Investigation of the flood event under global climate change with different analysis methods for both historical and future periods

Global climate change is one of the biggest challenges of today's world. This event, which occurs as a result of the complex interaction of human activities and natural factors, causes radical changes in the climate systems of our planet. There are many causes of climate change. These reasons can be stated as follows. Fossil fuel use: the burning of fossil fuels (coal, oil, and natural gas) causes large amounts of carbon dioxide (CO₂) and other greenhouse gases to be released into the atmosphere. These greenhouse gases accumulate in the atmosphere, preventing the sun's rays from reflecting back from the Earth's surface and leading to the warming of the planet (Maina et al. 2020; Romanello et al. 2022; Gaillot et al. 2023; Tumala et al. 2023). Deforestation: the rapid destruction of forest areas increases the CO₂ level in the atmosphere. Forests absorb and remove CO₂ from the atmosphere through photosynthesis and are therefore part of the carbon cycle. With the destruction of forests, this natural source of carbon absorption disappears and its accumulation in the atmosphere increases (Panday et al. 2015; Wolff et al. 2021; Zvobgo & Tsoka 2021). Industrialization and industrial activities: growth in the industrial sector leads to an increase in activities such as energy production, production processes, and transportation. These activities result in energy use and the release of by-products into the atmosphere. Fossil fuels, especially used for energy production, increase the emission of greenhouse gases. Agricultural practices: agricultural activities, especially livestock and some practices such as rice farming, result in the release of greenhouse gases such as methane and nitrogen oxides into the atmosphere. In addition, the cutting of forests for agricultural production is also a factor that triggers climate change (Saad et al. 2019; Q. He et al. 2022; Vandrangi 2022; Dolinska et al. 2023; Fatoki 2023; Kolapo & Kolapo 2023; Minh et al. 2023). Rapid population growth: rapid population growth leads to an increase in resource demands and more consumption in areas such as energy, water, and food. This, in turn, leads to more greenhouse-gas emissions and faster depletion of natural resources (Garba & Abdourahamane 2023; McCool et al. 2023; Zoll et al. 2023). Atmospheric particles and aerosols: particulate matter and aerosols in the atmosphere are factors that have complex effects on climate. Some aerosols reflect the sun's rays, while others can trap heat, causing the atmosphere to warm (Zhang et al. 2016; Li et al. 2019; Haywood 2021). Natural processes: natural processes such as volcanic activity, changes in solar radiation, and changes in seawater circulation can also contribute to climate change. However, the contribution of factors under human influence is seen more clearly and effectively.

As there are factors that cause climate change, many negative consequences are seen due to these reasons. Temperature increase in the atmosphere: with the accumulation of greenhouse gases, the average temperature in the atmosphere is increasing. This causes changes in climate systems and air quality (Sanchis et al. 2020; Abbasian et al. 2021; Araújo-Silva et al. 2022). Extreme weather events: global climate change is leading to more frequent and severe extreme weather events. These include extreme rainfall, floods, droughts, storms and hurricanes (Qin et al. 2023; Bolan et al. 2024; Calabrese et al. 2024). Sea-level rise: global climate change causes sea levels to rise with the melting of glaciers and the increase in water resources. This leads to effects such as erosion, salinization and loss of habitats in coastal areas (Roy et al. 2023; Tuan et al. 2023; Calabrese et al. 2024). Impact on water resources: climate change has a significant impact on water resources by affecting the water cycle. Drought can lead to dwindling water resources and water crises, while excessive rainfall can increase the risk of flooding (Alhamid et al. 2022; Dzirekwa et al. 2023; Tuan et al. 2023). Changes in agricultural production: global climate change causes changes in climatic conditions that have significant effects on agricultural production. Temperature increase, drought or excessive rainfall can lead to loss of productivity and food-security problems in agricultural areas (Li et al. 2023; Omotoso et al. 2023; Shah et al. 2024). Biodiversity reduction: climate change reduces biodiversity by having negative effects on natural habitats. Consequences such as habitat loss, difficulty in migration of species and deterioration in ecosystem balance occur (Sangha 2022; Suppula et al. 2023). Health risks: global climate change carries risks that also have negative effects on health. Factors such as increasing temperature, air pollution, and pollution of water resources can accelerate the spread of infectious diseases and lead to an increase in health problems (Faye & Braun 2022; Guihenneuc et al. 2023; Thaker et al. 2023). Economic impacts: global climate change also has serious effects on economic activities. Effects such as the destruction of economic resources by natural disasters (floods, hurricanes, etc.), a decrease in agricultural production, infrastructure damage, and an increase in insurance costs can challenge economic stability (Farajzadeh et al. 2023; Jiao et al. 2023).

