1 Introduction

Floods frequently occur as one of the most common natural disasters on Earth (Diley 2005; Njoku et al. 2018). This natural hydrological event causes damage to the socio-economic structure and results in loss of life and property. There are natural and anthropogenic factors that affect flood formation (Njoku et al. 2018). The main causes of flood formation are intense rainfall in watershed, melting snow in the spring season, inadequate drainage systems, overflow of main drainage channels, and the conversion of forest and pasture areas into residential and agricultural lands (Adeoye et al. 2009). Population growth, especially in urban areas, is one of the most significant triggers of flood disasters (Adeoye et al. 2009; Moazzam et al. 2018). The rapid urbanization process leads to changes in land use, an increase in impermeable surfaces (roads, structures, etc.), and a decrease in urban natural vegetation and river networks. The process of urbanization leads to a decrease in urban drainage capacity. In areas where the soil’s water retention capacity is reduced, intense rainfall results in surface runoff. In urban areas, where there is a lack of vegetation and steep slopes, water flows as channels and carries a significant amount of suspended solids into streams (Zhang and Chen 2019). The inability of the riverbed to accommodate the simultaneous influx of water into the stream also contributes to flood formation. This situation results in irreparable loss of life and property, inflicting severe damage on human life and the economy (Moazzam et al. 2018).

Recently, the global warming caused by climate change has further exacerbated the situation (Ahmad et al. 2020; Moazzam et al. 2018). Carbon dioxide emissions, such as greenhouse gases, continue to be released due to human activities, accelerating the process of global warming. This warming manifests itself through effects such as rising sea levels and melting ice caps in the atmosphere and oceans (Njoku et al. 2018). Increased extreme rainfall events associated with climate change inflict significant and challenging damages to both humans and the environment (Njoku et al. 2018). This has led to an increasing number of flood events day by day. It is estimated that approximately one-third of the world’s land (82% of the population) is at risk of floods. Over 90 countries worldwide are vulnerable to flood disasters (Moazzam et al. 2018). Between 2000 and 2019, flood disasters affected 1.65 billion people globally, and 104,614 individuals lost their lives due to floods (UNDRR 2019). Additionally, it is reported that the economic loss caused by flood disasters between 2000 and 2019 amounted to 651 billion dollars (Kharb et al. 2022; Njoku et al. 2018; UNDRR 2019). Floods constitute 40.16% of all natural disasters that occurred worldwide between 1980 and 2019, making them the most common type of natural disaster. In 2022, floods surged globally. Pakistan saw devastating floods, affecting 33 million people, with 1,739 fatalities and $15 billion in losses. Monsoon floods wreaked havoc in India (2,035 deaths, $4.2 billion damages), and China ($5 billion damages). Nigeria suffered 603 deaths and $4.2 billion losses, South Africa lost 544 lives, and Brazil faced a fatal flood with 272 casualties. Eastern Australia also witnessed floods, costing $6.6 billion. Floods showed a significant increase (4.76%) compared to 2002–2022, making them one of the deadliest natural disasters, alongside storms (3.84%), earthquakes (14.82%), droughts (27.27%), and wildfires (36.36%) (EM-DAT 2023). In recent years, environmental issues such as soil erosion and nutrient loss have increased, particularly in areas with intense urbanization and agricultural activities worldwide. The rising flood risk is a significant spatial and temporal event. Alongside soil erosion, soil productivity decreases, and soil particles are transported to streams as sediment through surface runoff, leading to water quality deterioration. This situation inflicts significant damage to aquatic ecosystems. Furthermore, sediment accumulation in areas where the slope of rivers decreases leads to shallowing of the riverbed and increases the risk of flooding in these areas. Soil erosion is a significant environmental issue worldwide, contributing to soil degradation, water pollution, and consequently, an increase in flood events. These environmental problems have become a major concern globally. Assessing flood risk in a rainfall catchment is of great importance for ensuring the sustainability of soil structure, continuity of land structure, preservation of water quality, and prevention of loss of life and property by mitigating natural disasters.

In Turkey, floods are usually triggered by heavy rainfall and sometimes combined with snowmelt, resulting in high water levels and overflowing of major rivers in the country. Floods are a significant type of natural disaster in Turkey and have unfortunately caused numerous casualties and property damage. In 2020, the total number of natural disasters in Turkey was 905, with 177 of them being floods (AFAD 2021). This indicates that flood events are among the most frequently occurring natural disasters in our country. In the Eastern Black Sea Region (EBSR) of Turkey, the average annual precipitation is significantly higher than the national average, recorded as 1009.19 mm (Coşkun et al. 2017; MBM 2020). These intense rainfall amounts, coupled with factors such as steep topography in the region, adverse effects occurring on valley and riverbeds due to unplanned urbanization, inadequate infrastructure works, and inappropriate agricultural practices, contribute to disasters such as floods and landslides. Flood disasters not only have a negative impact on the livelihoods of local communities but also cause significant damage to infrastructure, agricultural lands, residential areas, and the economy. In the past thirty-five years, 606 fatalities have occurred in flood events in our country, with 224 of them happening in provinces such as Trabzon, Gümüşhane, Rize, Bayburt, and Giresun (ÇOB 2008). Particularly, the flood disaster that occurred in Trabzon in 1990 has been recorded as the largest disaster in the region to date. This disaster affected the central district of Trabzon, as well as the districts of Maçka, Akçaabat, Vakfıkebir, Çarşıbaşı, Tonya, along with 12 towns and 266 villages connected to these districts (URL-1). One of the most significant factors contributing to the loss of life and property in flood disasters in our country is the desire of people to settle in areas prone to flooding. The affordability of land and housing in these high-risk areas encourages land and property acquisition (Görcelioglu 2003; Dölek and Avcı 2017; Njoku et al. 2018). Therefore, the management of flood risk and flood prevention measures are of great importance in our country.

