Introduction

Floods are one of the worst natural disasters that can happen anywhere in the world, causing significant damage and having a significant socioeconomic impact on various sectors (Sun et al. 2020; Zhao et al. 2022; Yang et al. 2023; Wu et al. 2023). An increase in the frequency of flooding is one of the effects of climate change (Xu et al. 2022; Gao et al. 2023). In terms of both quantity and losses, floods represent 34 and 40% of all-natural disasters worldwide, respectively (Lyu et al. 2019; Xie et al. 2021; Yuan et al. 2023). There are significant varieties that influence the environment such as  elevation differences and surface coverage, socioeconomic factors (e.g., population and built-up areas), and disaster-prone bodies (e.g., transportation facilities in different locations) (Zhang and Chen 2019; Huang et al. 2021; Zhou et al. 2021a).

It is important to manage and control floods that affect the agricultural economy, which is the mainstay of densely populated areas (Miller and Shaw 2015; Zhou et al. 2021b; Yin et al. 2023d; Ma et al. 2023). This is technically impossible without the accurate mapping of flood hazards and risks. Rapid population growth is rapidly accelerating, increasing pressure on farmland, which has increased the change in land use and land cover (LULC) that could lead to flooding (Li et al. 2021; Moisa and Gemeda 2021; Negash et al. 2021; Moisa et al. 2022; Merga et al. 2022).

In Ethiopia, the most frequently flooded areas include the Awash River Basin, the Wabi Shebelle River Basin, and the Baro-Akobo Basin. The Wabi Shebelle River Basin is one of the Ethiopia’s major river basins and is prone to flooding (Wondim 2016). Flooding has forced thousands of people to flee their homes and caused massive economic losses and environmental damage (Getahun and Gebre 2015).

The Wabi Shebele River Sub-basin is one of the major areas being affected by flooding. The majority of flood damage in the study area is seemingly caused by a few days of intense rainfall with an average intensity of 10–200 mm/hr and a total quantity of precipitation of 100 mm. Flood events have occurred frequently in the basin, manifesting as flash floods in the lowland areas, as shown by the state of the river banks and sheet erosion (Ethiopian Ministry of Water Resource 2004).

Previously, several studies have been conducted on the significant impacts of flooding on most parts of Ethiopia. According to Abaya et al. (2009) studies, flooding frequency and magnitude have significantly increased in the Gambela region due to a lack of flood-specific policy, little risk assessment, and weak institutional capacity, which directly result in death, injuries, and disease outbreaks in the local communities of the region. Flooding from Lake Tana's surface water discharge displaced the local community from their land each year (Tarekegn et al. 2010). The study done by Getahun and Gebre (2015) indicates that the upstream and middle are more vulnerable to flood risk than the downstream parts of the Fetam Watershed of the Upper Abbay Basin of Ethiopia.

However, most previous scholars did not emphasize the identification of flood risk-prone areas, which enhances prioritization for future conservation practices in monitoring the significant impacts of flood risk on natural resources and human life. Identifying and mapping the spatial distribution of the flood-prone zone enhances land-use planners and natural resource manager’s ability to apply risk management strategies at the right time. Geographic Information Systems (GIS) and remote sensing, along with multi-criteria evaluation (MCE) techniques, play a great role in assessing flood risk. Therefore, the present study aimed to fill the existing research gap by assessing flood risk-prone areas using geospatial techniques and the analytical hierarchy process (AHP) in the Wabi Shebele River Sub-basin of southestern Ethiopia.

Materials and methods

Description of the study area

The Wabi Shebele River Sub-basin (current study area) is located in the West Hararghe Zone of Oromia Regional State. Geographically, it is situated between latitude and longitude of 8°0′0′′ to 8°48′0′′ N and 40°24′0′′ to 41°12′0′′ E. The study area covers about 5596.34 km2 with an elevation ranging from 703 to 2988 m above mean sea level (Fig. 1).

Fig. 1
figure 1

Location map of the study area

Full size image

The major rainy season in the study area lasts from July to September, and the minor wet season from February to April. The annual average rainfall varies between 760 and 1200 mm. The wettest months are June, July, and August, while the driest months are May, June, March, September, and October. The local rain is heavy and showery, with low volumes and a sluggish onset. Typically warm winter weather prevails. Summer and winter average temperatures are 25 °C and 30 °C, respectively (Zonal Irrigation Development Authority 2014).

