1 Introduction

Floodplains are fundamental parts of natural river systems (Funk et al. 2019; Knox et al. 2022; Meli et al. 2014) and provide vital ecosystem services (Costanza et al. 1997; Petsch et al. 2023; Wantzen et al. 2016). When a floodplain is degraded, interrupted, or impaired, valuable and crucial ecosystems that are home to a large number of both aquatic and terrestrial species are lost (Costanza et al. 1997). Floodplains are, therefore, often referred to as biodiversity hotspots (Schindler et al. 2016). At multiple scales, biochemical processes in floodplains are dynamically intertwined and controlled by multidimensional interactions of flow, nutrients, and sediments (Gordon et al. 2020; Robinson et al. 2002; Schindler et al. 2014; Villa and Bernal 2018). Moreover, the public has long since discovered floodplains for recreational purposes and has begun to acknowledge their cultural value (Petsch et al. 2023; Stammel et al. 2021).

Intact floodplains rely on hydraulic connectivity and overbank flows due to flood events. Despite the essential importance of floodplains for natural flood retention and flood risk mitigation (Opperman et al. 2009; Schober et al. 2020; Serra-Llobet et al. 2022) floodplains have been impaired since the beginning of civilization, resulting in the loss or degradation of about 90% of former floodplains in North America and Europe. A new global data set reveals a floodplain loss of ~ 600,000 km2 between 1992 and 2019 (Rajib et al. 2023). At our study site, the Danube, a recent study (Eder et al. 2022a, b) and earlier reports (Hein et al. 2016; Schwarz 2010) confirmed a loss of 79% of former floodplains.

The consequences of the steady decrease in floodplains are evident on several levels. From an ecological point of view, the loss of floodplains or their isolation from the river results in the disappearance or severe degradation of the habitats for flora and fauna that live in them (Di Baldassarre et al. 2013, 2018; Opperman et al. 2009). The decline in floodplains, however, also leads to a significant increase in flood peaks and an acceleration of flood runoff (Habersack et al. 2015; Helton et al. 2014; Kundzewicz and Menzel 2005). In the Danube River, flood wave travel time has been significantly reduced compared to the nineteenth century (Blöschl et al. 2013).

The loss of floodplains is discussed as a key reason for the increase in extreme flood events on rivers. Due to changing climatic conditions, conventional flood protection, such as structural measures, could fail more frequently in the future, putting the area behind the protective structures, which is characterized by a high level of development and thus vulnerability, at risk (Di Baldassarre et al. 2018; Jüpner 2018; Ludy and Kondolf 2012; Schober et al. 2013).

Despite efforts to promote floodplain preservation and restoration, many floodplains are under pressure due to economic development (Steffen et al. 2007; Vörösmarty et al. 2010) and remain ecologically impaired or disconnected. The prevailing economic pressure, societal changes, opposed interest, and human-related impacts on our ecosystem pose a great challenge to floodplain preservation, restoration, and sustainable river management (Auerswald et al. 2019; Hein et al. 2016; Klijn et al. 2018).

In order to overcome barriers, facilitate conservation progress, and find win–win solutions among stakeholders, the floodplain evaluation matrix (FEM) is used as an approach to systematically quantify the influence of floodplains on flood risk reduction, ecology, and socio-economics. Floodplain evaluation is an integral part of flood risk analysis and aims to understand the characteristics, features, and dynamics of a floodplain by studying the topography, hydrology, geomorphology, vegetation, land use, and other factors. The field of flood risk analysis and assessment has seen a substantial increase in publications over the past decade, indicating the ongoing need for further research (Díez-Herrero and Garrote 2020). Numerous studies (Cammerer and Thieken 2013; De Bruijn et al. 2008; Eder et al. 2022a, b; Elmer et al. 2012; Hall et al. 2005; Klijn et al. 2015; Kundzewicz et al. 2014; Linde et al. 2011; Winsemius et al. 2016) systematically address future developments, including population growth, increasing urbanization, and the potential impacts of climate change on flood discharge, in order to evaluate future flood risks. Floodplain evaluation studies (Abell et al. 2023; Karpack et al. 2020; Lininger and Polvi 2020; Rohde et al. 2006; Schober et al. 2020) often deal with human modifications to floodplains, their effects, and floodplain restoration. Following the objective that restoration should be attempted at the watershed scale (Bernhardt and Palmer 2011; Wohl et al. 2005), we aim to demonstrate the advantages of a comprehensive transdisciplinary floodplain assessment capable of delivering basin-wide water management information, resulting in recommendations for floodplain preservation and restoration. The evaluation of floodplains with a set of parameters can show the importance of the floodplains, leading to more preservation and restoration efforts. We further developed the FEM to address the outlined research questions at any river system:

  • Why are active floodplains important from a perspective of flood protection, ecology, and socio-economics?

  • How is the value of active floodplains distributed within a large river system, and what factors influence this distribution?

  • Is the Floodplain Evaluation Matrix suitable and applicable to determine the importance of floodplains for an entire large river?

  • Can the Floodplain Evaluation Matrix provide a basis for future floodplain management and promote floodplain preservation and restoration?

The method was first applied to the Danube River, known as the most international river worldwide. The application was done within the EU-funded “Danube Floodplain” project (https://www.interreg-danube.eu/approved-projects/danube-floodplain#! accessed on 12 November 2024) that aimed at enhancing international water management and flood risk prevention, focusing on floodplain management. A consortium of representatives from national water agencies, universities, research institutes and NGOs along the Danube River worked together in this project to develop and apply the methods presented in this paper.

2 Material and methods

2.1 Study area—Danube River

The Danube River Basin has an area of 800.000 km2 and forms the second largest catchment in Europe. Located in the centre and the southeast of Europe, it includes territories of 19 riparian countries. It is regarded as the most international river basin in the world. A population of about 80 million people lives along the Danube and its tributaries (ICPDR 2011, 2015). The Danube River originates in the Black Forest Mountains in Germany and discharges into the Black Sea, covering a distance of 2857 km. The mean annual discharge of the Danube at the mouth is 6486 m3s−1 (Sommerwerk et al. 2009). The basin is divided into three sub-regions (Upper, Middle, and Lower Danube) corresponding to a characteristic gradient and landscape features from source to delta. Rapid currents occur in the upper catchment, and alpine tributaries dominate the flow and sediment characteristics. In the middle and lower sub-regions, the Danube flows predominantly in a lowland landscape. The transport capacity has dropped after passing the Hungarian Gates Gorge, creating a series of riparian plains and islands. Downstream the confluence of the tributaries Drava, Tisza, and Sava it has gained size, and the flow rate has almost tripled. About 80 km upstream of the Black Sea, the Danube River splits up into three branches, forming a delta of about 6500 km2 (ICPDR 2011).