Indeed, various strategies are being developed worldwide to mitigate the destructive impact of floods and prevent loss of life and property. To ensure the success of these strategies, comprehensive information about high-risk areas in flood-prone regions is essential. Prioritizing cost reduction measures in areas with a high risk of recurrent floods and predicting potential flood-prone areas in advance are important steps to reduce casualties and property damage. Therefore, comprehensive flood risk maps need to be created and flood-prone areas should be modeled, taking into account climate change. In recent years, mathematical models, statistical analysis, and geographic information systems have been used as tools to predict and analyze disaster risks, aiming to reduce flood risk. These methods allow for a better understanding of disaster risks and the implementation of more effective measures against disasters (Boroushaki and Malczewski 2010; Chen et al. 2011; Papaioannou et al. 2015; Zhang and Chen 2019). Among these methods, the Analytic Hierarchy Process (AHP), which is a multi-criteria analysis technique, is widely used for predicting flood-prone areas (Papaioannou et al. 2015). AHP integrates with Geographic Information System (GIS) to organize parameters in a hierarchical structure and analyze the relationships between components in order to solve complex problems. This method is commonly used in situations where multiple criteria are identified and compared. AHP serves as an effective tool in creating flood risk maps by calculating parameters that determine flood risk using Digital Elevation Model (DEM) data and the analytical hierarchy process. This method assists in identifying flood-prone areas by determining the parameters and evaluating their relative importance. This method is commonly used by different researchers worldwide for predicting flood-prone areas. Kourgialas and Karatzas (2011), utilized ArcGIS to create a flood risk map. They used geological data, flow accumulation, land use (vegetation), slope, elevation, and rainfall intensity maps for flood risk analysis. Özkan and Tarhan (2012) prepared the flood risk map for Izmir using a digital elevation model and GIS techniques. They integrated flow accumulation, land use, slope, rainfall intensity, and elevation maps for assessing flood risk. Arianpour and Jamalı (2015) created a flood risk map using ArcGIS and ILWIS software. They incorporated geological data, soil texture, erosion, land use (vegetation), slope, rainfall, and drainage intensity maps. Dölek and Avcı (2017) utilized the weighted overlay method in ArcGIS to create the flood and inundation map for the Muş province. They used parameters such as slope, aspect, elevation, vegetation, and soil layers to construct the map. In Moazzam’s study (2018), they selected parameters like elevation, slope, aspect, curvature, plan curvature, profile curvature, proximity to roads, proximity to rivers, proximity to streams, and land use/land cover during flood susceptibility analysis. Njoku et al. (2018) employed GIS Multi-Criteria Evaluation techniques to create a flood risk map. They emphasized that floods predominantly occur in low-lying areas near rivers during the rainy season, where people tend to settle. Therefore, they highlighted the importance of using parameters related to rivers, topography (elevation and slope), and rainfall when creating flood risk maps. Planners can utilize this information to implement measures against flood risks, develop emergency plans, and respond promptly to flood events. It is possible to monitor and analyze real-time data such as precipitation, water levels, and hydrological data. This enables a quick response to flood events and facilitates more effective emergency management. The wide range of applications of AHP, including predicting flood-prone areas, demonstrates its effectiveness as a tool for disaster management and planning. Therefore, conducting flood risk analysis and compiling risk maps hold significant importance in determining and implementing disaster management strategies. These methods play a crucial role in reducing the impacts of flood disasters and developing strategies to prevent loss of life and property.

The aim of this study is create a flood risk map using GIS and AHP methods in the Söğütlü Stream watershed of Trabzon Province based on site-specific measurements. Furthermore, we attempted to address the parameters that most significantly influence flood events in our study, aiming to create a flood risk map specific to our area. We used to seven parameters which are rainfall, elevation, slope, aspect, soil type, distance to the stream, land use that make up the flood risk map in this study. Looking at other studies conducted globally and in Turkey, it is generally observed that the creation of flood risk maps relies on average values of large soil groups and existing soil data is being used rather than site-specific measurements (Dölek and Avcı 2017; Njoku et al. 2018; Nkonu et al. 2023; Oğuz et al. 2016; Özkan and Tarhan 2012). In our study, direct soil samples were taken from the watershed to create an accurate soil map specific to the watershed unlike other studies conducted. Moreover, unlike the studies where values are interpolated in the rainfall map, meteorological stations located upstream and downstream were separately evaluated. Thus, the rainfall values were interpolated specifically for the area. As a result of this, the impact of soil parameters, rainfall and other factors on the flood risk map will be more accurately reflected. Within the scope of the study, flood-prone areas of the watershed will be identified, contributing to the implementation of necessary measures. The obtained results will encompass spatial and quantitative information regarding flood events at the watershed scale, providing valuable insights to decision-makers and planners regarding soil and water conservation and watershed management. Additionally, the flood risk map values will determine the measures to be taken at the watershed. This study is significant as it sheds light on future watershed studies by considering factors such as different land uses, a large watershed area, and the occurrence of landslides and floods leading to loss of life and property.

