Flood susceptibility mapping in the Bilate catchment, Ethiopia

Extreme hazards afflict both wealthy and poor countries creating massive financial disruptions, and major human sufferings (Hosseinzadehtalaei et al. 2021). Among other types of natural disasters such as volcanic eruptions, wildfires, explosions, and tsunamis, floods are the most frequent. It occurs in every corner of the planet (Das 2021). Floods have a lot of environmental and human consequences (Maskong 2019). Uncontrolled floods cause an overflow of water to destroy farming land, built environment, and infrastructure.

Flood susceptibility mapping is, however, necessary for identifying flood-prone areas to develop mitigation strategies (Antzoulatos et al. 2022). It helps to identify locations with flood susceptibility useful for effective flood risk management (Das 2018).

Flood can be predicted using hydraulic and hydrologic models. Hydraulic models such as MIKE, HEC-RAS, and SOBEK can simulate flood in one-dimensional (1D) or two-dimensional (2D) states (Teng et al. 2015; Malik et al. 2021). The hydrodynamic models can be used to create a physical framework for simulating various flows and sediment transport scenarios. The flood simulation models generally focus on river processes, taking overland processes as inputs of 1D or 2D hydrodynamic or hydrologic models. Due to the discrete structure of overland flow and unknown-dynamic boundary conditions, such classical approaches are not capable of fast and reliable spatio-temporal estimations of overland flows, and require detailed and well-organized spatial data (Ozcelik & Gorokhovich 2020).

Hydrologic models are also used to study the water and sediment transport in a basin where excess water can be represented in the form of flood. However, from the various models and their respective accuracy and applicability, it can be concluded that there is no single best model because of the nature of environmental predictions. Every model can produce only an approximation of the reality it is attempting to illustrate (Galavi et al. 2013). Rather, there are many plausible solutions, depending on the purpose and needed complexity. Therefore, the practice of hydrologic modelling has, in general, included too much reliance on mathematics at the expense of true knowledge, and suffers from a need for a more rigorous evaluation of appropriateness. Typically, model selection tends to be more of a function of familiarity than appropriateness.

Many distributed models that have been developed over the years have gained popularity for their performance and capabilities. A few examples of such models are the MIKE SHE, TOPMODEL, and Soil and Water Assessment Tool (SWAT). Distributed hydrological models use a large number of parameters in simulating a watershed and accordingly are subjected to model uncertainty (Ayele et al. 2022; Edamo et al. 2022a, 2022b, 2022c; Ukumo et al. 2022a, 2022b). Although, their uncertainty can be quantified for real-world applications (Mirzaei et al. 2021), hydrological and hydraulic models do not suit the objectives of this study.

After all, the correct representation of flood propagation has proven to be a significant challenge for researchers and engineering experts all over the world, even on a small-scale watershed, due to the complicated natural processes (Di Baldassarre et al. 2010). The most comprehensive hydraulic models therefore must use simplification to simulate the flooding process. Consequently, the model accuracy is affected by assumptions (Ali 2018). Modelling approaches are restricted by the lack of data on ever-changing river sections and long-time hydrological observations (Edamo et al. 2022a, 2022b; Ukumo et al. 2022a, 2022b). A topographical method based on digital elevation models (DEMs) was developed in conjunction with the use of real-time predictive hydraulic models to improve large-scale awareness of flood risks (Ukumo et al. 2022a, 2022b). This may be a way to cope with the current flood mapping gaps, decrease the data dependency requirements, and promote hydraulic modelling and remote sensing applications. The scarcity of data in the Bilate catchment further limits its simulation (Kruczkiewicz et al. 2021).

Studies were conducted in various locations throughout the world to create an accurate flood susceptibility mapping method utilizing various decision-making and machine-learning methodologies via geospatial approaches. Edamo et al. (2022a, 2022b, 2022c) attempted to identify highly likely flood-prone zones using an analytical hierarchy process (AHP). Tehrany et al. (2014) used a weighted overlay analysis in geographic information system (GIS) to identify possible flood-prone areas in West Bengal, India. However, their weighted overlay method does not quantify each location's possibility of fitting to specific sets from numerous input rasters (Sepehri et al. 2020). Many academics have also adapted similar methods to quantify flood-prone zones using geospatial modelling (De Risi et al. 2018). Tehrany et al. (2014) used a machine-learning algorithm and numerous multivariate statistical methodologies to forecast flood susceptibility.