2.2 Floodplain Evaluation Matrix (FEM)

The Floodplain Evaluation Matrix (FEM), developed by Habersack et al. (2015), is a holistic floodplain evaluation method that uses hydrological, hydraulic, ecological and socio-economy parameters to assess the multiple functions of a floodplain. The method’s concept and framework were presented in Habersack et al. (2015). Schober et al. (2013) and Habersack & Schober (2020) applied the method to rivers reaches in Austria (Danube, Raab, and Inn) to evaluate the role of floodplains in reducing flood hazards using hydrological and hydraulic parameters. The FEM has a flexible design and can be adapted to research questions and project needs. Eder et al. (2022a, b) expanded the method for a future-oriented flood risk management and considered land use developments and climate-related changes in flood discharges.

We further developed and adapted the FEM in this study to apply the method to an entire large river for the first time, leading to a comprehensive and systematic evaluation of floodplains. Since the Danube River is an international river, we had to adapt the method to the data availability in each country and find parameters that were accepted and could be applied by the representatives of the different states. The adaptations were necessary to develop a schematic approach of the FEM that allows an application at any river according to the data availability and research objectives. We also introduced the novel concepts of “need for preservation” and “restoration demand”, offering recommendations for floodplain preservation and restoration through commonly agreed parameters and thresholds established in collaboration with representatives from national water agencies, universities, research institutes, and NGOs from ten countries. The developed method consists of four consecutive steps, each of which is described in the following paragraphs. Figure 1 illustrates the developed approach in a flow chart.

Fig. 1
figure 1

Flow chart of the floodplain evaluation matrix (FEM). In step I., different data sets are collected and analyzed. The next two main steps (II. and III.) represent essential tasks leading to the first two FEM outputs: delineated floodplains and selected FEM parameters. In step IV, the first three tasks (1., 2., 3.) are mandatory, while all other tasks are optional, dependent on research questions or goals. In this study, all options were applied and led to results/outputs (1., 2., 3a., 3b., 3c., 4). The available data and selected tasks for this study are highlighted by the dotted rectangle, while optional data and tasks are presented in italics

Full size image

2.2.1 Step I: data acquisition and analysis

The first step is a comprehensive acquisition and analysis of diverse data sets. The data basis is individual for each river. As subsequent steps rely on the availability and quality of the data, this step is crucial.

Hydrological and hydraulic data such as observed flood waves, measured water levels, and hydraulic models are necessary to calculate the mandatory parameter of the FEM (relative peak reduction, flood wave translation, and water level change). Furthermore, depending on the research objectives, additional data sets such as geospatial, ecological, and other relevant data may be gathered during this stage.

2.2.2 Step II: delineating floodplains

The second step is a task to delineate floodplains based on available data and the chosen approach. Various methods exist in the literature for identifying active floodplains, including the consideration of inundation areas from flood events with specific recurrence intervals, examination of recent alluvial deposits, or utilizing ecological approaches to identify areas colonized by organisms adapted to flooding (Nanson and Croke 1992; Schwarz 2010; Tockner and Stanford 2002). The FEM provides the user the flexibility to select from these diverse methods found in the literature or use delineated floodplains from previous studies. The key requirement is to generate floodplains with clearly defined starting and ending points as the output of the delineation process.

2.2.3 Step III: selecting FEM parameters

The selection of FEM parameters depends on data availability and research goals. The mandatory FEM parameters are the relative peak reduction, flood wave translation, and water level change. This step provides the flexibility to include any additional parameter deemed necessary or valuable for addressing specific research questions. The chosen parameters are the output of this step. In Sects. 2.3.1–2.3.4, the mandatory FEM parameters are described along with additional ones.

2.2.4 Step IV: FEM application

In this step, hydrological and hydraulic scenarios are defined, the selected parameters are calculated, and optional tasks are performed based on the set research goals.

The first task is to define hydrological and hydraulic scenarios. It is mandatory to use a HQ100 at each floodplain, derived from observed data. Furthermore, it is advisable to include both steep and flat flood waves to assess the floodplain's sensitivity to varying hydrological conditions. Other hydrological scenarios can include flood waves with shorter or longer recurrence intervals, such as HQ30 and HQ1000, or consider the impact of increased flood peaks due to climate change. All other hydrological scenarios, such as mean or bankfull discharge, can be defined in this step as well if they are necessary for the selected parameters.

When considering hydraulic scenarios, two are mandatory: simulating the entire existing floodplain with the current dyke heights and a pure river channel model, in which the floodplains are disconnected by dykes with unlimited height. Additionally, further scenarios may involve gradual reductions of the floodplain and adjustments in dyke heights, ranging from 25 to 75%.

The next task is to calculate the selected parameter by performing the necessary numerical simulations using the defined scenarios, the available models, and the chosen approaches. Afterwards, thresholds for different performance classes can be established to evaluate the performance of a floodplain for a certain parameter. This can involve defining various performance classes and implementing a point system, such as high (5 points), medium (3 points), or low (1 point). These performance classes can then be used to assess the "need for preservation" and determine the "restoration demand." For instance, one approach could be to decide whether a floodplain should be preserved based on whether it achieves a high-performance rating (5 points) for at least one parameter in the evaluation, indicating a "need for preservation." Following this, the "restoration demand" can be determined by categorizing floodplains into low, medium, or high restoration demand groups based on the total points received during the application of the FEM.

The last optional task consists of analyzing and comparing the results of floodplains in different river reaches or the entire basin to get an overview of the importance of the floodplains in different reaches.

The following outputs can be obtained if all optional tasks are applied:

  1. 1.

    Evaluated floodplains.

  2. 2.

    Performance classes for floodplains.

  3. 3.

    Ranked floodplains.

    1. a.

      Based on points.

    2. b.

      Floodplains that need to be preserved

    3. c.

      Restoration demand of floodplains.

  4. 4.

    Basin-wide analysis of floodplains

2.3 Floodplain Evaluation Matrix (FEM) parameters

2.3.1 Hydrological parameters

Hydraulic models are used to determine hydrological parameters by computing the flood wave transformation between in- and output hydrograph for each floodplain. Since the design discharge for flood protection measures mostly corresponds to a 100-year return period (HQ100—medium frequency), it is necessary in the FEM to calculate the retention effects of each floodplain for such an event. Therefore, all input hydrographs had to have a peak discharge at the beginning of the floodplain that corresponded to a HQ100. In general, the hydraulic models should be split into smaller models to represent a single floodplain with an in- and outflow section. This ensures that the peak of the inflow hydrograph corresponds to the HQ100 discharge and that only the effects of that floodplain are investigated.