2 Methodology

2.1 Study area

The Söğütlü Stream watershed (G42, G43 and F43) belong to the Trabzon, and are located at the EBSR in Turkey (Fig. 1). It occupies 27,480.9 ha which is consisted of 12,596 ha forest (Alnus glutinosa, Carpinus betulus, Quercus, Fagus orientalis, Picea orientalis), 2430 ha pasture, 12,130 ha agriculture, 316 ha residential, 25.154 ha other areas and its perimeter is 101.174 km. Its climatic characteristics are similar to the EBSR with the annual average rainfall is 701.32 mm and annual average temperature of 15.66 °C (during 2015–2019) in Akçaabat. The average slope and elevation are %48.30 and 1030.12 m, respectively. Natural disasters such as landslides and floods occur because of continuous and heavy rainfall in the Söğütlü Stream watershed. In addition, it has been chosen as the study area because it contains touristic places such as Çal-Camili Natural Park and is ecologically important.

Fig. 1
figure 1

The study area

Full size image

2.2 Selection and evaluation of flood parameters

The selection of appropriate parameters is crucial importance in determining the flood sensitive areas correctly (Njoku et al. 2018; Nkonu et. al. 2023). Literature studies including Dölek and Avcı (2017), Njoku et al. (2018), Moazzam et al. (2018), Papaioannou et al. (2015) used various criteria such as elevation, drainage density, slope, aspect, soil texture, population density, flow accumulation, precipitation, reflect on geographical features of the area that triggered the occurrence of floods. Dölek and Avcı (2017) and Özkan and Tarhan (2012) created flood maps of İzmir and Muş using ArcGIS weighted overlay method, respectively.

2.3 Data processing and analysis

The determination of the classification and boundaries used in the preparation of parameters (Table 2) typically relies on a research design and objectives. Researchers can draw from sources such as literature reviews, expert opinions, previous studies, and personal experience. But it is important for this approach to be based on a scientific foundation and to remain impartial. Therefore, using literature references or valid standards to support and ensure the scientific validity of such classifications is a common practice. In particular, classifications and boundaries used in this study was identified through a literature review, personal experience and expert opinions that is influenced classifications. The parameters that make up the flood risk map (rainfall, elevation, slope, aspect, soil type, distance to the stream, land use) are divided into 5 classes: very low risk (1), low risk (2), moderate risk (3), high risk (4), and very high risk (5). The data for these parameters have been converted into raster data with a cell size of 30 × 30 in the arcGIS environment. The percentage contribution of each parameter to be used in creating the map has been determined using the analytical hierarchy process.

2.3.1 Rainfall

The amount, intensity and distribution over the area of rainfall, which is one of the climate parameters, has a great contribution to the occurrence of flood events in a watershed. The waters accumulating on the land surface with heavy rains reach the river at the same time causes overflowing (Njoku et al. 2018; Özkan and Tarhan 2012). The greater the intensity of rainfall in a watershed, the greater the risk of flooding (Zhang and Chen 2019).

By obtaining data from the Regional Meteorology Directorate of Trabzon and using ArcGIS with the IDW interpolation method which can be used to determine the location and properties of unknown points relative to known points, the rainfall values were distributed throughout the region and area of interpolated values was generated. The fact that data was collected over a period of at least 5 years is also important as it provides a more comprehensive understanding of the rainfall patterns in the area. Approximately 123 virtual meteorology stations were placed at 1500 m intervals in the watershed (Özdemir and Tatar 2016). Then, precipitation values of 123 points was calculated, assuming that the annual precipitation increased by 54 mm per 100 m as recommended by Schreiber. This is important to ensure that the map accurately represents the rainfall patterns in the area.

In the case of the Söğütlü Stream watershed, the rainfall amounts vary between 279 and 1472 mm and are divided into 5 classes. Assigning the value of 1 to the class with the least rainfall and the value of 5 to the class that receives the most precipitation was used for assigning values to rainfall classes. This allows for a simplified way to compare and analyze the rainfall data, as well as to identify areas with the highest and lowest rainfall amounts.

2.3.2 Elevation

The elevation of the land surface affects the density of runoff. When rainfall reaches the land surface, it accumulates in high elevation areas and flows towards lower elevation areas. As a result, flood events typically occur in low-lying areas (Njoku et al. 2018; Nkonu et al. 2023). Therefore, areas with lower elevations are more prone to flooding (Nkonu et al. 2023; Özkan and Tarhan 2012; Zhang and Chen 2019). According to Njoku et al. (2018), elevation is a critical factor in the spread of floods. The elevation layer was generated from a DEM created using 1/25000 contour lines with ArcGIS. The raster data is divided into 5 classes, ranging from low elevation areas (0–500 m) to very high areas (2000–2500 m). The elevation layer was then reclassified into 1 to 5 values, with 1 representing the highest elevation with the least susceptibility to flooding and 5 representing high-sensitivity areas with the lowest elevation.

2.3.3 Landuse

The land use affects the infiltration rate and surface flow of precipitation water. The urban areas with extensive pavement and buildings have lower infiltration rates and higher surface runoff compared to rural areas with more vegetation cover. The forest and pasture areas, especially have low flood susceptibility due to high infiltration rates. Vegetation cover such as forests and pastures can significantly reduce the occurrence of floods by providing a protective layer over the soil that allows for greater water infiltration. As a result, preserving and restoring vegetation cover in watersheds is an effective strategy for reducing the impacts of floods. Therefore, land use/land cover is an essential factor to consider when creating a flood risk map (Njoku et al. 2018).