Experts from all around the world have recently emphasized the importance of cost- and time-effective decision-making approaches based on GIS techniques in flood mapping. The weights that have been assigned determine the precision of the AHP approach. It is worth noting, however, that various researchers in separate studies assign different weightings and rankings to different parameters. The availability of data for numerous elements, their quality, and geographical conditions all play a role in the effort to build a precise flood susceptibility map.

In this regard, multi-criteria decision analysis (MCDA) has been considered a suitable method to perform flood susceptibility mapping and evaluations as a result of its flexibility and the possibility of facilitating the dialogue between stakeholders, analysts, and scientists (Cinelli et al. 2014). The AHP is simple to understand and has a good software support. The MCDA method could be used to overcome the limitations of the hydrodynamic models (Odu 2019). As a result, the MCDA has gained widespread acceptance as a useful technique for assessing complicated choice issues (Abdel Hamid et al. 2020). One of the MCDA strategies is the AHP, which is a decision-making procedure that contains combining multiple-choice criteria into a hierarchy (Wind & Saaty 1980). One of the shortcomings of the AHP is the rejection of certainty in the spatial decision-making process (Singh et al. 2021). However, for flood susceptibility studies, the AHP is yet a viable option (Parsian et al. 2021).

Concluding from the reviewed literature and practiced methodologies, considering the limited data availability in Bilate, the MCDA approach is selected to map the flood-susceptible area in the catchment. However, an option for assessing flood risk is to combine remote sensing and GIS tools with the AHP, particularly in areas with data scarcity (Mundhe 2019), and it is also followed here.

The data used in this study were collected from different sources. The rainfall data was obtained from the National Meteorological Agency of Ethiopia (NMAE) for the periods of 2010–2021. Soil was downloaded and used to prepare the soil map of the study area and supplied from the Food and Agriculture Organization (FAO). The source is https://www.fao.org/soils-portal/data-hub/soil-maps-and-databases/en/. The LULC data with resolution of 30 × 30 m was downloaded from United States Geological Survey (USGS): https://earthexplorer.usgs.gov/. Digital Elevation Model (DEM) with the resolution of 12.5 × 12.5 m were downloaded from Alaska satellite Facility (ASF): https://asf.alaska.edu/. The geology and geomorphology was obtained from Ministry of Water Irrigation and Energy of Ethiopia (MOWIEE).

The flood drivers are introduced in this section. Elevation, as the first factor, is the most critical characteristic of a basin that influences flood occurrence. Flooding is unlikely in high-elevation places, but it is more likely in lower-elevation areas. Water naturally flows from a higher point to a lower point. As elevation increases, the possibility of flooding decreases (Das & Pardeshi 2018; Ogato et al. 2020a). Flooding may occur more quickly in low-elevation locations than in higher-elevation places with a sharper slope.

Next, LULC is a key variable in predicting which areas have a high storm surge (Norman et al. 2010). In vegetated areas, flood threat is reduced due to the negative connection between flooding and green density. Storm runoff increases in residential areas and highways (Tehrany et al. 2013). The type of land used by sentient creatures as well as natural processes is displayed in the LULC pattern of an area (Ajin et al. 2013). Landscape management in the rural context offers a number of benefits in terms of environmental interactions. Changes in the land surface over time, as well as the effects of social and biological forces, must all be examined (Malik et al. 2021). These behaviours are regarded as catalysts for natural disasters that are catastrophic. Natural calamities such as floods are exacerbated by human actions such as deforestation and urbanization. Different forms of LULC might have a great effect on the water flow; for example, when bare land is present, rainwater runoff is often higher, whereas agricultural land or grass coverings hinder water overland flow and produce stagnation.

Soil type (ST): Soil parameters (permeability, soil layer thickness, rate of infiltration, and moistness of the soil prior to rain) have a direct impact on the rainfall–runoff process (Rimba et al. 2017). The ability of the soil to function as a sponge and absorb water will be influenced by its structure and infiltration capability. As soil infiltration capacity decreases, more surface runoff occurs, increasing the danger of flooding.