2.3.1.1 Relative flood peak reduction—ΔQrel

This parameter considers the effect of a floodplain on the peak of a flood wave. The difference between the peak of an in- and output hydrograph at the beginning, respectively at the end of a floodplain, results in the peak reduction ΔQtot (m3 s−1) for the investigated floodplain. However, it is necessary to calculate the retention effect of the river channel to show only the effect of the floodplain on peak reduction. For calculating the peak reduction ΔQRC (m3 s−1) of the river channel, “river channel” models, where the floodplains are disconnected from the river, are used. The floodplains can be disabled, or hypothetical dykes could be implemented that cannot be overtopped to simulate the loss of the floodplain areas. The same input hydrograph is used as for the calculation of ΔQtot. In Fig. 2, the in- and output hydrographs for the river channel model (ΔQRC, ΔtRC) and the hydraulically active floodplain (ΔQtot, Δttot) are visible. It is shown that the retention effect of the floodplain is significant. In the absence of inundation areas, the peak reduction for the entire river reach would be close to zero. The flood wave translation would be reduced as well.

Fig. 2
figure 2

The input hydrograph is the same for the river channel model and the current case where the floodplain is active. The blue-dotted line shows the output hydrograph for the river channel model. The black-dotted line represents the output hydrograph for the model where the floodplain is connected to the river. ΔQtot—flood peak reduction total (m3s−1); ΔQRC—flood peak reduction river channel model (m3s−1); Δttot—flood wave translation total (h); ΔtRC- flood wave translation river channel model (h)

Full size image

ΔQRC has to be subtracted from ΔQtot to demonstrate only the effect of the floodplain on the peak reduction ΔQ:

$$\Delta {\text{Q}} = \Delta {\text{Q}}_{{{\text{tot}}}} {-}\Delta {\text{Q}}_{{{\text{RC}}}} \left( {{\text{m}}^{3} \;{\text{s}}^{{ - {1}}} } \right)$$
(1)

The relative peak reduction ΔQrel (%) is computed to allow a comparison of the peak reduction effect in different river reaches (Eq. 2). Qmax is the highest discharge value of the input flood wave, and Qbankfull is the discharge where the river starts overtopping its bank. To calculate the relative peak reduction ΔQrel, Qbankfull is subtracted from Qmax because the floodplain is not active until Qbankfull is reached.

$$\Delta {\text{Q}}_{{{\text{rel}}}} = \Delta {\text{Q}}/\left( {{\text{Q}}_{{{\text{max}}}} {-}\Delta {\text{Q}}_{{{\text{bankfull}}}} } \right) \, \times {1}00\left( \% \right)$$
(2)
2.3.1.2 Flood wave translation—Δt

The flood wave translation is the second parameter required to investigate the process of wave attenuation and shows the deceleration of the flood wave through the floodplain. It is determined in a similar way as the peak reduction, by calculating the time difference Δt (h) between the occurrence of the out-/input hydrograph peak (see Fig. 2). The same hydrographs are used as for calculating the peak reduction. For demonstrating only the flood wave translation due to the floodplain, the following equation is used:

$$\Delta {\text{t }} = \Delta {\text{t}}_{{{\text{tot}}}} {-}\Delta {\text{t}}_{{{\text{RC}}}} \left( {\text{h}} \right)$$
(3)

2.3.2 Hydraulic parameter

2.3.2.1 Water level change—Δh

The hydraulic models were also used to illustrate the effects of a total loss of a floodplain on the water levels. The “river channel”, is a case where the river is fully embanked and completely disconnected from the floodplain used to determine the water levels without the effects of the floodplain (hRC) for a HQ100. The same hydraulic models and input hydrographs that were employed to calculate ΔQtot and Δttot are used to determine the water levels with the connected floodplain (htot). The water levels htot and hRC are observed at a defined cross-section in the middle of the river channel, and the mean value is determined to calculate the water level change Δh (cm) by subtracting hRC from htot (see Eq. 4). This yields the water level increase due to the floodplain loss.

$$\Delta {\text{h}} = \Delta {\text{h}}_{{{\text{tot}}}} {-}\Delta {\text{h}}_{{{\text{RC}}}} \left( {{\text{cm}}} \right)$$
(4)

2.3.3 Ecological parameters

2.3.3.1 Existence of protected species

This parameter is used to assess the number of protected plant and animal species in a floodplain according to different sources such as the Natura 2000, Emerald Network, national databases, or other studies.

2.3.3.2 Connectivity of floodplain water bodies

Longitudinal, lateral, and vertical connectivity is crucial for the functionality of riverine ecosystems. However, given the complexity of the assessment in all three directions and the project goals, a simplified method was chosen to apply this parameter to all floodplains along the Danube. Therefore, only lateral connectivity, which refers to the connection of the river channel and the floodplain, was investigated. The parameter was determined considering three scenarios:

  1. 1.

    Mean water level

  2. 2.

    Bankfull flow

  3. 3.

    Above bankfull flow

The connectivity is determined using the same hydraulic models as for calculating the hydrological parameters. Only the input hydrographs have to be changed according to the investigated scenario (mean water level, bankfull, above bankfull). The inundation areas of each scenario are used to determine the lateral connectivity of water bodies (e.g., side branches, oxbows) in the floodplain. First, the discharge has to be found where the floodplain water bodies are connected (flooded). The next step is to determine the “natural (historical)” status of the water bodies on the floodplains. For this, historical river maps prior to major river regulation activities or written historical documents are used. A possible source of historical maps is the website (https://static-cdn.mapire.eu/en/ accessed on 20 January 2024) from the company Arcanum. There are four possible outcomes when comparing the current and historical status:

  1. 1.

    No “natural” (historical) water bodies on the floodplain.

  2. 2.

    Existing water bodies on the active floodplain in the current and historical condition.

  3. 3.

    On the historical “natural” (historical) water bodies exist, but in the current floodplain, no water bodies are left due to human activity (e.g., dykes, etc.).

  4. 4.

    On historical maps, “natural” (historical) water bodies exist that are also still present on the current floodplain but have been cut off by human structures.

The river type must also be considered to evaluate the lateral connectivity. For example, the connectivity naturally starts at bankfull discharge in case of a meandering river system. If this is the case, the floodplain receives a high performance (5 points) in the FEM, and no further steps are needed. Historically braided or anastomosing river types receive a high performance when the side channels are already connected below mean discharge. The detailed scenarios for the FEM are listed below:

  1. 1.

    Water bodies connected up to mean water level/No “natural” (historical) water bodies on the floodplain/meandering river systems connected above bankfull discharge (high performance—5 points).

  2. 2.

    Water bodies connected at mean water level up to bankfull discharge (medium performance—3 points).

  3. 3.

    Water bodies not connected above bankfull discharge/On the historical maps “natural” (historical) water bodies exist, but at the current floodplain no water bodies are left (low performance—1 point).