The land use of Söğütlü watershed has been classified into 12 categories, including settlement, building areas, orchards, meadow, mixed cultivation model, agriculture, leafy forest, coniferous forest, mixed forest, natural pastures, transitional woodland shrubs, and sparse vegetation areas. Furthermore, to evaluate the flood risk, the land use in the Söğütlü watershed has been further classified into 5 sub-risk classes. The forest areas have the least sensitivity value of 1, followed by farmland, shrubs, and bare land with a value of 2, while the most sensitive areas are built-up areas with a value of 5 (Njoku et al. 2018).

2.3.4 Slope

The slope factor is a parameter that can accelerate soil erosion and surface runoff in areas where there is no protective vegetation cover on the soil. Less sloping areas can experience faster waterlogging compared to steeper slopes, which are typically found in high-elevation regions (Njoku et al. 2018). The slope layer was generated via DEM using the ArcGIS Arctoolbox 3D tool and the resulting map is expressed as a percentage. The slope classes have been divided into 6 sub-classes. Additionally, the slope layer has been divided into 5 sub-classes, which will impact the flood risk map. The class with a value of 1 represents the least sensitive class (steepest slope; 45–76%), whereas the class with a value of 5 (most gentle slope; 0–5%) represents the most sensitive class.

2.3.5 Soil texture

Soil type plays an important role in predicting floods and mapping because it affects the permeability and infiltration capacity of soil. Soils with high permeability and infiltration rates, such as sandy soils, can absorb more rainfall and produce less runoff, while soils with low permeability, such as clay soils, produce more runoff and are more susceptible to flooding (Njoku et al. 2018; Özcan 2006). To determine the soil type in the watershed, we collected 123 soil samples from a depth of 0–30 cm. Soil texture was determined using the Bouyoucos hydrometer method. The soil types in the Söğütlü watershed were classified into seven classes, including loam, loam sand, clay, clay loam, sandy loam, sandy clay, sandy clay loam. Among these, clay soils were assigned the highest risk value of 5 due to their low permeability and infiltration capacity, followed by clay loam and sandy clay soils with a risk value of 4. Loamy soils were classified as medium risk with a value of 3, while sandy loam and sandy clay loam soils were considered low risk with a value of 2. Lastly, loam sandy soils were given a very low risk value of 1 (Njoku et al. 2018).

2.3.6 Aspect

Aspect is the direction that a slope faces, and it is known to influence climate parameters such as temperature and precipitation in a region. South aspect receive more direct sunlight and heat up faster, have less vegetation, leading to higher temperatures, greater evaporation rates and increase the risk of flooding. In contrast, north aspect receive less direct sunlight and remain cooler and more moist, which can affect the vegetation growth and infiltration rates of precipitation (Çepel 1995; Dindaroğlu and Canpolat 2017). The aspect of the land can have a significant impact on the occurrence and severity of floods in a given area (Wang et al. 2022). By including aspect in the flood risk map, we can better understand the potential flood risk in different areas and develop appropriate strategies to mitigate it.

It is common to divide aspect classes into 8 or 9 categories (Nkonu et al. 2023; Zhang and Chen 2019). The aspect layer can be further classified into 5 subclasses to reflect its impact on the flood risk map. The value of 1 was assigned to north-facing slopes with little or no direct sunlight, while the value of 5 was assigned to south-facing slopes with a high degree of exposure to direct sunlight. Other aspects, such as east and west aspect, were assigned intermediate values based on their degree of exposure to sunlight.

2.3.7 Distance to streams

The river overflows are one of the main causes of flooding. When the water level of a river exceeds its capacity, water spills over the banks and floods the surrounding areas. Flood risk assessment should take into account the proximity of a location to a river or other bodies of water. Areas closer to the riverbanks have a higher flood susceptibility than those farther away (Njoku et al. 2018; Predick and Turner’a 2007).

In the study, distances from the river were determined by examining studies conducted worldwide and in Turkey (Karymbalis et al. 2021; Njoku et al. 2018; Predick and Turner 2007; Osei et al. 2021; Özer 2008). The authors evaluated flood risk values not only based on a specific template but also considering expert opinions, landform, watershed structure, population density, surface features, and more. In our study, taking these factors into account, the riverbank interval distances in our watershed area were determined based on the land structure, watershed characteristics, topographic, and climatic features with the input of expert opinions and a literature-based approach.

To integrate river distance as a parameter in the flood risk map, the distance classes were assigned values ranging from 1 to 5, with 1 indicating the least risky areas located farthest from the river, and 5 indicating the most risky areas located nearest to the river (Table 2). As the river buffers were originally in vector data format, they were converted into raster format using ArcMap in order to make them compatible with the other layers used in the analysis (Njoku et al. 2018; Karymbalis et al. 2021).

2.4 Mapping of Söğütlü flood risk

2.4.1 Integration of MCDA and AHP into geographic information systems

Multiple Criteria Analysis (MCDA) is a structured approach used to analyze a set of alternatives or targets and rank them based on a set of criteria, in order to determine their level of preference from the most preferred to the least preferred (Boroushaki and Malczewski 2010; Chen et al. 2011; Papaioannou et al. 2015). GIS based MCDA, as introduced by Malczewski in 2006, has found widespread use across various fields in recent years (Papaioannou et al. 2015).

In this study, we used the AHP that involve multiple criteria analysis, the most commonly used pairwise comparison method worldwide and in Turkey, to develop a flood risk map for the Söğütlü stream watershed (Ersoy and Bulut 2009; Kuru and Terzi 2018; Leal 2020; Papaioannou et al. 2015; Tokgözlü and Özkan 2018; Zhang and Chen 2019).