Slope and aspect: The slope is a key topographic feature in regulating the surface water flow in hydrology (Mojaddadi et al. 2017; Das et al. 2018). When the slope rises, the velocity of flow will also rise. A quick fall in the slope reduces infiltration but intensifies the surface runoff. The slope reduction stagnates large volume of water causing flooding. It is a crucial metric for surface zones that are extremely susceptible to flooding. One of the important criteria in defining whether or not a given rainfall will generate runoff is slope. The infiltration capacity of the soil is affected by the slope. The runoff increases because the soil does not have the time to absorb the water through its pore spaces. Slope and flood sensitivity are inversely related in that a steeper slope results in higher runoff and lower flood risk in that area. In contrast, in flat locations, the drainage rate is low and it causes water to stay on the surface of the ground and increases the level of the flood risk. When water is provided at a rate that exceeds the soil's capacity for infiltration, it rushes off the sloping terrain, causing flooding.

Another consideration is the aspect, which has an impact on the direction of flooding water flows as well as soil humidity (Cea & Costabile 2022). As a result, the feature has a secondary effect on flooding.

Rainfall: The reason for river flooding is mainly heavy rain; any water that does not enter the earth runs downstream as runoff. Excluding glacier portions, rain is the only origin of surface water. It has a tough link with river discharge and thus has a direct influence on the occurrence of floods. Flash floods in semi-arid locations can be triggered by unexpected rainfall.

Distance from the river (DR): The majority of flood-prone areas are typically situated along rivers. Because the distance from the river influences the frequency of floods and the river flow, it is an important criterion for identifying flood-prone areas in a catchment. The farther you are from the river, the less likely you are to get flooded (Liu et al. 2021).

Curvature: It is the degree to which a curve deviates from a straight line, or a curved surface deviates from a plane. Curvature describes the shape of the soil surface and reflects the capacity for water accumulation (Costache 2019). Curvature affects the flooding water budget. The negative value regions are involved in the runoff convergence operation (Towfiqul Islam et al. 2021). A positive value of curvature represents a convex surface, zero a flat surface, and a negative value a concave surface (Das 2019). The curvature splits the divergent and convergent runoff zones.

Flow accumulation (FA): Flow accumulation was calculated using the flow direction raster. Every column in the flow accumulation raster is awarded a discharge profile based on the number of cells that flow into it. An increase in flood sensitivity should, in this case, be accompanied by an increase in flow accumulation (Abdel Hamid et al. 2020).

Flow direction (FD): The ability to discern the flow direction of each raster pixel is one of a surface's hydrologic properties. The flow direction is a grid whose value represents each cell's steepest point in terms of flow direction.

Terrain ruggedness index (TRI): It measures the uniformity in the topographic distribution of altitude (Riley et al. 1999). This technique is quite useful for determining whether a region is flat or rugged (Das 2021). Because of their flat nature, regions with a low TRI rating have a greater risk of flooding. Therefore, very low TRI values can be found in flood plain.

Population density (PD): Population density affects the occurrence of flood. It is understood that the higher the amount of flood protection, the larger the growth in population density and the number of assets vulnerable to flooding over the last few decades (Bibi et al. 2019). If the population is densely distributed over the given area, the more prone it is to floods (Ogato et al. 2020b). Population growth increases the possibility of flooding as well as its potential effects by putting more pressure on drainage systems and encouraging urban development in flood-prone areas (Kablan et al. 2017). The rising sea levels and changing rainfall patterns are expected to make floods more frequent and intense over the coming decades. The flood occurrence and intensity will increase due to climate change and the increasing population The world's impoverished are particularly at risk of flooding and susceptible to its effects. The population data of the study area were collected from Central Statistical Agency of Ethiopia. The average population density calculation method was used to calculate the number of people per square kilometre.

Drainage density (DD): In general, the density of drainage increases with decrease in infiltration capacity of soil. The DD layer of the basin was produced in ArcGIS using line density tool. DD is calculated using the formula: DD = L/A, where L is the length of the drainage channel in the catchment (km); A is the area of the watershed (km²) (Ouma & Tateishi 2014).

Surficial geology (SG): Deals with the Earth's resources, their structure, and the processes that operate on them (Das 2019). The surficial geology includes drainage feature (pattern, density, etc.) and parent rock. The surficial geology affects the catchment runoff by formation of different permeable and impermeable rocks.