In this study, flooding, which is one of the important consequences of global climate change, is emphasized. Floods are natural disasters that occur as a result of water exceeding the carrying capacity of normal flow channels due to reasons such as excessive rainfall, snowmelt, and water release by dams. Climate change is increasing the intensity and frequency of flood events due to factors such as temperature increase in the atmosphere and changes in weather cycles. Increasing temperature increases the amount of water vapor in the atmosphere, which can lead to heavier rainfall (Xu et al. 2023). In addition, climate change can increase the severity of storms and hurricanes by increasing the amount of moisture in the atmosphere. This increases the risk and impact of flooding. However, climate change can also affect the snowmelt process, leading to an increase in flooding events (Chen et al. 2023). Increased temperature leads to melting of glaciers and reduction of snow cover. This causes water resources to melt and become fluid at a faster rate, increasing the risk of flooding. Climate change is also associated with sea-level rise. Sea-level rise increases the risk of flooding in coastal areas (Y. He et al. 2022). With the increase in population density and residential areas in coastal areas, sea-level rise can exacerbate the impact of flooding events. Flooding events have serious impacts on society, the economy and the environment. Economic impacts such as property damage, destruction of agricultural land, and damage to infrastructure and transportation systems occur. At the same time, floods can lead to loss of life and injuries (Wasko 2022). Environmental impacts such as infrastructure damage, pollution of water sources, and destruction of natural habitats also occur. Effective flood response and risk mitigation measures must be taken into account in conjunction with climate change. Many studies have been conducted on flood risk analyses.

Some of these studies have been done on the creation of IDF (intensity–duration–frequency) curves (Mailhot et al. 2007; Silva et al. 2021; Zhao et al. 2021, 2022; Mianabadi 2023; Agakpe et al. 2024). IDF curves are curves that show precipitation intensities of various durations (5, 10, 15, 30 min, 1, 2, 6, 12, 24 h) according to various rotation periods. These curves are used in the project design of water structures, design flow calculation and flood risk analysis. However, most of the IDF curves are generated from historical data from the study areas. Since data that include future climate change are not used, it is not considered appropriate to use the existing IDF curves in the water structures planned for future periods. Therefore, the first of the analyses made in this study was to estimate the IDF curves of the study area to cover the years 2023–2098 by using the equidistance quantile matching (EQM) method using global climate-model data. The data obtained was used to determine whether there will be flood risk in the future and to interpret the flood areas in the flood risk map created through the analytical hierarchy method, which is the second method used in the study.

There are many methods and studies for the decomposition of global climate--model data. Through the HYETOS program within the R program, daily maximum precipitation data can be separated into standard-time precipitation data (Tayşi & Özger 2022). Similar parsing can be done through artificial neural networks (Agarwal et al. 2021, 2022; Ahmmed et al. 2022; Ghobadi & Ahmadipari 2023; Waqas et al. 2023). By means of the ratio method, the historical data sets can be divided by the 24 h precipitation data, respectively, and the standard duration precipitation data, that is, the decomposition, can be performed by multiplying the calculated rate values with the 24 hour precipitation data of the climate models. Damé et al. (2008) compared methods by decomposing with the Method of Relations and Bartlett–Lewis Model of Modified Rectangular Pulse. The EQM method is an important method for updating the precipitation IDF curves for precipitation under climate change (Srivastav et al. 2014). This method involves spatially shrinking the maximum daily precipitation values taken from global climate models to the daily maximums observed at specific stations, and explicit temporal reduction to reflect changes between the historical period and future periods. By incorporating changes in the dispersion characteristics of climate models, EQM outperforms traditional spatial downscaling methods, accurately capturing past densities and frequencies while also predicting future extreme rainfall (Olorundami 2015). The simplicity, computational efficiency, and ability to project future climate scenarios of the EQM method make it an important contribution to water resources planning and management in the face of evolving extreme conditions (Madjidi et al. 2023). As seen in most studies in the literature, the decomposition of climate models is based on the relationship between the observed data and the model data.