2.4.2 Criteria weight assisnment using AHP

The comparison of different flood parameters is often conducted using AHP or similar multi-criteria decision analysis methods. The AHP is a method developed by Thomas L. Saaty in the 1970s to support decision-making in complex problems (Saaty 1980). The AHP involves creating a hierarchy of decision criteria and sub-criteria, assigning weights to each criterion based on its relative importance (Lee et al. 2008; Zhang and Chen 2019). The AHP creates a correlation matrix by comparing each factor (Saaty 1980; Mutlu and Sarı 2017; Rincon et al. 2018; Zhang and Chen 2019) (Table 1). In the Analytic Hierarchy Process, parameters are compared by conducting surveys with experts or by directly comparing parameters according to their degrees of importance. Both of these methods have their advantages and disadvantages. The survey method allows for the consideration of expert opinions but can be subjective. The advantages of the direct comparison method include a quick decision-making process and cost savings, especially in large watersheds. However, its disadvantages include a lack of objectivity and the possibility of making incorrect decisions. Researchers should take these factors into account when choosing their method and interpreting the results. In this study, the direct comparison method was applied, taking into consideration the most commonly used expert opinions worldwide.

Table 1 AHP comparison scale
Full size table

2.4.3 Conformity check and model validation

The AHP correlation process results in a calculation of the percentage of each alternative, as described by Saaty in 1980. To ensure consistency and accuracy of the model, the calculated consistency ratio of the parameters should be below 0.10, as recommended by Saaty in 1980. According to the Saaty method, if the consistency ratio (CR) is lower than 0.1, it indicates that the pairwise comparisons made during the AHP are generally consistent and reliable. In this case, researchers can proceed with confidence in the relative importance of the criteria or alternatives that were being evaluated. However, if the CR exceeds 0.1, it suggests that there may be inconsistency in the pairwise comparisons, which could lead to less reliable results. In such cases, researchers should reconsider their pairwise comparisons, review the judgment scales used, or reevaluate the AHP model to make it more consistent. Consistency is important in AHP to ensure that the model’s output accurately reflects the decision maker’s preferences and priorities. Furthermore, correlation analysis was used to investigate the relationship between flood risk and other factors such as slope, aspect, land use, and soil type. For this purpose, quantitative validation was performed with 123 systematic selected points. This approach allows for the identification of spatial patterns and trends in the data and helps to determine the influence of various factors on flood risk. The model has been successfully applied in diverse environmental conditions in other watersheds globally, including Turkey. To validate the results, the estimated flood risk measurements were also compared with those of other watersheds in the region. Furthermore, the values in the flood risk map were compared with past flood data in the area and the accuracy of the map was verified (Nkonu et al. 2023).

2.4.4 Flood risk map

In the present study, 30 * 30 m resolution raster maps consisting of soil texture, rainfall, landuse, slope, aspect, elevation and distance to the stream parameters were used as a base for mapping of flood risk. The flood risk map was created by applying the overlapped weighting process using the AHP and weighted overlay tool in the ArcGIS (Arianpour and Jamali 2015; Papaioannou et al. 2015; Zhang and Chen 2019).

The flood risk map values are divided into five risk zones: very low risk (1), low risk (2), moderate risk (3), high risk (4), and very high risk (5), as defined by Dölek and Avcı (2017) and Moazzam et al. (2018).

3 Results

3.1 Flood risk parameters analysis

3.1.1 Rainfall

According to Fig. 2 and Table 2, the areas where rainfall that is the greatest risk are the upper sections of the watershed, ranging from 1000 to 1500 m in elevation. In these areas, the railfall amount exceeds 1000 mm. The vegetation in these regions consists of dense forests, meadows, and natural grasslands. As you move from the source region to the downstream of the watershed, the rainfall amount decreases due to the effect of elevation. However, starting from the downstream and in the middle section where Düzköy district is located, the rainfall amount increases. In the Düzköy region, the rainfall decreases up to a certain elevation and then increases again. The annual total precipitation at the Düzköy meteorological station, located in the middle section of watershed, is approximately 615.3 mm. The annual total precipitation at the Akçaabat meteorological station, which is closest to the downstream of the watershed, is approximately 701.32 mm (MBM 2020).

Fig. 2
figure 2

Flood risk map based on rainfall (a), elevation (b), distance to streams (c), land use (d), slope (e), soil (f) and aspect (g) data in the Söğütlü stream watershed

Full size image
Table 2 Statistics and ratings of flood parameters
Full size table

3.1.2 Elevation

In the elevation map, 17.86% of the study area is shown to have a very high risk of flooding, particularly in the downstream of the watershed, where the elevation is below 500 m (Fig. 2b). This region exhibits a high vulnerability to flood hazards caused by the high discharge of water from surrounding high areas and lower sub-watersheds (Kazakis 2015). The areas with low elevation have been designated as the highest-risk zones, as indicated in Table 2.

3.1.3 Landuse

One of the most important factors influencing surface runoff in a watershed is the density of vegetation cover in the area (Ahmad et al. 2020; Arabameri et al. 2018; Karymbalis et al. 2021; Özcan 2006). Forested areas in the Söğütlü Stream watershed generally start from the middle elevations of the watershed, around 900 m, and extend up to 2200 m. The agricultural areas of the watershed, on the other hand, extend from the downstream of the watershed to the beginning of the source area along the valleys formed along the banks of the Söğütlü Stream (OGM 2019). Therefore, approximately 45.51% of the Söğütlü watershed is very high and moderate risk.