If a human structure cuts off parts of the existing floodplain water bodies (e.g., a dyke), the determined performance is downgraded to the next class because we recognize this as negative for the connectivity. For example, floodplain water bodies are connected up to mean flow, leading to high performance in the FEM. However, by checking the historical maps or a DEM, it was discovered that the existing water bodies have been cut off, resulting in a downgrade to the medium performance class.

2.3.4 Socio-economical parameters

2.3.4.1 Potentially affected buildings

This parameter determines the number of buildings on each floodplain and divides it by the total area of the floodplain to allow a comparison between them. Aerial photographs, high-resolution satellite imagery, or cadastral information can be used to determine the number. Cadastral information is more accurate than satellite-based detection, but their public access is often difficult or non-existent. If access to these datasets is available, they should be preferred since they are more accurate. The determination of the number of buildings on each floodplain is carried out with the help of GIS software such as ArcGIS or QGIS.

2.3.4.2 Land use

Land use adapted to potential flooding minimizes the socio-economical vulnerability of a floodplain. Hence, less vulnerable land uses on the floodplain such as forests or natural grasslands receive high performance, while highly vulnerable land uses such as settlements get a low performance. The CORINE land cover data set (https://land.copernicus.eu/pan-european/corine-land-cover/clc2018 accessed on 20 January 2024) from the Copernicus database summarizes land cover into five categories: “Artificial surfaces”, “Agricultural areas”, “Forest and semi- natural areas”, “Wetlands”, and “Water bodies”, and can be used to analyse the current land use of each floodplain in a GIS. Each land use class was assigned to one of three performances (low, medium, high) based on the vulnerability against flooding (see Table 6—Appendix). For example, land use urban fabric units or industrial units are highly vulnerable against flooding, resulting in low performance (1 point) in the FEM. The different land uses are combined proportional to their areas to one evaluation value for the whole floodplain by multiplying all highly vulnerable areas by 1, the area of medium vulnerability thereby 3 and the low vulnerability areas with 5. The respective results are summed up and then divided by the sum of the total floodplain area, resulting in a weighted FEM-value for the floodplain.

2.4 Thresholds for FEM parameters

Generally, the thresholds can be defined individually for each river, taking into account specific characteristics of the river and its floodplains (e.g., basin size, river type). The thresholds for the hydrological and hydraulic parameters can be defined based on the previous FEM studies (Habersack et al. 2008; Schober et al. 2013, 2020) and may incorporate findings from studies on preparation time for population and emergency forces (Friso et al. 2008), as well as on damage functions based on water levels (de Moel AND Aerts 2011). All thresholds should be discussed and defined within an expert consortium that can consists of local and regional experts who possess comprehensive knowledge about the river's characteristics, hydrology, geomorphology, ecology, and socio-economic dynamics. By bringing together this multidisciplinary expertise, the consortium can collaboratively define thresholds and develop strategies tailored to the specific needs and challenges of the river under investigation.

2.5 Recommendations for preservation and restoration

Providing recommendations for floodplain preservation and restoration based on a multi-criteria approach is desirable for effective floodplain management. Giving preservation recommendations derived from such methodology could assist stakeholders and decision-makers in prioritizing floodplain preservation efforts. Assessing a floodplain's restoration demand offers insights into its current status and the extent of human interventions. Typically, higher restoration demand indicates increased human development on the floodplain and greater deviation from its natural state.

3 FEM application at the Danube River

This section presents the application of the FEM at the Danube River, beginning with an overview of each step outlined in 2.2.1–2.2.4. Subsequently, it provides a basin-wide overview of the calculated FEM parameters and their performance allowing a comparison across three distinct river sections. Finally, the number of floodplains that should be preserved based on our results and the distribution of restoration demand along the Danube River are presented.

3.1 Step I: data acquisition and analysis

Following the research goals, hydrological, hydraulic, ecological, geospatial, and other data were collected and analyzed. An overview and a more detailed description of all used data is presented in Table 4 (see Appendix).

3.2 Step II: Delineating floodplains

Numerous approaches have been documented in literature to delineate active floodplains. These include analyzing inundation areas corresponding to flood events with defined recurrence intervals, studying recent alluvial deposits, or employing ecological methods to identify zones inhabited by organisms adapted to flood conditions. We used the 50 active floodplains (> 500 ha) that Eder et al. (2022b) identified in their study based on hydraulic data. We also used their naming convention for floodplains, including country code (DE = Germany, AT = Austria, SK = Slovakia, HU = Hungary, HR = Croatia, RS = Serbia, BG = Bulgaria, RO = Romania), followed by the river code (DU = Danube River), and the floodplain type (AFP = active floodplain). The fourth and last part is the sequential number within the country in flow direction of the river. If a floodplain crosses international borders, each nation is listed in the code, starting with the one upstream.

3.3 Step III: selecting FEM parameters

At the Danube River, we started with parameters that were investigated in previous studies on floodplain evaluation (Habersack et al. 2008, 2015; Schober et al. 2013) to have a data basis and comparative values. As described in Sect. 2.2.3, the mandatory FEM parameters are the relative peak reduction, flood wave translation, and water level change. Besides that, each partner of the “Danube Floodplain” project could suggest new parameters for one of the four categories of interest (Hydrology, Hydraulic, Ecology, Socio-Economics) that met our requirements. First, the parameter should be applicable in each country (data availability), and partners from each country should be able to apply the parameter using a description. In the end the project team selected the parameters considering the parameter's application and informative value. Table 1 shows the selected FEM parameters for the application at the Danube River.

Table 1 Overview of the selected FEM parameters. The hydrological and hydraulic parameters are mandatory in the FEM
Full size table

3.4 Step IV: FEM application

First, we defined our hydrological and hydraulic scenarios. Since the applied method requires that the peak discharge at the beginning of the floodplain corresponds to a HQ100, we used the shape of one of the two major flood events in the years 2006 and 2013, depending on which flood event was closest to a HQ100 in the respective floodplain. Table 5 (see Appendix) details which flood event was chosen in each country. In some cases, up- or downscaling of the flood wave was required to reach the exact HQ100 discharge at the beginning of the floodplain. The mean and the bankfull discharges were required to determine the ecological parameter “connectivity of floodplain water bodies”.

We applied the two mandatory hydraulic scenarios by simulating the entire existing floodplain with current dyke heights and the river channel model (i.e., disconnected floodplains by unlimited dyke heights along the river channel).

