Flood risk assessment using analytical hierarchy process: A case study from the Cheliff-Ghrib watershed, Algeria

Floods are considered one of the most dangerous natural phenomena that spread and recur worldwide. Urbanization and irregular demographic growth, in addition to climate change, are all reasons that lead to the emergence of floods and usually result in human and economic losses (Abdelkebir et al. 2021). The extent of the impact of floods includes 170 million people each year (Hagos et al. 2022). On the local level, the phenomenon of floods is considered as the most dangerous geographical phenomenon that Algeria has known in recent decades (Goumrasa et al. 2021). In recent years, the study of the analysis and impact of flood maps has witnessed rapid development in the analysis methodology and the strategy of extracting flood impact maps (Hu et al. 2017).

Many studies have documented the assessment and analysis of flood risks in several regions of the world, through multiple approaches and strategies, for example, multiple criteria analysis (Thinh & Vogel 2007; Papaioannou et al. 2015; Arya & Singh 2021); principal component analysis and varied statistical models (Akay 2021); cluster analysis (Baborowski et al. 2012; Fernandez et al. 2016; Rahman & Rahman 2020); frequency ratio (Tehrany et al. 2015; Rahman & Rahman 2020); and an analytical hierarchy process (Chakraborty & Mukhopadhyay 2019).

One of these approaches, the analytical hierarchy process (AHP), has been widely used in several studies to analyze and assess the effects of flood risk (Yang et al. 2013; Danumah et al. 2016; Gigović et al. 2017; Das 2018; Luu et al. 2018; Hammami et al. 2019; Cai et al. 2021).

The AHP was developed by Saaty in 1980 and it is considered a mathematical technique of multi-criteria decision-making (Hammami et al. 2019). The AHP is a general theory of measurement. It is used to derive ratio scales from both discrete and continuous paired comparisons. These comparisons may be taken from actual measurements or from a fundamental scale, which reflects the relative strength of preferences and feelings (Saaty 1987).

Researchers have relied mainly on their strategy to assess the impact of flood hazards on integrating AHP approach with other approaches such as GIS and Artificial Intelligence (AI) to estimate flood risk in their location (Aydin & Sevgi Birincioğlu 2022).

One of the advantages of the AHP method is its extensibility and robustness; if the decision maker wishes, he can modify the value of a criterion or add or eliminate criteria that he deems relevant. The method allows him to readjust the evaluation previously carried out without repeating the entire hierarchy already established.

In the context of this study, the objective is not to choose one or more (alternative) solutions but to prioritize different targets according to their importance in relation to the risk of flooding using the AHP method. This allows us to deduce a flood risk map that includes several criteria apart from the classic layout of flood zone maps. A correct and accurate diagnosis by researchers of the effects of flood risk using the sequencing approach contributes to giving decision-makers the ability to develop strategic plans to reduce the effects of floods.

This mode of risk assessment by multi-criteria analysis, in this case, the AHP, on the entire basin, offers the advantage of dispensing with the hydrodynamic modeling of free surface flows by the equations of Saint-Venant and the computational convergence problems that pose this kind of physical modeling. This modeling also requires topographic surveys of watercourses, which are very costly in terms of money or need high-resolution satellite images and are difficult to obtain at affordable prices by local financial means. The criteria proposed here aim to identify, as far as possible, the factors that influence the risk of flooding, using as much free-access and free-of-cost data as possible. Indeed, this study uses 14 flood risk assessment criteria. Nine criteria (elevation, surface slope, drainage density, distance to rivers, topographic wetness index, modified normalized water index, rainfall, normalized difference vegetation index, and lithology) are used to assess the hazard and the remaining five (total population, population density, land use and land cover, distance to hospital, and distance to road) are used to obtain the vulnerability map.

Table 2 shows the type of data acquired, resolution, sources, date of acquisition, as well as the relevance of the entire primary and secondary data used for this study.

Following the building of a pairwise comparison matrix, we were able to obtain a relative significance of relevant factors using the AHP. The weight of each parameter zone is determined once it has been classified according to its relative importance. The relative relevance score was set at 1 to 9, with lower numbers indicating less essential factors and higher numbers indicating much more important factors. The pairwise comparison matrix is shown in Table 4 using a 9 × 9 matrix with diagonal elements equal to 1. To calculate the rating score, the values of each row are compared to the values of each column. Rainfall intensity, for example, is substantially more essential than land use and hence it has been given a value of 7. Row emphasizes the significance of land use. As a result, the row bears the inverse of the pairwise comparison (for example, 1/7 for rainfall intensity) (Danumah et al. 2016).