There are many ways to conduct flood risk analyses. Remote sensing, GIS techniques, mathematical applications, machine-learning models, statistical analysis, and evaluation methods are among the most-used applications in flood analysis (Wang et al. 2012; Ogato et al. 2020; Thapa et al. 2020; Xafoulis et al. 2022; He et al. 2023; Jodhani et al. 2023; Bhattacharyya & Hastak 2024; Faisal et al. 2024; Maharjan et al. 2024; Paulik et al. 2024; Wu et al. 2024). Oborie & Rowland (2023) examined the flood situation in the Niger Delta region with the help of GIS by using morphometric parameters such as drainage density, flow frequency, texture ratio, form factor, solidity index, relief rate and infiltration number. Flood risk analysis using the analytical hierarchy process (AHP) is an important method used in various studies. AHP, combined with GIS and remote sensing, helps identify flood-prone areas (Ahmed et al. 2023; Baalousha et al. 2023). It allows the integration of multiple factors such as land cover, soil type, rainfall, elevation and runoff accumulation to assess the flood risk (Cikmaz et al. 2023). AHP enables the creation of flood hazard zoning maps by helping to determine the relative weight of each parameter (Skonieczna & Walczykiewicz 2023). In addition, the AHP method is used to rank the importance of exposure criteria and improve risk assessment in urban areas that are vulnerable to severe flooding (Mourato et al. 2023). The AHP decision-making process, which includes expert opinions, is also effective in creating flood-sensitivity indices for the assessment of flood risk in different regions.

In this study, a flood risk map of Diyarbakır province was created by working in seven different layers (slope, aspect, soil classes, land use classes, precipitation, geology, distance to the river) with the AHP in the GIS environment. The flood risk map was created with historical data. Then, future precipitation data created by separation by the equidistance quantile matching method and future interpretations were made on the flood risk map obtained with AHP. In line with its purpose, the study tried to evaluate the flood risk situation of the study area from different perspectives. According to the findings obtained, it was observed that precipitation that would cause flooding may occur according to the HadGEM-ES model, and that there will be a decrease in precipitation and situations that may cause flooding will not occur according to the MPI-ESM-MR and GFDL-ESM2M models. In the current flood risk map, it is seen that the HadGEM-ES model is in harmony with these regions, especially since the Bismil and Çınar regions are located in high flood risk areas and floods have occurred in these regions both in the past and in the recent period.

In the study, precipitation data of the RCP4.5, RCP8.5 scenarios of the HadGEM-ES, MPI-ESM-MR, and GFDL-ESM2M models were used in the equidistance quantile matching method (Figure 2). The time range of the data covers the years 1971–2000 for the historical period and 2023–2098 for the future period. The reason for choosing these time intervals is that they are within the range of the data in the data pool of TSMS climate models. All data in the data pool were used. All precipitation data were obtained from the General Directorate of Meteorology of Turkey. HadGEM-ES is a climate model developed by the UK Meteorological Office. This model simulates a complex climate system that includes the atmosphere, ocean, and glaciers, as well as vegetation, the carbon cycle, and other important components. HadGEM-ES is used to analyze climate change scenarios and examine the possible effects of global warming. MPI-ESM-MR is a climate model developed by the Max Planck Institute for Climate Research. This model performs a comprehensive simulation of the climate system, which includes components such as the atmosphere, ocean, glaciers, and the carbon cycle. MPI-ESM-MR is used to investigate climate change impacts, assess the accuracy of climate models, and inform policy-makers. GFDL-ESM-2M is a climate model developed by the Climate Dynamics Laboratory of Princeton University in the United States. This model simulates a complex climate system that includes components such as the atmosphere, ocean, glaciers, vegetation, and the carbon cycle. GFDL-ESM-2M is a powerful tool used to investigate climate-change scenarios, develop climate forecasts and shape climate policies. RCP4.5 and RCP8.5, on the other hand, are two different scenarios that express how climate scenarios may change depending on future greenhouse-gas emissions. RCP stands for ‘Future Climate Change Project’ (Representative Concentration Pathway). RCP4.5 is a scenario in which more limited greenhouse-gas emissions occur. In this scenario, it is aimed to keep global warming at a limited level. RCP8.5, on the other hand, is a scenario in which high greenhouse-gas emissions continue, and under this scenario, more serious effects of global warming are expected. These climate models and scenarios help scientists examine future scenarios on climate change and make scientifically based decisions in the creation of climate policies. These models are used as important tools to assess climate-change impacts, understand possible risks, and develop strategies to combat climate change. One of the biggest reasons for the selection of the climate models used in the study is that the models used in the study are the most suitable models for Turkey's climate dynamics among the global climate models, as a result of the analyses and studies conducted by TSMS (TSMS 2023).

The process steps of the EQM method, which is the first method used in the study, are shown as follows.

• First, annual maximum precipitation data are obtained from the historical data set of global models, the future data set and the data of the observed precipitation data sets of Diyarbakır province.

• In the second stage, the distribution parameters that best match the Gumbel distribution are calculated for each data set. These parameters are the characteristics that define the statistical distribution of the data set.