3.1.4 Slope

In the watershed, areas with high slope degrees prevent surface waters from infiltrating the soil and cause accumulation in the direction of the slope. Floods generally occur in low-slope areas. Approximately 8.72% of the study area consists of areas with slope degrees lower than 10°. Particularly, regions with agricultural areas in the middle of the watershed and settlements in the mouth of the watershed enable the accumulation of surface waters in high volumes. In general, low-lying and flat or gently sloping areas are more susceptible to floods and water inundation because steep slopes facilitate faster flow, thereby allowing rapid drainage of surface waters (Tehrany and Kumar 2018). These surface waters carry the potential to inundate low land areas and lead to flood formation. The slope degrees of the study area is shown in Fig. 2e and Table 2.

3.1.5 Soil texture

Approximately 37.45% of the study area consists of areas with clay soil, which carries a very high risk. Clay soils are generally found in the downstream and southwestern parts of the watershed where agricultural fields and settlements are located, while sandy soils are found in natural grassland areas in the source region of the watershed. Clay soils have poor drainage conditions and can be prone to seasonal flood inundation. This condition negatively affects the water holding capacity of soil. The spatial distribution of soil types is shown in Fig. 2f.

3.1.6 Aspect

The watershed consists of shaded aspect covering an area of 157.9 km2, sunny aspect covering an area of 116.27 km2, and flat areas covering an area of 0.62 km2 (Fig. 3g). It can be observed that the majority of the Söğütlü watershed is comprised of shaded surfaces. In the aspect map, 42.48% of the study area is shown to have a very high and modarete risk of flooding.

Fig. 3
figure 3

Flood risk map (a, b) Söğütlü stream watershed

Full size image

3.1.7 Distance to stream

The rainfall on the watershed is collected through surface runoff and leaves the watershed through the drainage network (Osei et al. 2021; Özer 2008). Generally, as the number of streams in a waterhsed increases, it becomes possible to discharge the rainfall without causing damage (Görcelioğlu 2003). Looking at the map in Fig. 2c, it can be observed that the risk of flooding is very high along the riverbanks as we move from the downstream of the watershed towards the north and the upstream regions. These areas particularly include residential areas, agricultural lands, and urban areas.

3.1.8 Flood risk mapping in Söğütlü stream

In the Söğütlü watershed, as seen in Fig. 3a, there are four risk classes: low risk (2), moderate risk (3), high risk (4), and very high risk (5). The low-risk areas cover an area of 18,327.76 ha, the moderate-risk areas cover 8048.73 ha, the high-risk areas cover 1102.34 ha, and the very high-risk areas cover an area of 2.07 ha. In the presented study, it was determined that the highest flood risk occurs in the downstream and middle parts of the Söğütlü watershed, where there are weak vegetation and areas of settlement and agriculture. In these areas, locations with high flood susceptibility were observed along the main stream in the downstream section, on the tributaries in the northwest part of the middle section, and along the main stream and riverbanks throughout the watershed (Moazzam et al. 2018). Additionally, some areas of natural pastureland with weak vegetation were also found to have a high flood risk. Nearly half of the Söğütlü watershed consists of dense forested areas. Forested regions extend from the middle section of the watershed towards the source area as the elevation increases. Settlements in the watershed are generally located near water bodies in the downstream and middle sections (Njoku et al. 2018). Particularly, the urban area in the downstream section of the watershed is classified as a 4th class in terms of flood risk. Zhang and Chen (2019) noted that floods occur in areas with lower elevation and slope, dense river network distribution, and high rainfall frequency. The correlation analysis of flood risk parameters in the Söğütlü stream watershed is provided in Tables 3, 4, and 5).

Table 3 Correlation matrix between flood risk parameters of Söğütlü watershed
Full size table
Table 4 Comparison matrix
Full size table
Table 5 The weights obtained for flood risk parameters according to the AHP method
Full size table

3.1.9 Identification and model validation

According to the AHP, the criterion of distance from the river had the highest weight value of 31.2%, while the aspect criterion had the lowest weight value of 3.2%. The consistency ratio (CR) was found to be 3.7% (Table 4). Our findings suggest that the results are acceptable for generating the flood risk map. The correlation analysis (α = 0.01) revealed that rainfall, slope, elevation, soil type, and land use had respective correlations of 0.18, 0.27, 0.39, 0.38, and 0.43 with flood risk. Based on these findings, it can be concluded that land use demonstrated the strongest association with the variation of flood risk (Table 3). Due to the fact that MCDA-AHP is mostly used in watershed for flood risk mapping in Turkey. Özkan and Tarhan (2012), Dölek and Avcı (2017), and Koralay and Kara (2023) have conducted studies to create a flood risk map. According to our flood risk map, it demonstrates consistency with past flood events in the watershed. In 2018, a flood incident occurred in Helvacı of Düzköy. According to our map, these areas are characterized by high and very high flood risks. Furthermore, a major flood disaster with loss of life and property occurred at the downstream of the watershed in 1990 (Fig. 3c).The results are consistent statistically, past flood and literature. Hence, the results are considered to be reliable (Gürgen 2004; Kankal and Akçay 2019).

