FLOOD RISK OF GRASSLANDS IN THE MOUNTAIN AREA OF WESTERN ROMANIA

. Floods, some of the most destructive natural phenomena, are occurring with increasing frequency, against the backdrop of climate change in recent years. Floods affect both built spaces and natural areas, implicitly the lands used as grasslands. In the context outlined above, the purpose of this study is to identify grasslands susceptible to being affected by floods and to group and map them according to the intensity of the manifestation of the flood risk. For the flood risk assessment, both for the entire analyzed area and for the grassland areas, in the GIS environment, seven factors (spatialized in raster format) determining the occurrence of floods were taken into account: land use, altitude, slope, drainage density, rainfall, distance to rivers and distance to roads. The flood vulnerability map of the analyzed territory, implicitly of the grassland areas, obtained by summing up and the weighted participation of the seven variables, was reclassified into five risk classes/areas: no risk or very low risk, low risk, medium risk, high risk and very high risk. It has been shown that the most vulnerable are the surfaces located on land with low slopes, located at low altitudes and in the proximity of water courses. Regarding the areas of grasslands, 50% fall into the classes without flood risk, with very low and low risk, 28% in the moderate risk class, 16% present a high risk and 6% of the grasslands are classified in the risk class very high to floods. From the point of view of practical applicability, it is important to identify and map the grasslands at risk of flooding, so that measures can be taken to mitigate the destructive effects.


INTRODUCTION
Flooding can be regarded as one of the most destructive natural phenomena occurring worldwide, leading to damage to property and infrastructure or even loss of human life (Yu et  Scientific research shows an increase in the frequency of extreme phenomena attributed to global climate change, and in the case of floods, the increase and manifestation of precipitation is in particular blamed ( In the specialized literature, different flood risk assessment methods are described and tested: the historical disaster method (De Moel et al., 2011), the scenario simulation method (Willems, 2013), the index system method (Guo et al., 2014), but also different methods based on GIS and remote sensing (Hall et al., 2005;Li et al., 2017;Cai et al., 2019;Cai et al., 2021;Desalegn and Mulu, 2021;Hagos et al., 2022).
GIS and remote sensing techniques and methods are essential tools for mapping and monitoring flood hazards (Bisht et al., 2016;Antzoulatos et al., 2022;Katipoğlu and Acar, 2022; Pinos and Quesada-Román, 2022; Souza et al., 2022).Currently, the development of flood hazard models has become feasible on a continental and global scale, and their application can be extremely useful for preparing the population and implicitly for reducing the catastrophic impact of these phenomena (Dottori et al., 2015).
Scientific research shows that grasslands have enormous potential in terms of increasing the resistance of landscapes to floods and erosion: in high areas, grasslands reduce soil erosion and contribute to mitigating downstream floods, and in low-lying areas, they favor water infiltration and have a resistance higher in floods compared to other land use (Macleod et al., 2013;Bengtsson et al., 2019;Sărăţeanu et al., 2020;Hussain et al., 2021;Strock et al., 2022).In this context, it is important to quantify the erosion and flood mitigation potential of grasslands, which will help to assess the impact of policies aimed at influencing land use (Milazzo et al., 2023).On the other hand, on the contrary, it is important to identify and map the grasslands subject to the risk of floods, with different intensities, so that measures can be taken to prevent and combat their destructive effects, both for the vegetation and for the "physical" surface of the grasslands.
In the context outlined above, the purpose of this study is to identify grasslands susceptible to being affected by floods and grouping and mapping them according to the intensity of the manifestation of the flood risk.

MATERIAL AND METHODS 1. Study area
The study area is located in the southwest of Romania, partially on the territory of five counties (fig.1): Timiş, Hunedoara, Gorj, Mehedinţi and Caraș-Severin.From a physical-geographic point of view, the study area overlaps the Banat Mountains, the Poiana Ruscă Mountains and the Retezat-Godeanu Mountains Group in the Southern Carpathians (Posea and Badea, 1984).The relief of the study area extends, from an altitudinal point of view, over an interval between 58 -2472 m, the lowest values being in the depression areas, and the highest altitudes are found in the mountain peaks of the Southern Carpathians (fig.1).

Study methodology
In accordance with the purpose of the work, the work methodology was structured in three main stages (fig.2): the delimitation of the study area, the assessment of the risk of flooding at the level of the entire region and the assessment of the risk of flooding in the case of grassland areas.The final stage is the grouping of grasslands according to flood risk, from grasslands with no or very low risk of flooding, to grasslands in areas with a very high risk of flooding.
The delimitation of the study area was made according to the major relief units.The units included in the Banatului Mountains group, the Poiana Ruscă Mountains and the relief units from the Retezat-Godeanu group, from the Southern Carpathians, were selected, according to the delimitation proposed by Posea and Badea (1984).
In the flood risk assessment, the following variables were taken into account: land use, drainage density, rainfall amounts, land altitude and slope, distance from rivers and distance from roads.The raster maps obtained for each variable were reclassified into five classes, in accordance with the specific values.
To assess the risk of flooding, for the seven previously listed raster maps, the Weighted Sum function from ArcGIS was applied, which represents the weighted sum by overlapping them and multiplying each one by the "weight" in the assessment and then summing them up.The weight of each variable in the flood risk assessment was assigned as follows: land use mode, 10%; drainage density, 15%; distance from roads, 5%;  In order to establish the "rank" of flood vulnerability, the grasslands were classified into five classes: (1) grasslands with no or very low risk of flooding; (2) grasslands with reduced flood risk; (3) grasslands with moderate flood risk; (4) high flood risk grasslands and (5) very high flood risk grasslands.