Geomorphology: The scientific study of landforms and the processes that shape them is known as geomorphology (Das 2019).

An AHP is a simple decision-making strategy that considers the relative importance of numerous elements (Németh et al. 2019). It combines multiple-choice criteria into a hierarchy, comparing numerous options for each criterion, evaluating their relative values, and determining an overall status of the alternative's affordability, value, and danger. By giving suitable weights to numerous input criteria, the multi-criteria decision-making technique has been widely used to address multifaceted difficulties (Odu 2019). In this study, the AHP was used to compute the weights of all input features for flood risk mapping. The pairwise comparison matrix comes in handy when determining weight values. After that, each pairwise comparison matrix element was divided by the total of each column to create the normalized matrix. The average value of each row was used to create the final weight value of the equivalent parameter. The comparison matrix A for any problem can be represented by the following decision matrix (Jabbar et al. 2019) (Equation (9)):

Step 1.

• A.

  Construct a pairwise comparison matrix (n × n) for criteria with respect to objective by using Saaty's 1–9 scale of pairwise comparisons as shown in Table 1. In other words, it is used to compare each criterion with each other criterion, one-by-one.

• B.

  For each comparison, we will decide which of the two criteria is most important, and then assign a score to show how much more important it is.

• C.

  Compute each element of the comparison matrix by its column total and calculate the priority vector by finding the row averages.

• D.

  Weighted sum matrix is found by multiplying the pairwise comparison matrix and priority vector.

• E.

  Dividing all the elements of the weighted sum matrix by their respective priority vector elements.

• F.

  Compute the average of this value to obtain λ_max.

• G.
  Find the consistency index, CI, as follows (Equation (4)):

  

  (4)

  where n is the matrix size.
• H.
  Calculate the consistency ratio, CR, as follows (Equation (5)):

  

  (5)
• I.

  Judgment consistency can be checked by taking the consistency ratio (CR) of the CI with the appropriate value as shown in Table 2. The CR is acceptable, if it does not exceed 0.10. If it is more, the judgment matrix is inconsistent. To obtain a consistent matrix, judgments should be reviewed and improved.

Step 2:

Step 3:

In pairwise comparison matrix A, the element a_ij of the matrix is identified as the relative importance of the alternatives ith and jth with consideration to criterion A as shown in Equation (9) where a_ji is the reciprocal values of a_ij.

Saaty (Wind & Saaty 1980) proved consistent reciprocal matrix, the largest Eigen value is equal to the size of comparison matrix, λ_max = n. Then, Saaty gave a measure of consistency, called CI as deviation or degree of consistency.

Knowing the CI, the next question is how do we use this index? Wind & Saaty (1980) proposed that we use this index by comparing it with the appropriate one. The appropriate CI is called Random Consistency Index (RI). The Random Index (RI) is the average of CI values of various sizes of comparison matrices (Edamo et al. 2022a, 2022b, 2022c). The RI represents the random index that refers to the consistency of the pairwise comparison matrix which is randomly generated. It is obtained as the average of the random consistency index, which was computed by Wind & Saaty (1980) using a sample of 500 matrixes randomly generated. Wind and Saaty randomly generated reciprocal matrix using scale 1/9, 1/8, …, 1, …, 8, 9 and get the random consistency index to see if it is about 10% or less. Then, they proposed what is called CR, which is a comparison between CI and Random Consistency Index (RI).

The CR was computed to determine the degree of consistency between the weight values of distinct parameters throughout the weight value determination step. A CR value of less than or equal to 0.1 indicates that the pairwise comparison matrixes are stable, but values more than 0.1 indicate that the matrixes should be reconsidered.

In flood susceptibility mapping, several flood-influencing factors are critical and multiplied by the associated weight value (Parsian et al. 2021) to prepare the flood susceptibility map. Weighted overlay was utilized to evaluate many factors of varying relevance in order to make a final decision. It helps to demonstrate how raster analysis and raster arithmetic can be utilized to solve spatial challenges (Aksha et al. 2020).

Researchers from all over the world have recently stressed the relevance of decision-making approaches based on GIS techniques in flood-prone zonation, which is both cost- and time-effective (Abdel Hamid et al. 2020). The probable flood susceptibility map of the Bilate catchment is developed using the AHP, which takes into account 18 topographic and climatic parameters. All of these elements, as well as the outcomes, are covered in the following.