Calibrated 1D and 2D hydraulic models were used to perform numerical simulations to determine hydrological, hydraulic, and ecological parameters, with the national hydrological agencies providing the necessary discharges and flood waves. In Table 5 (see Appendix), all the models used and the discharges for the parameters are presented for each country. Generally, hydraulic models were divided into smaller segments to represent individual floodplains, each featuring an inlet and outlet section. This approach ensures that the peak of the inflow hydrograph aligns with the HQ100 discharge, enabling focused investigation solely on the effects of that specific floodplain. In our study, all countries except Hungary, used this approach. Due to practical reasons, the Hungarian team selected a different modelling approach and used a continuous hydraulic model to calculate the hydrological and hydraulic parameters for the Hungarian floodplains. Therefore, also a continuous model was used for the river channel with disconnected floodplains, and they calculated the impact on the water level with this model.

We calculated our defined FEM parameters with the described methods from Sect.  2.3.1–2.3.4 and the results are presented in the following Sect. 3.5.

At the Danube River, thresholds were defined for each parameter, categorizing performance levels as high (5 points), medium (3 points), or low (1 point). The thresholds and the point system for the hydrological and hydraulic parameters were defined based on previous FEM studies (Habersack et al. 2008; Schober et al. 2013, 2020) and consider studies on preparation time for population and emergency forces (Friso et al. 2008) as well as on damage functions based on water levels (de Moel and Aerts 2011). All thresholds were discussed and defined in several meetings within the expert consortium of the “Danube Floodplain” project. These thresholds were determined through testing with Danube River data, and variations were applied considering the input of all involved representatives from national water agencies, universities, research institutes and NGOs along the Danube River. This collaborative effort resulted in, for example, two key thresholds for the “protected species” parameter. One for “need for preservation” and another for “restoration demand”, because only one protected species could lead to the need for preserving the floodplain, but clearly, there is a high restoration demand if only one protected species is present. To determine “the need for preservation”, a floodplain is considered low-performing if no protected species are present, medium-performing if 1–20 species are present, and high-performing if more than 20 protected species exist. For the “restoration demand”, floodplain performance is categorized as low if less than 40 protected species are present, medium if the count is between 40 and 100, and high if it exceeds 100. In Table 2, all thresholds for the chosen FEM parameters are presented. For the connectivity parameter, no thresholds are defined because the described methodology for applying this parameter leads directly to one of the three performance classes.

Table 2 Thresholds for all FEM parameters
Full size table

We developed a two-step approach to provide recommendations for the preservation and restoration of floodplains based on the FEM results. First, we assessed the “need for preservation” to determine whether a floodplain should be preserved. Following that, we evaluated the “restoration demand”. The higher the restoration demand, the more human interventions (e.g., land developments) have occurred on the floodplain and the further the floodplains are away from a natural state. A floodplain must be preserved if at least one parameter of the minimum set is rated with a high performance (5 points). After that, the restoration demand is defined. Using the parameters and their ratings, each floodplain is assigned to one of three groups (low, medium, and high restoration demand). The floodplains were assigned to one of these groups using the total number of points they received in the FEM. For the first time, the “restoration demand” of a floodplain was determined using the FEM. The thresholds were established by the expert consortium of the “Danube Floodplain” project. The consortium consisted of stakeholders and decision-makers who regularly met to discuss different suggestions and variations of the thresholds to define the final thresholds for the project and this paper. Establishing the thresholds through an expert consortium ensures that it is based on the collective expertise and input of relevant decision-makers and stakeholders, considering the diverse perspectives and needs of the stakeholders involved. The following thresholds are intended to serve as a starting point for further discussion, as new results may require adjustments to be made to the system.

The sum of the points received can be maximum 35 (seven minimum parameters × 5 points each) and has to be ≥ 27 to result in a low demand for restoration. Floodplains with total points between 26 and 23 are considered to have medium restoration demand. All floodplains with < 23 points show a high demand for restoration (Table 3).

Table 3 Used thresholds for the floodplains to determine their “restoration demand”
Full size table

3.5 Basin-wide analysis

This section presents a basin-wide overview of the calculated FEM parameters and their performance allowing a comparison across the Upper, Middle, and Lower Danube. This allows to identify potential differences between the three sections.

Sixteen active floodplains with an average size of 50 km2 were identified and evaluated using the FEM at the Upper Danube. Ten are located in Germany, two in Baden-Wuerttemberg that were not evaluated in the scope of this project and eight in Bavaria. The remaining six active floodplains are in Austria and at the Austrian-Slovakian section (see Fig. 7—Appendix). Twenty-four active floodplains with an average size of 54 km2 were identified and evaluated using the FEM at the Middle Danube. Five are located in the Hungarian-Slovakian section, eight in Hungary, one in the Croatian-Hungarian section, five in the Serbian-Croatian section, and five in Serbia (see Fig. 8—Appendix). Ten active floodplains with an average size of 78 km2 were identified and evaluated using the FEM at the Lower Danube. Six are located in the transboundary section of Bulgaria and Romania. Four active floodplains are found in Romania (see Fig. 9—Appendix).

The table of the final FEM, covering all sections, is provided in Table 7 (see Appendix), offering an overview of all results at a glance. Since this work builds on the “Danube Floodplain” project, the results and further outputs are accessible to the public through the spatial database “Danube Floodplain GIS” (http://earth.geo.u-szeged.hu/dfgis/ accessed on 20 January 2024).

3.5.1 Relative flood peak reduction

Figure 3 shows the relative flood peak reduction results for all active floodplains along the Danube River. The calculated values range from 0 to 17%. Thirteen floodplains exhibit high performance (ΔQrel > 2%), six medium performance (1–2%), and the remaining 29 floodplains a low performance. 62% (eight) of the floodplains with a high performance and one-third (two) of the floodplains with a medium performance are found in the Upper Danube section. All floodplains at the Lower Danube display a low performance.

Fig. 3
figure 3

Relative flood peak reduction for all active floodplains along the Danube River. Blue color indicates high, green medium and orange low performance in the Floodplain Evaluation Matrix (FEM) based on the defined thresholds. Insert shows the thresholds for the presented parameter

Full size image

3.5.2 Flood wave translation

Due to the flow processes in the floodplains, the flood wave is decelerated in a range of 0.25–41.5 h. Twenty-eight floodplains show a medium (1–5 h), fourteen a high (> 5 h), and six a low (< 1 h) performance for the “flood wave translation” parameter. Six high-performing (> 5 h) floodplains are distributed in both the Upper (around 43% of the floodplains in this section) and Middle Danube (25% of the floodplains in this section). In the Lower Danube only medium (eight floodplains) and high (two floodplains) performances can be found. The median in the upper section is 5 h, in the middle section 2.75 h, and in the lower Section 2 h.

3.5.3 Water level change

Figure 4 shows all water level changes observed in the river channel if the active floodplains were disconnected from the river. In nineteen cases, the water level would rise at least 50 cm (= high performance). Nine each are located in the Upper (64% of the floodplains in this section) and Middle Danube (around 38% of the floodplains in this section). Only one floodplain in the Lower Danube reached the highest performance class. The median in the upper section is 66 cm, in the middle section 29.5 cm, and in the lower section 12 cm.