4 Discussion

The increasing population and economic development in the EBSR are placing growing pressure on natural resources to meet demands. With the population growth and technological advancements, urbanization and development are taking place, leading to the degradation of vegetation. The uncontrolled expansion of residential, commercial, and industrial areas has resulted in rapid urbanization, encroaching upon agricultural lands, and converting forest and pasture areas into agricultural fields. As a consequence, the physical urban infrastructure, informal settlements, and inadequately planned housing are increasing around the city, causing harm to natural resources. Similar studies conducted in other regions support the notion that population growth and urbanization exert pressure on natural resources by expanding into vast areas. In the EBSR, the combination of population growth and topographic conditions has greatly contributed to the establishment of settlements in the watershed and especially along riverbanks, transforming other land uses into developed areas (Asabere et al. 2020). These changes in land use, coupled with the effects of global warming, have intensified pressure on the hydrological cycle and impacted the environmental sustainability needed for the conservation of natural resources. The EBSR and Trabzon have been grappling with severe flood disasters over the years. Over the past thirty-five years, our country has experienced 606 fatalities due to flood events, with 224 of these occurring in provinces like Trabzon, Gümüşhane, and Rize (ÇOB 2008). Among these, the flood disaster that occurred in Trabzon in 1990 stands out as the largest disaster in the region to date. The disaster had a significant impact on the central district of Trabzon, as well as on districts such as Akçaabat, Vakfıkebir, Tonya (URL-1). Therefore, detailed studies on the status and evolution of flood hazards would be beneficial in identifying flood-prone areas and recommending appropriate measures. Generally, models used to identify flood-prone areas take into account certain parameters that contribute to flood formation (Hong et al. 2018). Kourgialas and Karatzas (2011) utilized the MCDA tool of ArcGIS to create a flood risk map by incorporating geology, flow accumulation, land use (vegetation cover), slope, elevation, and precipitation intensity maps. Özkan and Tarhan (2012) determined flood risk areas using GIS techniques and considering factors such as flow accumulation, land use, slope, precipitation intensity, and elevation maps. Dölek and Avcı (2017) utilized the ArcGIS weighted overlay method to assess flood and inundation risks by considering factors like slope, aspect, elevation, vegetation cover, and soil layers. Tehrany et al. (2015) emphasized that floods in a region are influenced by its morphological, geological, topographic, and hydrological characteristics. Therefore, selecting the parameters that contribute to flood formation is a crucial step in identifying flood-prone areas. Dekongmen et al. (2021) proposed that infiltration and surface flow velocity are primarily determined by high slope and soil properties. The GIS-based MC-AHP approach used in our study determined the spatial distribution of flood-prone areas in the Söğütlü Stream watershed in Trabzon and evaluated the results. The most commonly used seven parameters, considered as the most influential factors in identifying flood-prone areas in the literature, were taken into account. The findings contain important results for predicting where floods have occurred or are likely to occur in the future. The flood risk map reveals four risk classes in the Söğütlü Stream watershed: low risk (2), moderate risk (3), high risk (4), and very high risk (5). The low-risk areas cover an area of 18,327.76 ha, moderate-risk areas cover 8048.73 ha, high-risk areas cover 1102.34 ha, and very high-risk areas cover 2.07 ha. The flood risk map indicates that 33.31% of the study area has a moderate to high probability of experiencing floods. The obtained map clearly shows that the majority of high-risk areas are located near residential areas. In the Söğütlü Stream watershed, the neighborhoods of Söğütlü, Özdemirci, Şinik, Cevizli, and Yenicami fall under the high-risk category from the mouth to the source of the stream. Additionally, areas near the Şinik and Cevizli neighborhoods are classified as very high-risk areas. The neighborhoods situated in the moderate-risk section include Komarlı, Helvacı, Çiçeklidüz, Çınarlık, Tatlısu, Fıstıklı, Yeşiltepe, Muratoğlu, Akdamar, Zaferli, Gümüşlü, Aptioğlu, Sertkaya, Cami, Kuruçam, Kıran, Acısu, Ortaalan, Ambarcık, Konaklı, Koçlu, Yeni Mahalle, Bahçecik, Yeşilce, Demirkapı, Aykut, Ortaorman, Alazlı, Doğankaya, and Çayırbağı. The remaining neighborhoods in the watershed are located in low-risk areas. Upon examining the entire watershed, it is evident that the majority of residential areas are highly vulnerable to flood disasters during the rainy season. A significant number of past flood events within the study area (such as Düzköy and its surroundings, Helvacı and its surroundings, and the downstream of the watershed) occurred in high-risk areas according to the flood risk map, which confirms our findings. This situation is further exacerbated by sudden heavy rainfall and settlement and urbanization along the riverbanks. Additionally, our results indicate that areas with forest cover have a very low flood risk. Therefore, it can be concluded that vegetation plays a role in preventing flood occurrences. Wang et al. (2022) suggested in a study conducted on Xuanwu District in China that higher population density increases the flood vulnerability of a region, indicating a significant correlation between population density and flood vulnerability. In the study by Diakakis and Deligiannakis (2013), it is highlighted that most flood events occur at night and in the rural areas of the country. Furthermore, considering the environmental conditions, it was determined that fatal incidents typically occur on flood-sensitive asphalt roads and bridges, with many of them taking place on asphalt river crossings constructed on the riverbed of usually dry torrents or waterways. Similarly, several studies have reported a direct correlation between land use changes and an increase in flood-prone areas (Adnan et al. 2020; Dams et al. 2013; Kuru and Terzi 2018; Njoku et al. 2018). According to Moazzam et al. (2018), floods generally occur frequently at elevations ranging from 274 to 317 m, and there are typically no floods at higher elevations in the study area. The analysis highlights that floods mainly occur near riverbanks and rarely at distant locations. Similarly, in our study, it was found that riverbanks and areas surrounding streams have a very high susceptibility to flooding. Ahmad et al. (2020) stated in their study that floods in Pakistan are commonly triggered by heavy rainfall and occasionally exacerbated by snowmelt, leading to high flood events in the country’s major rivers. The risk of flood disasters in the coastal areas of Guangdong province is shown to be higher than in the northern mountainous regions. The spatial distribution of flood risk exhibits specific patterns along the coast and rivers, which can be influenced by economic and human activities (Zhang and Chen 2019). In our study, it was observed that areas with high flood risk were characterized by low rainfall, low elevation, and low slope. Li et al. (2019) reported that low-lying areas (with an elevation lower than 300 m) have a higher probability of flood occurrence. Adjei-Darko (2017) stated in their study that intense rainfall is the primary cause of floods and inundations. In their study, Chatzichristaki et al. (2015) concluded that, in addition to heavy rainfall, human intervention within the torrent bed played a significant role in the cause and mechanism of the flash flood. Nkonu et al. (2023) demonstrated that geomorphological factors such as high elevation and gentle slopes in the lower parts of the study area create favorable conditions for flood formation. They noted that the northern portion of the study area, where the topography changes abruptly, exhibits low to very low flood susceptibility due to higher elevations and slope angles greater than 12°. This indicates that our study area is highly vulnerable to flood occurrences. Similarly, Karymbalis et al. (2021) observed that low slope angles increase the probability of flooding. They also reported that areas with low flood risk are predominantly mountainous regions. Because rainfall flows rapidly from higher slope areas to lower slope areas and accumulates there, increasing the risk of flooding. Settlement areas are also typically concentrated in low-slope areas. Consequently, these areas experience an increase in both loss of life and property. When examining our map, it becomes evident that the region with the highest flood risk is located in the downstream section of the watershed, characterized by low slope and dense residential areas. Oğuz et al. (2016) mentioned in their study that flood risk is high in areas with high rainfall, low slope gradient, and widespread agricultural fields. In our study, areas with high rainfall are predominantly covered by dense forest areas, grasslands, and pasturelands. Therefore, the fact that areas with high rainfall exhibit low flood risk can be attributed to the presence of such vegetation cover, which reduces the amount of surface runoff that can lead to flood disasters, as indicated by Kazakis et al. (2015). Additionally, areas with higher elevations and steeper slopes are less prone to flooding as they do not promote water accumulation and stagnation. However, the areas with the highest flood risk, according to our findings, are located near streams and rivers, as well as in densely populated areas. Zhang and Chen (2019) stated that flood risk is prevalent in agricultural areas, coastal regions, and along rivers. Moazzam et al. (2018) reported in their analysis results that floods occur rarely in distant areas and mostly near the banks of rivers. They emphasized that the areas with the highest occurrence of floods are characterized by shrubs, meadows, and water bodies. These findings align with our study, which supports this information. In terms of soil type, areas with higher sand content in the soil may exhibit lower flood risk values (Adjei-Darko 2017; Karymbalis et al. 2021; Njoku et al. 2018). The findings of this study strengthen previous findings by indicating that the Söğütlü watershed has a high likelihood of flood occurrence due to its low slope and elevation characteristics.