There were 18 flood-influencing factors discovered in this research. The identified factors are the most deriving agents of flood hazard in the Bilate catchment. All the flood-causing factors have no equal impact (Bui et al. 2019).

The soil groups of the study area are reclassified into the following five classes: Pellic vertisols, Euric fluvisols, Leptosols, Chromic vertisols, and Orthic solonchaks (Figure 3(b)). The Leptosols and Pellic vertisols are the dominant soil type in the Bilate catchment (Table 4) and cover an area of 2,123.3 km² (37.7%) and 1,830.7 km² (32.5%), respectively. The nature of the soil in a given area influences the flood levels and possibly regulates the infiltration capacity. Leptosols have a high infiltration rate and a high degree of a flood risk.

The slope of the Bilate catchment was classified into five slope classes: very high slope, high slope, medium slope, low slope, and very low slope. The slope of the Bilate catchment fluctuates from 0° to 66° (Figure 3(c)). It is observed that the slope range (0°–3.4°) covered the largest land 2,133.2 km² (37.9%) (Table 5). In hydrological research, slope is a significant topographic factor in controlling the flow of surface water (Tehrany et al. 2013). According to Zaharia et al. (2019), regions with slopes greater than 15° have less water accumulation. The obtained average slope causing flood in the catchment was 5.13°.

The LULC of the Bilate catchment was grouped into five classes (Figure 3(e)). The LULC map of the Bilate catchment include 3,892.2 km² (69.2%) of intensively cultivated, 882.3 km² (15.7%) of moderately cultivated, 590.5 km² (10.5%) of shrub-land, 221.9 km² (3.7%) of grassland, and 38 km² (0.7%) of water body (Table 7). Land use and land cover are crucial factors in identifying areas that are at risk of flooding (Ghosh & Kar 2018).

Figure 3(f) shows that the upstream of the catchment receives the high annual rainfall, whereas the downstream of the watershed receives the low annual rainfall. Even though the upstream rainfall amount is slightly higher than that of the downstream rainfall of the catchment, the downstream areas have relatively flat topography and flat slope, which indicate it is highly vulnerable to flood hazard. According to Radwan et al. (2019), flood hazard increases with an increase in precipitation. Table 8 shows the various average rainfall classes, which are very low (925–1,000 mm), low (1,001–1,100 mm), medium (1,101–1,253 mm), high (1,254–1,300 mm) and very high (1,301–1,400 mm) rainfall, respectively.

Flow accumulation is among the most important criteria in determining flood hazard regions (Kazakis et al. 2015). Flooding occurs in regions with a lot of water (Tehrany et al. 2015). Flow accumulation is often lower for the lower classes of rivers in the upstream. In the downstream of the watershed, many tributaries join the main channel, which increases the flow accumulation and the flood risk. Therefore, high flow accumulation can be found in the downstream region of the study area. An increase in flow accumulation exacerbates the occurrence of flood (Abdel Hamid et al. 2020) (Figure 3 (g)). Flow accumulation is often lower for the lower classes of rivers in the upstream. In the downstream of the catchment, many tributaries join the main channel, which increases the flow accumulation and the flood risk. Therefore, high flow accumulation can be found in the downstream region of the study area. The flow accumulation and the area coverage of the Bilate catchment are given in Table 9.

One of the fundamental advantage of the GIS is its ability to differentiate the flow direction of each raster pixel. The flow direction in the map of the Bilate catchment is presented in Figure 3 (h). The flow direction is a grid whose value shows how each cell flows with respect to its closest downward slope neighbour (Sanyal 2004). Flow direction was divided into the following five categories: 1–2, 2–8, 9–32, 33–64, and 65–128. Low flow directions imply a high probability of flooding, whereas high flow directions show a low probability of flooding (Table 10). The flow direction ranged (1–2) covered the largest area (48.2%) of the catchment.

Index of aspect is derived from the elevation that defines the pixel directions per unit degree, allowing for more precise flood risk mapping judgments. It is often referred to as the horizontal direction of the mountain facing, and it determines the local climatological conditions. Aspect of the catchment was ranged from −1 to 360 (Figure 3(i)). It was divided into the following five categories: –1 to 75, 76–150, 160–220, 230–280, and 290–360 (Table 11).