Fig. 4
figure 4

Water level change for all active floodplains along the Danube River. Blue color indicates high, green medium and orange low performance in the Floodplain Evaluation Matrix (FEM) based on the defined thresholds. Insert shows the thresholds for the parameter shown

Full size image

3.5.4 Connectivity of flood water bodies

From an ecological perspective, the lateral connectivity between the river channel and floodplain is impaired by human interventions in all active floodplains along the Danube River, leading to no high performance. At the Upper Danube, only one floodplain received a medium performance. All the other twelve showed a low performance. The middle section presented a more balanced picture, with ten low and fourteen medium performances. All floodplains at the Lower Danube received a medium rating for the connectivity parameter.

3.5.5 Protected species

For the protected species parameter, two thresholds were defined by the expert consortium (see 3.4). In this subsection, we use the “restoration demand” thresholds to assess the different performances. The number of protected species per floodplain is notably highest at the Lower Danube (see Fig. 5). On the majority of the floodplains at the Lower Danube (nine out of ten), more than 100 protected species (= high performance) can be found, resulting in a median of 162. In the upper and middle sections of the Danube, the median is 58 and 65, respectively, with three and four floodplains in the highest performance class.

Fig. 5
figure 5

Existence of protected plant and animal species at all active floodplains along the Danube River. Blue color indicates high, green medium and orange low performance in the floodplain evaluation matrix (FEM) based on the defined thresholds. Insert shows the thresholds for the “restoration demand “

Full size image

3.5.6 Affected buildings per area

At all floodplains along the Lower Danube, less than one building per km2 (= high performance) can be found (see Fig. 6). At the Middle Danube, the highest value of around 35 buildings per km2 is observed, but the median of this section lies at around four buildings/km2 and is thus significantly lower than at the Upper Danube, which yields about 13 buildings/km2. In the middle section, nine floodplains are in the lowest performance class, eight in the middle class, and nine in the highest class. At the Upper Danube, nine floodplains are also in the lowest class, and only two floodplains are in the highest. The remaining three floodplains have a medium performance.

Fig. 6
figure 6

Affected buildings per km2 at all active floodplains along the Danube River. Blue color indicates high, green medium and orange low performance in the floodplain evaluation matrix (FEM) based on the defined thresholds. Insert shows the thresholds for the shown parameter

Full size image

3.5.7 Land use

Evaluating land use and linking it to vulnerability enables the determination of the floodplain's suitability for flooding. The analyses reveal a distinct pattern in each section. In the Upper Danube River, eleven floodplains exhibit medium vulnerability to flooding based on the land uses, resulting in a corresponding medium performance. Only on three floodplains, the vulnerability is low (= high performance). In the middle section, the majority (twenty-one out of twenty-four floodplains) showcase a low vulnerability, aligning with high overall performance; the other three demonstrate a medium performance. The vulnerability of floodplains in terms of land use is uniformly low in the Lower Danube, translating to only high performances across this entire section.

3.6 “Need for preservation” and “restoration demand”

A primary objective of the “Danube Floodplain” project was to identify floodplains that should be preserved and restored. To address this goal, we further developed the FEM by incorporating the “need for preservation” and the “restoration demand”. Detailed discussions of this novel approach are provided in the Sects. Based on the developed approach and evaluated parameters, we determined that all active floodplains along the Danube River should be preserved, since at least one parameter received the highest-ranking class in each floodplain. Furthermore, our analysis revealed that 50% (24) of the active floodplains showed a high restoration demand, 31% (15) a medium, and only 19% (9) a low restoration demand (see Table 7–Appendix). Notably, all floodplains with a high restoration demand are concentrated in the Upper or Middle Danube regions. Along the Lower Danube, eight floodplains exhibit a medium restoration demand, while two exhibit a low restoration demand.

4 Discussion

Our developed and applied method evaluates floodplains along an entire large river, providing an overview of their importance based on various parameters. While the method leads to recommendations for preservation and restoration, it is essential to recognize and address certain inherent limitations and uncertainties associated with our approaches. Hence, we start the discussion by thoroughly examining the identified limitations (see Sect. 4.1) to properly address and acknowledge them before discussing our findings in the following subsections.

4.1 Limitations

4.1.1 Hydrological and hydraulic parameters

Using hydraulic models to calculate hydrological and hydraulic parameters entails certain uncertainties and limitations, which are well-known through several studies with such models (Aronica et al. 1998; Hall et al. 2005; Pappenberger et al. 2005; Tritthart and Gutknecht 2007). In our case, this starts with the input flood waves. We used the shape of flood waves that were measured at the closest gauging station next to the investigated floodplain. The form and volume of a flood wave can vary significantly depending on the flood event and its origins. For example, flood waves can be shallower or steeper than those we used. Overall, the rainfall-runoff pattern is very complex and individual for each river and event (Haider 2014). Schober et al. (2013) showed that, in general, steep flood waves with shorter duration are attenuated stronger. Our calculated hydrological parameters only represent the results for one particular flood event. To compare the parameters, we ensured that the input hydrograph for each floodplain corresponded to a HQ100, which is a value where statistical inferences play a role since these extreme values are extrapolated from historical data. Other sources for uncertainties include terrain representation and roughness parameterization in the hydraulic models. Each country used available calibrated and validated models to calculate the parameters and reduce uncertainties. All countries, with the exception of Austria, utilized 1D models to compute the parameters. While 2D models are generally considered to be the preferred method for calculating these parameters (Connell et al. 2001), it is also possible to obtain acceptable results using 1D models (Ferstl 2020; Schober et al. 2013). Schober et al. (2013) and Ferstl (2020) showed that with suitable databases (measured hydrographs of flood events) and expertise, 1D models can accurately simulate the observed reduction in flood peak and translation of flood wave. In the study by Ferstl (2020), the deviation between measured and simulated flood peak reduction was approximately 3%.

In our study, the hydraulic models should be split into smaller models to represent a single floodplain with an in- and outflow section. As described in Sect. 2.3.2, the Hungarian modelling approach was different for practical reasons. They used a continuous hydraulic model to calculate the parameters. As long as the hydrograph at the beginning of each floodplain corresponds to a HQ100 this approach has no effects on the hydrological parameter. In the case of the hydraulic parameter, it is different since the water level change becomes larger when using a continuous model for each floodplain. Hence, the results of the hydraulic parameter for the Hungarian floodplains should not be compared with results from other countries.