• A statistical relationship is established between the daily maximum data of the global climate models and the observed standard duration maximum data series. This relationship is achieved using the principle of quantile matching. A statistical relationship is established by equating the cumulative probability distribution of the global climate model with the cumulative probability distribution of the standard duration series (Equations (1) and (2)):

  

  (1)

  

  (2)

In these equations, indicates the max data of the observed data, indicates the duration represents the max data of the historical data set of the global climate models, and and refer to the coefficients of the linear equation.

• A statistical relationship based on a similar quantile matching is established between the maximum data of the historical data sets of the global climate models and the cumulative probability distributions of the future maximum precipitation data of the global climate models (Equations (3) and (4)):

  

  (3)

  

  (4)

In these equations, indicates the max data from the historical data set of the global models, represents the 24 h maximum precipitation data of global climate models, and represent coefficients of the linear equation.

• In the next step, Equations (3) and (4) are combined to obtain the following equation (Equation (5)):

  

  (5)

In this equation, represents disaggregated standard duration global-model data and represents the global model 24 h max precipitation data. Other coefficients are specified in Equations (9) and (10).

In the last step, the decomposed data obtained from Equation (5) and the future IDF curves are drawn according to the Gumbel distribution (Srivastav et al. 2014).

The second method used in the study was the analytical hierarchy method. The analytical hierarchy method is a method used to analyze and solve multi-criteria decision-making problems. This method parses the different criteria in a decision-making process and evaluates the relationships between these criteria. The main purpose of this method is to systematically structure the decision-making process and to provide the decision-maker with the opportunity to make an objective evaluation. The analytic hierarchy method is a powerful tool that can be used in complex and ambiguous decision-making situations. The implementation of this method proceeds through several steps. In the first step, the decision-making problem is identified and the criteria and sub-criteria that make up the problem are determined. Next, the relationships between these criteria in the decision-making process are determined. Relationships are weighted according to the priorities of the decision-maker. For this weighting process, the decision-maker is often asked to indicate their relative priorities. For example, the decision-maker is helped by asking questions such as ‘How important is Criterion A compared with Criterion B?’ In this way, the relative importance of the criteria to each other is determined. Next, the weighted values of the criteria are calculated. This calculation is usually done by mathematical methods, and analytic hierarchy matrices are used. Matrices show the relationships of criteria with each other and are used in weighting operations. Finally, the decision-making problem is solved using the weighted values of the criteria. These values are used to evaluate different alternatives and help determine the most appropriate option. The analytical hierarchy method is thus an approach that facilitates the analysis and solution of complex decision-making problems. This method is used and helps to achieve useful results in many areas such as business, engineering, health and strategic planning.

Table 1 shows the degrees used in decision matrices and which degree corresponds to what as an expression.

The process steps of the AHP method are as follows.

• 1. The problem definition and content are determined exactly. The order of precedence is defined among the parameters to be used.

• 2. A matrix of comparisons is created in pairs. Grade coefficients in the ranges 1–9 and 1/2–1/9 are assigned to parameters according to the degree of importance (Table 1). Matrices are formed as a square matrix with diagonals of 1 (Equation (6)):

  

  (6)

In this equation, is the value of the comparison of criterion I and criterion J relative to each other, and corresponds to the value of 1/⁠.

• 3. In this step, each element in the matrix is normalized by dividing it by its column total. The sum of each normalized column is 1 (Equation (7)):

  

  (7)
• 4. The row sum of each normalized matrix is averaged by dividing it by the size of the matrix. The values found indicate the importance weight of each parameter. These weight values form the priority vector:

  

  (8)

Using Equation (8), percentage distributions are obtained that show the importance of the parameters relative to each other.

• 5. It is necessary to determine the consistency ratio of the created comparison matrix at this stage. In this context, it is necessary to calculate the coefficient called the Consistency Index (CI) (Equations (9) and (10)):

  

  (9)

  

  (10)

  

  (11)

  

  (12)

  

  (13)

If the value obtained from Equation (14) is less than 0.10, the comparison matrix is judged to be appropriate.

• 6. To prioritize decision options, a matrix is created by making comparisons between criteria. This matrix can also be defined as the weight vector for each criterion.

• 7. When sorting decision options, priority vectors are used. By combining the calculated priority vectors for the criteria, a matrix of all priorities is obtained. Then, the matrix of all priorities and the priority vectors of the decision options are multiplied and summed to obtain a result vector. The decision option with the highest weight in this vector is determined as the decision option that should be preferred for the solution of the problem (Faisal et al. 2024).