The curvature map of the study area is presented in Figure 3(j). Table 12 implied that the curvature is classified into the following five classes: −15 to (−0.76), −0.75 to (−0.31), −0.3 to (−0.02), 0.021–0.35, and 0.36–14, respectively. A positive value of curvature represents a convex surface, zero a flat surface, and a negative value a concave surface (Das 2019). Hudson & Kessel (2000) observed that curvature between 1.0 and 2.0 had a greater probability of flooding. Hence, the probability of flood is very high between the curvature from 1.6 to 22 and very low between −19 and (−2.2), respectively. Similarly, curvature is also an important factor and represents the morphology of the topography (Das 2018).

The STI of the Bilate catchment ranges from 0 to 140 (Figure 3 (k)). It was classified as very high (0–48), high (49–260), medium (270–720), low (730–1,300), and very low (1,400–2,000) (Table 13). Due to silt deposition, the carrying capacity of stream channels in this catchment will be greatly diminished, and it could cause flooding. The highest STI covers an area of 2,666.3 (47.4%), and the lowest STI covers an area of 1,102.4 (19.6%). STI is sediment transport caused by flowing water (Shafapour Tehrany et al. 2019). The higher STI indicates the area with more sediment transport, and the lower STI indicates less sediment transport. The chance of flooding will be high in areas with low STI values (0–48) because these are the depositional zones.

Flood is more likely in areas with a bigger TWI (Mojaddadi et al. 2017). Lower TWI zones, on the other hand, are less vulnerable to flooding (Figure 3 (l)). The higher value was assigned for the higher TWI and the lower value was assigned for the lower TWI, respectively. As a result, the downstream of the Bilate catchment has the highest TWI and has a larger probability of flooding than the upstream of the catchment, which has the lowest TWI and has a lower probability of flooding. The lowest TWI covers an area of 265.9 km² (4.7%) in class between 6.8 and 8.3 whereas the highest TWI cover an area of 2,077.2 km² (36.9%) in class between 12 and 14, respectively (Table 14).

The lithological properties of an area influence porosity and permeability. Such factors/properties of a rock can influence flood intensity. The geology affects the catchment runoff by formation of different permeable and impermeable rocks. The Bilate catchment is separated into four primary divisions geologically such as paleozoic sedimentary rocks, tertiary volcanic rocks, quaternary volcanic rocks, and quaternary sedimentary rocks (Figure 3 (m)). The area coverage of the surficial geology of the Bilate catchment is presented in Table 15.

Geomorphology is the driving agent of flooding. The landforms of the Bilate catchment are presented in Figure 3(n). Volcanic landforms are sensitive to flooding which covers 47.1% of the study area. The area coverage of the geomorphology of the Bilate catchment is presented in Table 16. The alluvial landforms covered 13.1% of the catchment.

The population is densely distributed in the Bilate catchment (Figure 3(o)). The population density layer was further reclassified into five density classes as very high, high, medium, low and very low population densities (Table 17). The reclassified population density of the study area are <52, 53–67, 68–115, 116–203, and >200, respectively. Therefore, the lowest population density <52 covered an area of 137.1 per square kilometres (2.4%) (Table 17). Floods can occur often in highly populated places due to a lack of appropriate drainage systems (Ogato et al. 2020a). Population density was reclassified based on the assumption that the denser the population, the more prone it is to floods (Ogato et al. 2020a).

According to the classification shown in Figure 3(p), a higher DD area was very heavily affected by flood and ranked in class five, whereas a lower DD area was very lightly affected by flood and ranked in class one. The DD has a big influence on flood peaks. The control related to the length of the stream network and hillslope routes has the most substantial effects. DD considerably influences the concentration-time and thus the peak flow magnitude because the river network has a larger flow velocity. As a result, as DD increases, flood peaks rise as well. Furthermore, a long concentration-time means there are more chances for water to permeate. If the DD is high, it creates a high runoff rate and causes a higher flood risk (Shekhar & Pandey 2015). The density of drainage is divided into the following five categories: 0–0.00021, 0.00022–0.0006, 0.00061–0.0012, 0.0013–0.002, and 0.0021–0.0037 (Table 18).