4.1.2 Ecological parameters

The two ecological parameters that we applied to the entire river also have certain limitations. First, all the uncertainties mentioned in the previous Sect. 4.1.1 regarding hydraulic models play a role for the “connectivity” parameter as well since such models were used for this parameter too. Another limitation is that few temporal changes in lateral connectivity were considered, as it was determined by a hydraulic model using only three different steady discharges. However, as the catchment of the Danube River is greatly influenced by climate change and hit by serious droughts and floods, applying only three steady discharges might provide only a rough snapshot of the lateral connectivity. Moreover, we did not evaluate longitudinal and vertical connectivity, though they play an essential role in habitat continuity and species migration. Undoubtedly, all three types of connectivity are important, but one goal of this study was to develop a method that could be applied along an entire river and be transferable to other rivers. Therefore, data sources and parameters were chosen, which are easier to access and apply. Since the applied FEM has a flexible design, other analyses that consider all three types of connectivity could be applied at other case studies if the data availability is given. Some ideas about the other directions of connectivity can be found in Woolsey et al. (2007), or Fryirs et al. (2007).

For the number of protected species, we had to rely on the database of Natura2000. However, its applied methodology differs among the countries (Hošek and Dusek 2016). All plant and animal species were considered to be of equal value so as not to favor any of the protected species. In addition, the occurring number of each protected species was not considered in the method. Here one could still improve the applied approach.

Additionally, invasive species are increasingly important elements of riparian ecosystems. However, in this study, their number and abundance were not evaluated due to a lack of information.

4.1.3 Socio-economical parameters

The evaluated socio-economical parameters were: “affected buildings” and “land use”. The land use is derived from the CORINE LU/LC data set for 2018. This data set is extracted from high-resolution satellite imagery and is uniform for all of Europe. The geometric accuracy is better than 100 m, and the thematic accuracy is better than 85%. The smallest size of the land use units in the database is 25 ha (Büttner et al. 2021). On the scale of the floodplains, errors due to misclassification or the geometry are smaller and averaged out. The FEM ranking of each class is fixed and has been applied uniformly to all floodplains. Therefore, inconsistencies are minimized.

The data sets applied to derive the “affected buildings” parameters are more diverse. For some floodplains, buildings were counted based on visual interpretation on aerial photographs, sometimes (restricted) cadastral information was used, and there were also countries where high-resolution satellite imagery formed the base data set. The spatial resolution of the base data influences the identifiable buildings. Cadastral information is more accurate than satellite-based detection. Also, the methodology in different floodplains may differ slightly. In some cases, each individual building was counted based on remote sensing data, while in the case of cadastral data, the function of the buildings was taken into account, which could reduce the number of identified buildings. For the purpose of this study, it is acceptable to use both approaches, as the goal was to obtain a rough estimate of affected buildings. The quality of the visual identification of buildings may also vary according to the experience of the human interpreter (Lillesand et al. 2015).

4.1.4 Thresholds for FEM parameters

Due to the research objectives, we introduced thresholds for each parameter to determine three different performance classes, subsequently defining the “need for preservation” and the “restoration demand”. These thresholds were established based on empirical data from previous studies (e.g., flood peak reduction) but also used values from flood risk analyses that showed, for example, that damages are higher when the water depth exceeds 50 cm (de Moel and Aerts 2011) or emergency forces can prepare the public much better if they have more time (Friso et al. 2008). To determine the thresholds, various options for the Danube River were tested in a thorough process and a consortium of experts from various organizations, including national water authorities, universities and non-governmental organizations, was consulted. This collaborative effort resulted in thresholds specific to the Danube River that should not be seen as fixed values but rather as a basis for further floodplain analyses. Generally, the thresholds need to be individually defined for each river, taking into account river-specific characteristics (e.g., river type, basin size). This approach allows the consideration of the unique features of each river system during the definition of the thresholds.

4.2 Application of the FEM at the Danube River

The evaluation of the active floodplains along the Danube River with the FEM showed that the floodplains are important for flood protection, ecology, and socio-economy. Some floodplains are more relevant for flood protection, and others fulfil more ecological or socio-economic functions and services.

As the results for the hydrological parameter “relative flood peak reduction” demonstrated, most floodplains (62%) at the Upper Danube significantly reduce the flood peak by more than 2% of a HQ100 flood peak. At the Lower Danube, no floodplain was able to reduce the peak by this amount. We identified following reasons for that. First, dykes of hydropower plants in the upper reaches lead to later flooding of the floodplains, as already shown by Schober et al. (2013) at the Austrian Danube, where at some floodplains, the dykes are overtopped at discharges around a HQ5. Previous studies (Haider 2014; Schober et al. 2013; Skublics et al. 2016) showed that if the bankfull discharge is higher or later in the rising limb of the flood wave, the peak reduction will be larger as when the floodplain is flooded already at the beginning of the rise. Secondly, at the Lower Danube, the percentage of the former floodplains is smaller than along the Upper Danube (Eder et al. 2022b). Thirdly, the flood waves in mountainous regions such as the Upper Danube are generally steeper than the flood waves in lowland rivers. In the lower section, the Danube is such a lowland river. Studies from Haider (1995) or Schober et al. (2013) demonstrated that the flood peak reduction for steep flood waves with shorter duration is stronger. The investigations of Schober et al. (2013) have shown that the peak reduction for every floodplain in Austria is stronger for a steeper flood wave. In contrast, the flood wave translation can also be higher for flatter flood waves. In our study, we see barely significant differences between the performances in the three Danube sections for the “flood wave translation” parameter.

The results of the hydraulic parameter “water level change” showed some differences between the sections. The loss of the floodplain would lead in the upper section in 64% of cases to an increase of the water level by at least 50 cm. At the Middle Danube, 38%, and at the Lower Danube, only 10%. One reason for that might be that a higher percentage of the former floodplains is preserved in the upper section Eder et al. (2022b) and disconnecting these areas from the river would lead to the higher water levels in the river channel.

In contrast to the hydraulic parameter and the relative flood peak reduction, the ecological parameter “protected species” showed a different picture. At the Lower Danube, the floodplains have the best values (most protected species) from an ecological perspective. In the upper section, a higher percentage of former floodplains is preserved, but as the "land use" parameter also showed, land use in the active floodplains has changed significantly and most of the natural properties were lost. Agricultural and urban uses are much higher in the upper section, and this, of course, subsequently reduces the potential habitats for various species and increases the vulnerability against flooding. Floodplains are used for agriculture, especially in mountainous areas (Upper Danube) with limited space. There are barely any agricultural uses on the active floodplains along the Lower Danube. Flood characteristics of the lower section might also play a role in that since floodplains at the Lower Danube can be flooded for several weeks due to lower flood dynamics (Nagy and Kiss 2016; Varadi and Goncz 1999).

Urban usage is also highest in the upper section and is reflected in the parameter “affected buildings”. The median of the Upper Danube is with 13 buildings/km2 significantly higher than in the middle section (four buildings/km2). Floodplains are attractive for agricultural and settlement development due to their topographic characteristics, fertility, and water availability (Schober et al. 2013). The limited space for urban development in the mountainous countries of the Upper Danube promotes more urban developments in the narrow valley bottoms along the rivers (Schober et al. 2020) than at the Lower Danube, where less than one building per km2 could be found on the active floodplains.

