A methodology  for determining the potential for  flash floods in small basins within  large geographical area was presented. Geographical Information System (GIS) technology was used to assimilate digital  spatial data, remotely  sensed data, with physically-based  hydrological  - hydraulic  models catchment  response. The  methodology  used digital terrain  elevation data, digital river reach data, and the US Geological  Survey land-use  and land-cover  data to produce estimates of the  effective rainfall volume of a certain duration required to produce flooding  in  small streams. This  flood potential index is called  threshold runoff.  For operational application, soil water accounting models were used to yield estimates of effective precipitation  over areas  of 1000km². Maps of flash flood potential could then be  constructed using  remotely-sensed and on-site data. Examples of application in various regions of the United State of America were  discussed.

(v) Geographical Information System (GIS)

Okoduwa (1999) applied  Geographic  Information System (GIS) in the prediction of urban flooding  in Benin City. Nigeria. This was  achieved by creating a digital database  of selected  variables such as land use, land cover and soil strength. The  software used was  Arcview 3.1 and the overlay  techniques in GIS was used for analysis. The  result of the analysis  showed  high flood  prone areas, medium flood prone areas  and low flood prone areas.