Data sources and descriptions

Several data sources were used to achieve the stated objective in the study area. The geographic area of the study area is located between 67 paths and 54 raw reference systems. Landsat 8 operational land imagery (OLI) and thermal infrared sensor (TIRS) of cloud free were downloaded from the USGS website for the year 2022 with a spatial resolution of 30 m for land-use and land-cover classification. The National Metrological Agency also used the five stations in the study area to obtain rainfall data for the year 2022: Bedesa, Gelemso, Abomsa, Sheki-Husen, and Seru stations. In addition, soil type data were collected from the Ministry of Water and Irrigation Engineering, while Shuttle Radar Topographic Mission (SRTM) was downloaded from the USGS website to calculate the slope of the study area. Finally, GIS data from 2018 were used to calculate the equidistance of the water source to the study area (Table 1; Fig. 2).

Table 1 Data sources and description
Full size table
Fig. 2
figure 2

Methodological flowchart

Full size image

Software package used

ArcGIS 10.3 software was used to analyze and visualize the geospatial data of the study area. In addition, IDRES Selva 17 software was applied for parameter pairwise comparison, whereas ERDAS 2015 software was used for image preprocessing, layer staking, and false color band combination during image classification. Similarly, ArcSWAT software was also used to generate the stream network and boundary of the study area.

Data analysis

Rasterization was performed for appropriate parameter weighted overlay analysis under GIS environments. The weighted importance value of each parameter was calculated in IDRISI Selva 17 software by using the artificial hierarchy process (AHP) method for pairwise comparison based on their percent of influence in determining the status of flood risk in the study area. During pairwise comparisons, the relative importance value for each parameter in identifying flood risk-prone areas was decided by an expert’s opinion based on Saaty’s 9-point weighing scale. The pairwise comparison matrix was made by assigning a relative importance value for each parameter and normalizing the eigenvector of the factors by their cumulative total (Wind & Saaty 1980).

Parameters used for flood risk assessment

Soil types

Soil types may be considered one of the critical factors in defining flood-prone areas. It has a significant impact on the rate of precipitated water infiltration and water-holding capacity. Four different major soil types existed in the Wabi Shebele River Sub-basin. These are Calcic Regosols, Eutric Cambisols, Haplic Xerosols, and Haplic Yermosols. According to Wondim (2016), infiltration has a significant impact on the availability and quantity of surface runoff produced by the rainfall runoff process. As a result, clay soil infiltrates at a much lower rate than sandy soil (Hagos et al. 2022).

Slope

The slope of the study area was generated from STRM DEM by degree. According to Hagos et al. (2022), slope was classified into 0°–2°, 2.1°–5°, 5.1°–15°, 15.1°–35°, > 35.1° and very high, high, moderate, low, and very low, respectively. Steep slopes tend to reduce the amount of infiltration of water into the ground; the water can flow quickly down to rivers as overland flows. In addition, steep slopes also cause more through flow within the soil. In reality, due to the fact that water accumulates gradually inside the process of moving alongside the slope from high to low and flood typically nearest happens within the lower place, so the slope length is an element that affects both floods as well as the opportunity of flooding.

LULC analysis

LULC in the sub-basin plays an important role in flooding by delaying or accelerating surface flow. These LULC types of the study area were classified from Landsat OLI/TIRS of 2022 by using supervised classification with maximum likelihood. As a result, the LULC types of the Wabi Shebele Sub-basin were classified into forest, farmland, grassland, shrubland, and settlement. Land use, population density, and population growth in the basin are all considered factors that contribute to flooding (Kundzewicz and Stoffel 2016; Yin et al. 2023c). Even though the study area is largely covered in forest, flooding has a significant impact on farmland and settlement areas. This demonstrates unequivocally that the flood magnitude is strongly negatively correlated with forest cover and positively correlated with cultivated land (Liu et al. 2017; Zhu et al. 2022).

Distance from river

Proximity to a water source can determine the severity and status of flood risks in the study area. The amount of water generated from rainfall, streams, tributaries, and underground recharge increases the volume of the water and causes surface water discharge in the form of floods. This phenomenon indicates that the area of the study area nearest to the water body was more vulnerable to flood risk than the fairest one (Ullah and Zhang 2020; Yin et al. 2023a).