The TRI is a secondary geo-morphometric measure for describing and quantifying local relief. The TRI influences the river flooding; however it was not considered in the flood susceptibility mapping in different studies (Riley et al. 1999). The TRI value of the Bilate catchment was presented in Figure 3(q). It provides new information on terrain morphology and is suggested for usage, particularly in characterizing flood damage sections.

The topographic ruggedness index measures the uniformity in the terrain distribution of altitude (Riley et al. 1999). This technique is quite useful for determining whether a region is flat or rugged (Das 2021). The regions with a low TRI rating have a greater risk of flooding because of their flat nature. As a result, very low TRI values can be found in flood plains. The TRI of the Bilate catchment is classified into five classes; as very high (0–0.37), high (0.38–0.45), medium (0.46–0.52), low (0.53–0.61) and very low (0.62–0.89), respectively (Table 19).

It plays a key role in identifying flood sensitive regions. As the distance from the river is far from the likelihood of flooding is less. The distance range in the Bilate catchment is found between 0 and 19,000 m (Figure 3(r)). Further development of this study's findings may focus on the inclusion of a greater number of potential conditioning factors. An area close to the river hence has more flood than areas farther away from river and vice versa. The majority of Bilate people live near the river bank, making them subject to flooding. The distances were classified into five classes as 0–1,520 m, 1,521–2,550 m, 2,551–5,010m, 5,011–9,000 m and 9,001–20,000 m, respectively (Table 20). The nearest distance to the watershed (0–1,520 m) covered an area of 834.7 km² (14.8%) whereas the farthest distance to the watershed (9,001–20,000 m) cover an area of 94.8 km² (1.7%). Hence, the largest area of the watershed covered an area of 2,270.7 km² (40.4%) which is in between 1,521 and 2,550 m (Table 20).

The weight of the factors were obtained and overlay process was accomplished in ArcGIS 10.3 environment. The CR, the Random Consistency Index (RI), and the CI were calculated as 0.061, 1.49 and 0.078, respectively.

The use of multiple flood conditioning factors is useful to determine the capacity of effective flood danger management in flood-prone catchments since there is variation in flood causing from catchment to catchment and region to region. Rimba et al. (2017) used six flood-influencing factors (LULC, slope, soil, DD, and rainfall) and found that rainfall was the most flood causing factor in Okazaki City, Japan. However, the number of variables was limited, making the outcome highly doubted.

A pairwise comparison matrix was constructed to determine the proper weight, indicating the contribution of each component that influences flood (Table 21). The matrix normalization method is used in this study to compute the importance of the various flood-related parameters.

The degree of flood susceptibility is divided into the following five categories: very low, low, moderate, high and very high, with 9.3, 32.6, 41.2, 10.8, and 6.1% of the entire Bilate catchment, respectively (Figure 4). In order to avoid future flood circumstances, authorities should pay special attention to regions classified as high flood zones. Using geospatial approaches (Das 2018) attempted to identify likely high flood risk zones using an AHP. The advantages of the multi-criteria approach was to prepare the flood map for the whole catchment. It is a challenging task to complete flood susceptibility mapping in a catchment using hydrodynamic models (Di Baldassarre et al. 2010; Tehrany et al. 2019). AHP helped to overcome the shortcomings of hydraulic/hydrologic models. The map developed in this study will play a significant role in flood risk management. It will also help as the source of information for further researches. The identified flood causing factors can be used in different parts of the world for flood-related investigations.

Flooding is a destructive incident that can occur nearly anywhere along a river's route, especially downstream. Using remote sensing and GIS methods, flood conditioning factors were identified to define flood-prone zones in the Bilate catchment and identify more vulnerable locations. To combine the decision measures in this study, the multi-criteria method was used. LULC (16.8%) is the most flood-influencing factor in the catchment. The whole Bilate catchment suffered flood susceptibility ranging from very high to very low. The very high flood susceptibility covered approximately (6.1%) of the total land area. The flood susceptibility map provided in this study will be a useful resource for construction managers, decision-makers, administrators, financiers, and administrative authorities involved in organizing response and emergency services during floods. In addition, it will play a significant role in assessing flood risks not just in the Bilate catchment but also in other flood-prone watersheds prone to flood events across the nation. As a result, it is advised that more elements should be employed as an effective technique for possible flood mapping studies to increase the successful control of flood damage.

No fund was provided from any source.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.