When the data in Figure 5 are examined, situations similar to the situation in the RCP4.5 scenario are seen. Compared with the other models, the HadGEM model shows an increase in continuous precipitation except for the 5-min duration in a two-year period. MPI and GFDL models, on the other hand, show a decrease in all durations of continuous precipitation. Relative to the RCP4.5 scenario, precipitation reductions in the MPI and GFDL models are further reduced in the RCP8.5 scenario.

As can be seen in Table 4, the matrix was chosen to show how areas at distances closer to the stream would be at higher flood-risk. The matrix is formed in such a way that the risk of flooding decreases as you move away from the stream. Since the consistency rate is less than 10%, the matrix is accepted as in Table 4.

In Table 5, ‘1’ was selected as urban areas and commercial units, ‘2’ as non-irrigated arable areas, ‘3’ as continuously irrigated areas, ‘4’ as pasture areas, ‘5’ as forest areas, ‘6’ as bare cliffs, and ‘7’ as water areas.

This study aims to evaluate the flood risk in different ways of analysis, to look at the area to be studied from different perspectives, and to predict the future flooding situation under the influence of climate change. In order to make future precipitation forecasts under climate change, precipitation forecasts were made with the RCP4.5 and RCP8.5 scenarios of the HadGEM-ES, GFDL-ESM2M, MPI-ESM-MR models.

Looking at the future precipitation data obtained through the equidistance quantile matching method, it was seen that the HadGEM-ES model increased precipitation in almost all of the rotation periods of the RCP4.5 and RCP8.5 scenarios, while the MPI-ESM-MR and GFDL-ESM2M models may decrease in precipitation. In the RCP8.5 scenario, it is seen that the increase and decrease in precipitation is higher than in the RCP4.5 scenario.

When the flood risk map obtained with AHP is evaluated together with the data obtained with the EQM method and the previously observed and recorded flood events are taken into consideration, it is seen that this model may be more consistent and accurate in predicting for future periods, since the HadGEM-ES model for Diyarbakır province shows that precipitation may increase at a level that may cause floods.

According to the data obtained, almost half of the study area is in the risky class. Since the flood points that have survived to the present day occur in the high-risk and very high-risk classes on the map, attention should be paid to these areas and necessary precautions should be taken for flood events that may occur in the future.

It has also been observed that the equidistance quantile matching method, which is used to create the daily maximum precipitation data of climate models and IDF curves and to decompose them into standard duration precipitation data, is very successful in decomposition, and is more useful and practical than many of the other separation methods available in the literature.

It has been observed that it may be appropriate to adjust the IDF curves for the future, especially since they are made with historical data in public institutions and organizations and do not contain the effect of climate change for the future, and according to the HadGEM-ES model, they predict significant precipitation increases for future periods, and the recent heavy rainfall and flood conditions support this situation.

The flood risk map created with the analytical hierarchy model provided the opportunity to present the flood risk situation visually more clearly for the studied area. Since AHP contains both objectivity and subjectivity, it has been seen that it is a method that allows it to contain expert opinion and to intervene in comparison matrices and maps according to the conditions. In addition, it offers the opportunity to analyze very large areas through the AHP method, as in this study.

The data used for AHP was selected at a resolution of 25 m due to the fact that it is more comfortable to work with given the hardware and software used and the working area is large. If the hardware and software are better, similar studies and analyzes will be more detailed and efficient due to the clarity of the resolution. The layers selected for AHP may differ depending on the workspaces. Seven data sets have been selected for the current area, and different and more detailed studies can be done by adding more layers in different studies. The weight coefficients and importance numbers in the comparison matrix used may differ according to the field of study and expert opinion.

The global climate models used for EQM will be able to ensure that the data are more relevant and consistent with the creation of newer and improved climate models. In addition, similar studies can be carried out with different separation methods. In the decomposition method, it is seen that decomposition methods that can easily calculate large data groups are more suitable in analysis due to the fact that climate models have very large data sets.

Since the accurate and reliable calculation of flood risk analyses is of great importance for the safety of life and property in the study areas, it is of great importance that the analyses in the study area are carried out with different analyses as in this study and that they are made with future-oriented analyses other than historical data, taking into account the effect of climate change.

According to the results obtained from the engineer's point of view, and considering the most dangerous situation obtained from the engineer's point of view, especially according to the data obtained from the HadGEM-ES model, due to the possibility of precipitation that may increase the flood risk, the existing designs of infrastructure and water structures, especially in very high risk and high-risk areas, should be updated in the flood risk map created as a result of AHP analyses and new structures should be built according to these analyses. It is thought that this will help public institutions in the region.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

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

The authors declare there is no conflict.