Rainfall

The status and seasonality of the rainfall are also key determinants for evaluating the impacts of flood risks in the study area. Both rainfall and flood risks possess positive relationships (Li et al. 2022; Yin et al. 2023b). Heavy rainfall conditions result in the occurrence of higher flooding risks on the surface of the study area and vice versa. In addition, rainfall also increases the volume of the water in the water body and increases the severity of surface water flooding in the study area (Assefa 2018).

Elevation

Elevation shows the location of the study area above or below sea level. It can determine the speed of runoff and the status of flood accumulation in the study area. An area with lower elevation was characterized by lower flood risks and higher flood accumulation and vice versa.

Accuracy assessment

Accuracy assessment was also made to evaluate the quality of classification output in this study. Error matrix based on assessment of the overall accuracy, producer’s accuracy, user’s accuracy, and kappa coefficient was utilized. The main goal of accuracy assessment is to quantitatively assess how efficiently the pixels were tested into the correct land-cover classes (Bhatt 2008). To validate the result with ground truth, field data were collected using handheld global positioning system (GPS), and finally, land-use and land-cover classes were identified and mapped for further analysis. The result of accuracy assessment provides an overall accuracy of the map based on an average of the accuracies for each class in class in the map (Girma et al. 2020; Moisa et al. 2023).

$${\text{OAC}} = \frac{\Sigma Xij}{N} \times 100$$
(1)
$${\text{Khat}} = \frac{{\left( {{\text{Obs}}{-}\exp } \right) }}{{\left( {1{-}{\text{Exp}}} \right) }}$$
(2)

where OAC is over all accuracy, UAC is user accuracy, PAC is producer accuracy, Khat is Kappa statistics, N is total number of samples, Xij is the diagonal values, Xi+ is the column total, X + i is row total, r is the number of categories, Obs is accuracy reported in error matrix (overall accuracy), and Exp represents correct classification.

Parameters rating scale for flood risk assessment

The scale variation in all targeted parameters may cause the presence of various flood risks in the study area (Table 2). Estimation of such variation enables land managers to identify flood risk-prone areas and prioritize future conservation.

Table 2 Scaled and weighted of flood risk and hazard factors
Full size table

Analytical hierarchy process

During this study, six ecological parameters, including soil type, slope, LULC, elevation, rainfall, and proximity to a water source, were assigned for accessing flood risk-prone areas in the study area. For each of the targeted parameters, weighted values have been calculated by using the analytical hierarchy process (AHP) method and the multi-criteria decision analysis (MCDA) tool based on a 9-point scale (Saaty 2008). The reliability of the parameter weighted value was evaluated based on the relative importance value assigned by exports for each parameter (Table 3).

Table 3 The relative importance value
Full size table

Pairwise comparison matrix has been performed by using assigned relative importance value of each parameter in EDRIS Selva 17 to evaluate the flood risk vulnerability of the study area (Table 4).

Table 4 Pairwise comparison table
Full size table

The pairwise comparison acceptability was based on the calculated value of consistency ratio (CR) which must be less than 0.1 (Moisa et al. 2023), and the consistence ratio in the present study was 0.06. The consistency ratio for parameters pairwise comparison was determined as Eq. (3).

$${\text{CR}} = \frac{{{\text{CI}}}}{{\text{ RI}}}$$
(3)

where CI represents consistency index, and RI represents random consistency index.

Consistency index was the measure of how all targeted parameters were interrelated and calculated as Eq. (4).

$${\text{CI}} = \frac{\lambda \max - n}{{n - 1}}$$
(4)

where n indicates the number of parameters, and \(\lambda \mathrm{max}\) refers to the principal eigen value of parameters that can be calculated from multiplication of the total horizontal summation of given intensity importance value and parameters weighted value. The normalized principal eigenvector was obtained from averaging the normalized relative weight of the parameters.

The random consistency index was an assigned constant number based on the number of parameters used in the study (Table 5).

Table 5 The random index scale
Full size table

Final flood risk analysis

All criteria map of the parameters were overlaid by using parameters suitability index to calculate flood risk-prone area in the study area. According to Debesa et al. (2020), flood risk area was calculated as Eq. (5).

$${\text{SI}} = \sum Wi \times Xi$$
(5)

where SI is suitability index, Wi is weight of factor I, Xi is normalized criterion of factors, and ∑ is their the summation.

Results and discussion

Flood risk factors analysis

To determine flood risk, environmental background conditions such as soil type, slope, elevation, rainfall, and distance from the river, as well as LULC, which are the key flood-producing factors, were analyzed in the study.