AHP is based on paired comparisons and employs hierarchical structures to depict a problem and then establish priorities for alternatives based on the user's assessment. The importance of the evaluation criteria and their weights must be decided. There are six steps in the procedure (Saaty 1977):

• Breaking a complex unstructured problem into its component factors;

• Development of the AHP hierarchy;

• Paired comparison matrix determined by imposing judgments;

• Assigning values to subjective judgments and calculating the relative weights of each criterion;

• Synthesize judgments to determine the priority variables;

• Check consistency of assessments and judgments.

One of the key points in AHP is the calculation of the consistency ratio (CR) (Saaty 1980). If CR is less than 0.1, then the mentioned matrix can be considered as having acceptable consistency.

The binary combination is based on the scale proposed by Saaty (1980) for element comparison in Table 3. The pairwise comparison is the fundamental component of the AHP process. For each pairing within each criterion, the better option is awarded a score, again on a scale between 1 (equally good) and 9 (absolutely better), whilst the other option in the pairing is assigned a rating equal to the reciprocal of this value. Each score record shows that option ‘X’ meets criterion ‘Y’. Afterwards, the ratings are normalized and averaged. Experts provide their judgment of the relative importance of one indicator against another. The pairwise comparison tables were completed by nine experts in the field of natural disasters. Their results were normalized and examined with the CR test (Danumah et al. 2016).

The principle of development is the following matrix:

• Determine the eigenvectors (Vp) of each criterion for each item is described in the following equation:

  

  (1)

With k = number of parameters and comparing Wk ratings main parameters;

• Calculate the weighting coefficients (Cp), the formula is given in the following equation:

  

  (2)

The sum of Cp of all parameters of a matrix must be equal to 1:

• normalize the matrix by dividing each element by the sum of the column;

• average each line to determine the priority vector [C];

• multiply each column of the matrix by the priority vector corresponding there to determine the overall priority [D];

• divide each global priority by the corresponding priority vector to determine the rational priority [E];

• determine the maximum eigen value (λ_max) by the following equation:

  

  (3)
• Calculate the consistency index (CI) expressed by the following equation:

  

  (4)
• Determine the CR using Equation (5). The ratio of coherence can be interpreted as the probability that the response values are completed in a random manner. In fact, the responses often have a certain degree of incoherence. The AHP method does not require that judgments are consistent or transitive, indeed, Saaty (1980) has defined the value of the CR. In the case where the value of the consistency ratio is less than 10%, the judgment is consistent and when it exceeds 10%, the assessments may require some revisions:

  

  (5)

(RI) is the random index, the values are shown in Table 4.

Hazard is considered a physical phenomenon, natural and non-managerial, occurrence of data and intensity that can cause damage by stream overflow and the extension of the field in the water flood. The hazard also refers to hydroclimatic processes and their effect on water flow. Geomorphological characteristics, including slope, drainage density, soil types (Meraj et al. 2015) and rainfall (because it is the intense rainfall that triggered flooding) are the various factors taken into account in the mapping of the hazard.

Vulnerability expresses the level of foreseeable consequences of a natural phenomenon on issues (Glynn et al. 2001) and on the other hand is the most crucial component of risk in that it determines whether or not exposure to a hazard constitutes a risk. Flood vulnerability mapping is the process of determining the degree of susceptibility and exposure of a given place to flooding. These issues include people, goods and socio-economic activities likely to be affected both quantitatively and qualitatively by a natural phenomenon. In this study, the vulnerability to flooding consists of three criteria, population density, drainage system and land use. The weights assigned to each element to determine the vulnerability extent are in Tables 7 and 8.

The term ‘risk’ is multidimensional in today's academic contexts since it is intertwined with safety, economic, environmental and social issues and has a range of meanings (Samuels 2006). A careful review of a large range of scientific papers indicated that modern flood management systems are related to three essential elements: risk (R), hazard (H) and vulnerability (V), which can be expressed in a simple form (Unisdr 2009; Masood & Takeuchi 2012; Wisner et al. 2014).