The assessment of the ecological parameter “connectivity of floodplain water bodies” showed the effects of the massive floodplain loss occurred at the Danube River, which was demonstrated in previous studies (Eder et al. 2022b; Schwarz 2010). No active floodplain has kept their natural (historical) lateral connectivity. One reason for this is that all floodplains lost their former extent due to human interferences, morphological changes, etc. Very often dykes cut off parts of the former floodplains, and in our applied method, as soon as dykes disconnect the former floodplain, a high performance (= natural (historical) status) is impossible. In the upper sections, most of the floodplains (thirteen out of fourteen) showed even a low performance (= poor lateral connectivity). Some floodplains are only connected to the river when the discharge corresponds to a HQ5, due to the dykes of hydropower plants (Schober et al. 2013). In the Middle and Lower Danube, the active floodplains have mostly a medium performance.

In this study, we introduced the “need for preservation” to provide floodplain managers and decision-makers with a basis for deciding whether a floodplain should be preserved. The expert consortium of the project determined that all functions and services provided by the floodplain should be treated equally. After exploring various approaches, the consortium reached the conclusion that as soon as one parameter was evaluated with a high performance, the floodplain must be preserved, irrespective if the parameter is an ecological or a hydrological one.

The “restoration demand” was introduced to assess the current status of the floodplains with the FEM. Humans use and modify floodplains for many purposes (agricultural usage, urban development, leisure activities, etc.), affecting and changing their natural functions and services (Rajib et al. 2021; Schindler et al. 2014; Tockner and Stanford 2002). Restoration measures such as reconnecting side channels are considered valid solutions to restore the lost ecological and flood protection functions and services (Sanon et al. 2012; Serra-Llobet et al. 2022; Stammel et al. 2021). In our assessment with the FEM, we identified the most natural floodplains in the basin to perform best. Therefore, their “restoration demand” was evaluated as low, although this should not imply that no measures are needed to improve their performance. It only indicates that compared with the other floodplains, these are the closest to their natural state. In our evaluation, all relatively well-preserved floodplains received at least 27 points. Hence, this value was considered a suitable threshold, and all floodplains that reached at least 27 points were considered as having a low “restoration demand”. Floodplains where humans significantly altered the natural functions, services, and former extent of the floodplain reached not more than 23 points in our assessment. Therefore, all floodplains that scored less than 23 points were evaluated with high “restoration demand”. The floodplains that reached between 23 and 27 points received a medium demand for restoration. Since this study is the first one that evaluates the floodplains of an entire large river using the FEM, these presented thresholds should be seen and used as a starting point for later studies on other floodplains. These thresholds are flexible and could be changed based on the results of further studies.

4.3 Future usage of the floodplain evaluation matrix

As the vulnerability and limitations of structural flood protection measures have been clearly demonstrated during major flood events (de Kok and Grossmann 2010; Hutton et al. 2019; Jüpner 2018; Kundzewicz and Menzel 2005), floodplain preservation and restoration have become essential elements of European environmental and water policies (EC: European Commission 2000, 2007, 2009), as they provide functions and services related to flood defence and ecology. Nevertheless, floodplains are still under pressure due to societal changes such as economic development, urbanization, population growth, etc. (Steffen et al. 2007; Vörösmarty et al. 2010). In Europe, more than 1000 km2 are sealed each year, although the EU-goal is to achieve a “no net land take” by 2050 (European Parliament and Council, 2013). The previous land consumption has left its mark, especially on river floodplains. 90% of former floodplains have been lost or are negatively impacted by human activities in Europe and North America (Tockner and Stanford 2002). Floodplain evaluation using the FEM provides decision-makers and stakeholders with a basis for floodplain management decisions by demonstrating the importance and impact of floodplains on flood protection, ecology, and socio-economy. The results of the FEM can contribute to strengthening the preservation and restoration efforts by underlying the importance and effects of the floodplains. The FEM results are also presented in the latest version of the Danube River Basin (ICPDR: International Commission for the Protection of the Danube River 2021b) and Flood Risk Management Plan (ICPDR: International Commission for the Protection of the Danube River 2021a), leading to the conclusions that all active floodplains must be preserved and a concrete large-scale restoration program is needed to restore floodplains as much as possible. Evaluating and collecting information about other floodplains along other large rivers (e.g., Nile, Mississippi) is desirable and would support floodplain managers in the preservation and restoration of floodplains. Admittedly, fully implementing the FEM is a time-consuming process. Therefore, we developed a QGIS plugin to accelerate and simplify the evaluation procedure by delineating floodplains, calculating specific FEM parameters, storing the results, and generating factsheets for each floodplain. More information on the current version of this FEM-Tool is available at https://www.interreg-danube.eu/uploads/media/approved_project_output/0001/49/976f5f357d4013348ba4aa13cc79aa42e7912c43.pdf accessed on 17 January 2024.

5 Conclusions

This study aimed to develop a method to determine the effects of active floodplains on flood protection, ecology, and socio-economics at different river systems. To achieve this, we used and further developed the floodplain evaluation matrix (FEM) (Habersack et al. 2015; Habersack & Schober 2020) to allow an application to the Danube River. The challenge of our study was to develop a method applicable in all countries, respecting the different data availability, and replicable at other large rivers. At the same time, we considered floodplains' multiple functions and services by assessing hydrological, hydraulic, ecological, and socio-economic parameters. The active floodplains exhibited a range of flood mitigation benefits, including significant reductions in flood peak (up to 17%), deceleration of flood wave progression (up to 41.5 h), and lower water levels (up to 174 cm) during a HQ100 event. Meanwhile, other floodplains may not have a noticeable impact on flood events but play a crucial role in preserving biodiversity by providing essential habitats for protected species.

Besides evaluating the effects of floodplains, a task was to identify floodplains where preservation and restoration efforts should be taken based on the FEM. Hence, the “need for preservation” and the “restoration demand” were introduced and assessed, showing that all active floodplains should be preserved and that all floodplains need restoration measures since their natural functions and services were compromised. 81% of the floodplains demonstrated a high or medium demand for restoration, emphasizing the urgency of intervention.

The results of the FEM not only highlight the diverse benefits of floodplains but also underline the need for actions, particularly in preservation and restoration efforts. Presenting the results could convince planners and funders to strengthen their support for the preservation and restoration of floodplains, leading to implementation of preservation and restoration measures as demanded in water and environmental policy plans such as the EU Floods Directive (2007/60/EC) (EC: European Commission 2007) and EU Biodiversity Strategy for 2030 (EC: European Commission 2020).