Soil types

Soil may be considered one of the critical factors in defining flood-prone areas. It has a significant impact on the rate of precipitated water infiltration and water-holding capacity. The statistical analysis of soil types in the study area shows that a large portion, which accounts for 69.04%, is covered by Haplic Xerosols, followed by Eutric Cambisols, 11.64% (Table 6; Fig. 3). According to Wondim (2016), infiltration has a significant impact on the availability and quantity of surface runoff produced by the rainfall runoff process. As a result, clay soil infiltrates much lower rate than sandy soil (Hagos et al. 2022). Eutric Cambisols were highly vulnerable to flood risk than other soil types in the study area. A northern part of the study area was highly susceptible to flooding in the Wabi Shebele Sub-basin. Ayenew and Kebede (2023) confirmed that Cambisol soil types were highly susceptible to flooding than nitisol soil types.

Table 6 Soil type
Full size table
Fig. 3
figure 3

Soil type of the study area

Full size image

Slope factor

Steep slopes tend to reduce the amount of infiltration of water into the ground; the water can flow quickly down to rivers as overland flows. In addition, steep slopes also cause more through flow within the soil. In reality, due to the fact that water accumulates gradually inside the process of moving alongside the slope from high to low, and flooding typically happens within the lower place, the slope length is an element that affects both floods and the opportunity of flooding. The slope in the study area ranges from 0° to 74.24°. A high rating is assigned to low slopes, and a low rating is assigned to high slopes. From the total area of the sub-basin, about 1924.39 km2 (34.39%) fall under the slope range between 2.1° and 15°, which was assigned as the high rate for flooding risk (Table 7). Geographically, the central and southeastern parts of the study area were at higher risk for flooding than other directions (Fig. 4). The slope has a great influence on assessing flood risk because it controls the amount of surface runoff (Alemayehu 2007; Hagos et al. 2022; Yin et al. 2023f).

Table 7 Rate of slope for susceptibility to flooding
Full size table
Fig. 4
figure 4

Rate of slope for susceptibility to flooding

Full size image

Land-use and land-cover factor

LULC of the sub-basin plays an important role in flooding by delaying or accelerating surface flow. In the study area, the main LULC types are forest, farmland, grassland, shrubland, and settlement areas. About 2277.20 km2 (40.69%) of the total LULC is covered by forest, followed by grassland and farmland, which take up about 1460.08 km2 (26.09%) and 1184.15 km2 (21.16%), respectively (Table 8). From LULC types, farmland, settlement, grassland, shrubland, and forest were very high, high, moderate, low, and very low for flood risk in the study area, respectively. Spatially, the central and western parts of the study area were at higher risk for flooding than the northern and southern parts of the study area (Fig. 5). The result is more consistent with the previous study (Ayenew and Kebede 2023) that agriculture is more vulnerable than forest for flood risk in the case of the Dikala Watershed in Kobo Woreda, Amhara Region, Ethiopia.

Table 8 Land-use and land-cover factors
Full size table
Fig. 5
figure 5

Land-use/land-cover map

Full size image

Distance from rivers

Discharge increases as more water enters the stream through rainfall, tributary streams, or groundwater seeping in. Generally, as discharge rises, the stream’s width, depth, and velocity all increase (Kiersh 1964). Any undeveloped low-lying land close to a river, lagoon, or lake is more vulnerable to flooding as the water level rises. Therefore, the risk of flooding decreases as the distance from the river increases, and flooding increases when the distance is closer to the river (Fig. 6). The river density of the study area was high and susceptible to flood inundation in the Wabi Shebele Sub-basin. The central and southern parts of the study area were at higher risk of flooding than the northern part of the Wabi Shebele Sub-basin. Negese et al. (2022) stated that flood inundation increased as the distance to the rivers decreased in the Dega Damot district, northwestern Ethiopia.

Fig. 6
figure 6

Distance from rivers

Full size image

Rainfall factor

The massive groundwater system’s primary source of recharge is rainfall on the highlands during the rainy season (Chen et al. 2023a). The area dominated by high rainfall was more vulnerable than the area captured with low rainfall for flooding in the Wabi Shebele Sub-basin. The regions with the highest recharge rates are the northwest, southeastern highlands, and upper basin. These aquifers are refilled by the streams that rise in the eastern highlands. Seasonal flooding occurs in the highlands during the summer, and the fragmented volcanic cover is favorable for groundwater recharge (Fig. 7). The amount, duration, and spatial distribution of rainfall in catchments are some of the most important factors influencing the risk of flooding (Assefa 2018; Chen et al. 2023b).