The geographical approach to risk has just developed in the recent several decades (Defossez et al. 2017). As previously established, the term ‘risk’ is multidimensional, implying that it has several definitions. The consistency can be found in the fundamental characteristics of hazard and vulnerability, which most definitions are built on.

For the present context, the definitions given by the United Nations International Strategy for Disaster Reduction (Unisdr 2009) have been followed. The definitions are (a) Hazard: ‘A dangerous phenomenon, substance, human activity or condition that has the potential to cause death, injury or other health effects, property damage, loss of livelihoods and services, social and economic disruption, or environmental degradation’; (b) Vulnerability is defined as ‘the traits and conditions of a community, system or asset that render it vulnerable to the destructive impacts of a hazard…. Vulnerability fluctuates dramatically within a community and over time.’ Flood risk management necessitates a better knowledge of the vulnerability, as a hazard becomes a disaster only when it strikes a system or community that is susceptible to its impacts (De Brito et al. 2018). However, the assessment of vulnerability and its combination with the hazard to obtain risk differs with natural events. Thus, defining vulnerability essentially depends on the purpose of the assessment. Vulnerability is in fact hazard-specific—‘it is perfectly possible for a system to be vulnerable to one hazard yet resilient in the face of another’ (Ciurean et al. 2013); and (c) Risk: ‘The combination of the probability of an event and its negative consequences’. A methodological workflow comprising a few consequent tasks has been designed to assess the flood risk and its hazard and vulnerability components for the Cheliff-Ghrib watershed.

This division of risk into ‘hazard’ and ‘vulnerability’ is classic. But in fact, it is purely conceptual and helps to analyze the phenomenon. With the ease and efficiency of multimeter analysis – in this case, AHP – it is not necessary to obtain a risk map.

According to the flood sensitivity map in Figure 7 and Table 9, 16.11% of the study area, or 222.11 km², appears to be very sensitive to flooding in the southern and southwestern parts of the watershed of the Cheliff-Ghrib. About 55.15% of the total area of the study area, or 760.34 km², is occupied by the moderately venerable parts. Parts of Ouled Antar, Ouled Helal and M'fatha, totaling 394 km², or approximately 28% of the territory, are regions of low sensitivity.

For those in charge of planning and managing activities, identifying flood risk locations as done in this study is crucial. Hydrologic and hydraulic models were frequently used in the past to evaluate probable flood damage and inundation zones for specific recurrence periods. These models essentially only take into account how waterways' flow and transportation are balanced. Two main phases make up the flood risk assessment used in this study. First, variables or causes that actually result in floods are identified. Second, AHP-based Multi-Criteria Evaluation (MCE) is used in a GIS system and these approaches are assessed for their effectiveness in identifying flood-prone locations.

To calculate the flood risk index for the different urban zones, the results are expanded. This strategy might be more practical than models that use hydraulics alone, hence, combining the two strategies is advised.

The multi-criteria analysis approach used in mapping areas at risk of flooding required a combination of a hazard map (Elevation, Slope, DD, DTR, TWI, MNWI, Rainfall, NDVI and Lithology) and vulnerability map (TP, PD, LULC, DTH and DR). The resulting map indicates that 22.63% of the study area is of high flood risk. The municipalities of Chorfa and Ouled Bouachra are extremely vulnerable to the possibility of floods according to the data. As a consequence, decision-makers may utilize the generated map as a reference for future preventative actions, better land use planning and flood risk management under climate change. To prevent further harm, policymakers must establish strict rules addressing unchecked development, the occupation of lands near rivers and the placement of obstructions in waterways. High-spatial-resolution satellite data must be used to do more in-depth mapping in high-risk areas in order to provide a research viewpoint and enhance and improve the findings. This study also emphasized the validity and crucial function of geoinformation methods in the evaluation.

The results of this study back up the idea that combining AHP and GIS methodologies can effectively employ spatial data for decision-making processes in flood hazard mapping.

Future research efforts might be concentrated on how AHP can be coupled with other approaches like fuzzy logic for additional investigations and to further benefit from the adaptability of AHP in urban flood studies, as recommended by Boroushaki & Malczewski (2010). AHP–GIS should be used to iteratively carry out the judgment process in studies on the effects of longer rainfall/flood records and flood risk assessment.

All relevant data are available from an online repository or repositories (see Table 2).

The authors declare there is no conflict.