Model analysis of the study, uncontrolled urbanization, settlement near riverbanks, and improper land use may result in high flood risk in these areas in the future. The findings of the study, along with the existing conditions in the study area, suggest that the GIS-based MCDA-AHP model can predict flood events and differentiate between high and low flood sensitivity classes. The model demonstrates an objective and impartial approach while making this distinction. In the flood risk map, an attempt is made to find the closest accurate result by comparing the parameters that have a significant impact on the occurrence of flood events in a watershed in a hierarchical structure, along with expert opinions. Accordingly, it provides guidance to planners by predicting where floods may occur in the future. Conducting studies related to floods in large watersheds is time-consuming and, as a result, can lead to loss of life and property. Through GIS, it is possible to find the closest accurate results while saving time and expenses. This way, practicality is provided for making quick decisions in the field, taking necessary measures, and preventing potential loss of life and property. Additionally, the model’s accuracy can be analyzed by comparing it with the flood risk map and past flood events, ensuring its consistency. Our results demonstrate between the predicted flood sensitivity map and actual data. The findings of this study are consistent with previous studies conducted by Dölek and Avcı (2017), Koralay and Kara (2023), Oğuz et al. (2016), Özkan and Tarhan (2012), and Tokgözlü and Özkan (2018).

5 Conclusion

In this study, the focus was on analyzing and predicting flood-prone areas in the Söğütlü stream watershed using a GIS-based MSCA decision analysis. The modeling method was chosen for the ease of measuring the watershed size and time-saving, and a flood risk map for the area was generated. In this model, the climatic, edaphic, biotic characteristics of the watershed, which contribute to flood formation, were selected as the determining factors. Using the GIS-based AHP model, the weight of each selected parameter was calculated, resulting in a flood risk map divided into 5 classes. The model results indicate that higher degrees indicate areas more prone to flood inundation. The flood risk map shows that 33.31% of the study area has a moderate to high probability of experiencing floods.

Areas with high flood risk were found to be generally located in the downstream part of the watershed, where the population is densely concentrated and close to riverbanks. These areas were characterized by low land morphology, low elevation and slope, intensive urbanization, and impermeable surfaces. This indicator confirms the reliability of the model based on historical flood records and particularly its effectiveness in distinguishing between areas of very low and very high flood sensitivity in the current study.

This study, utilizing the information obtained about areas prone to flooding, will be useful in reducing the damages caused by floods and implementing future land use planning measures. In this context, engineers, planners, and decision-makers can benefit from research findings in selecting the most appropriate measures for sustainable land use in soil and water conservation efforts.