Fig. 7
figure 7

Rainfall distribution map

Full size image

Elevation factor

Elevation was the main parameter used to assess the flood inundation of the study area. As a result, the area with moderate to low elevations was highly susceptible to flooding, whereas the area with high elevations was less vulnerable to flooding in the Wabi Shebele Sub-basin. In this study area, the upper and intermediate catchments (i.e., the northern and northwest parts of the sub-basin) have high elevation and experience low levels of flooding (Fig. 8). The central and southern parts of the study area were at high risk for flooding. However, the northern part of the study area was at low risk for flooding. The previous studies (Wondim 2016; Choubin et al. 2019; Ogato et al. 2020; Yin et al. 2023e) confirmed that low elevation was highly vulnerable to flooding and vice versa.

Fig. 8
figure 8

Elevation map

Full size image

Accuracy assessment

Overall accuracy assessment and kappa coefficients of the LULC types of 2022 were 92.5% and 0.89, respectively, in Wabi Shebele Sub-basin.

Potential flood risk map of the Wabi Shebele Sub-basin

Based on the significance of the elements and classes in the case study, weight and rank values were assigned to them. The element with the highest weight, according to expert knowledge and AHP results, was elevation (0.2899), followed by distance from rivers (0.2119), slope (0.1688), land use/land cover (0.1419), soil type (0.0849), and rainfall factor (0.1025) (Table 9). The resulting map was given a value corresponding to each pixel's factor and normalized weight after total scores were used to weight factors and rank classes.

Table 9 Level of flood risk
Full size table

According to the flood risk map (Fig. 9), the study area was subjected to have very high 527.31 km2 (9.42%), high 2133.66 km2 (38.12%), moderate 1906.44 km2 (34.06%), low 847.35 km2 (15.14%), and very low 182.54 km2 (3.26%) flood risk, respectively (Table 8).

Fig. 9
figure 9

Flood risk map

Full size image

The levels of danger for the elements at risk in this study area are different. The worst-affected districts of the sub-basin are Daro Labu, with a very high flood risk 160.6 km2 (14.7%) and Hawi Gudina district, with 284.78 km2 (13.1%) (Table 10). The sub-districts of the study area are subject to varying degrees of flood risk. This difference is primarily the result of various environmental characteristics. The results of this study suggest that areas with a high concentration of farms and bare land, as well as low elevations, are at risk of flooding. Due to all of these factors, Woredas such as Daro Lebu, Burka Dintu, and Hawi Gudina are at high risk of flooding. This finding supports Assefa (2018) and Hagos et al. (2022) who found that factors such as unpredictable drainage areas, rapidly changing topography, and low ground surface infiltration capacity all increase flood risk. While flooding occurs frequently in nature, human land-use activities typically involve destroying natural vegetation (for farming or construction), which can significantly increase runoff and raise the risk of flash floods and river floods.

Table 10 Districts of the sub-basin
Full size table

Conclusion

Assessment of flood risk is an important strategic planning tool for monitoring flood risk. This study investigated flood risk areas using geospatial techniques in the Wabi Shebele River Sub-basin of Southeast Ethiopia. The datasets used in the study include Landsat OLI/TIRS of 2022, soil types, climate (rainfall), and topographic data (slope and elevation). The collected data were analyzed by integrating geospatial techniques with an analytic hierarchy process for generating a flood risk map in the study area. Environmental factors such as soil type, slope, elevation, rainfall, and distance from rivers, as well as land-cover types, were the essential parameters for assessing flood risk in the Wabi Shebele Sub-basin.

The result revealed that about 9.42, 38.12, 34.06, and 15.14% of the total area are at risk of flooding, with very high, high, moderate, and low flood risk, respectively. This result indicates that the study area was extremely susceptible to flood risk due to topographic variation and the expansion of agricultural land in low and high land areas, which aggravates the rate of deforestation and surface runoff. As a result, within the six districts in the study area, four districts were in very high flood risk classes. Identification of the flood risk-prone area was essential for prioritizing future management strategies in the study area. Based on this study, we recommended that environmental and climate change resilience specialists, natural resource managers, and any other concerned bodies assume crucial responsibility for appropriate flood control strategies in the study area.