RESULTS AND DISCUSSIONS
In accordance with the physical-geographical conditions in the study area, the grasslands are distributed differently, depending on the altitude and the mountain units; they add up to approx.15% of the total area of the area of interest (fig.3).The distribution of grasslands on altitudinal levels shows an uneven distribution, a specific situation for other mountain areas (Samfira et al., 2010).In the low areas, below 300 m, 11% of the total grasslands are located, on small areas, dispersed among other agricultural lands; in the area of the hills, between 300 -600 m, 32% of the grassland areas are concentrated, thus being the altitudinal floor with the largest share of grasslands; between 600 -1600 m altitude, the participation of grasslands decreases progressively, forest areas being dominant and registers a new increase above 1600 m altitude, in the alpine hollows of the high mountains (13% of the total grassland areas).
From a qualitative point of view, the analyzed grasslands are differentiated according to the physicalgeographical sub-areas in which they are located and the way of use (Moisuc et al., 1994 To assess the territory's vulnerability to flooding, a multicriteria analysis was applied in the GIS environment, based on a set of raster maps of all the involved parameters.The factors that contribute to the occurrence of floods, respectively the variables involved in the analysis, were established according to the specialized literature (Cabrera and Lee, 2020; Desalegn and Mulu, 2021; Hagos et al., 2022).
The land used in a hydrographic basin is of particular importance in terms of the manifestation of surface runoff, depending on the category of use they are slowed down or accelerated.Based on this principle, the land use map in the study area was reclassified into five categories: shrub vegetation, agricultural land (category in which grasslands are also included), anthropogenic surfaces, areas without vegetation and water surfaces (fig.4).Altitude is another major factor in the occurrence of floods.Flat areas, located at low altitudes, are susceptible to liquid runoff accumulation, compared to areas located at high altitudes, on steep terrain (Kazakis et al., 2015;Cabrera et al., 2020;Ogato et al., 2020).For the study area, the elevation map was obtained based on a Digital Elevation Model (DEM) with a spatial resolution of 25 m.To participate as a factor in the flood vulnerability analysis, the elevation map was reclassified into five classes, from 53 m, up to 2473 m, according to figure 5.
The slope is one of the most important factors in flood risk assessment because it determines the amount, duration and speed of water runoff (Hagos et al., 2022): on land with small slopes, the speed of water runoff is lower and the stagnation period is longer , these being more vulnerable to floods compared to lands with steep slopes (Bapalu, 2006;Rimba et al., 2017;Singh et al., 2020).For the study area, the slope map, expressed in percentages, was generated from the DEM, in the ArcGIS software, later being reclassified into five classes, depending on the degree of participation in the evaluation of the territory's susceptibility to flooding (fig.5).Drainage density, another relevant factor in flood risk assessment.In areas with high drainage density, the risk of flooding is lower (Wondim, 2016).In calculating the drainage density, the total length of the watercourse segments is measured, which is divided by the surface unit and expressed in km/km 2 (Elkhrachy, 2015).The drainage density map was obtained in the GIS environment through specific procedures (ArcGIS Documentation, 2023) and was later reclassified into five classes (fig.6).The distance from rivers is necessary in the flood risk assessment for the following reason: floods occur when extreme rains saturate the drainage systems and overflow, thus areas located in the proximity of watercourses are more prone to floods (Gigović et al., 2017).Maps with distances to roads and rivers have been reclassified, according to figure 6, to be included as factors in the assessment of the vulnerability of the territory to floods.
The seven previously described factors, represented in the form of reclassified raster maps, were "aggregated" to generate the flood vulnerability map of the analyzed territory (fig.7).The map has been reclassified into five classes, depending on the degree of flood risk.The most vulnerable are the surfaces located in low areas, with reduced slopes, located near water courses.According to the model presented in figure 7, the surfaces with the lowest degree of vulnerability are located in the high mountain areas.
On the flood susceptibility map generated for the entire study area, the areas used as grasslands were superimposed and extracted.It was thus possible to identify and assign them to different levels of flood risk (fig.8).Of the total areas used as grasslands, 14% are not at risk of flooding.These surfaces are mostly located in mountainous areas with high altitudes and steep slopes.Of the total grasslands, 36% have a low risk of flooding, and 28% have a moderate risk.Of the areas used as grasslands, 16% are at high risk of flooding, and 6% of the total fall into the very high risk class.These grasslands are located in river meadows or in their proximity, on lands with a very low slope and unfavorable drainage conditions.

CONCLUSIONS
The present study demonstrates the utility and applicability of geospatial models and methods for flood risk assessment.These methods and workflows can be applied to any land area.
At the level of the study area, seven parameters with a significant role in the production and manifestation of these extreme phenomena were taken into account to assess the susceptibility to flooding: altitude, land slope, atmospheric precipitation, drainage density, land use, distance from rivers and distance from roads.Based on these variables, it was established that the most vulnerable are the areas at low altitudes, with low slopes, located near watercourses and with unfavorable drainage conditions.Mountainous areas, especially at high altitudes, generally fall into classes with low flood risk.
Of the grassland areas analyzed, 50% are classified as very low and low flood risk, 28% are in the moderate risk class, 16% are high risk and 6% of the grasslands are classified as very high flood risk.In most cases, grasslands that are subject to the risk of flooding are located at low altitudes, on horizontal or gently sloping land, in river meadows.
The use of models for flood risk assessment offers the possibility of analyzing the region, as a whole, but also the spatial location and framing according to the intensity of extreme phenomena, thus they can be considered starting points in the management strategies of the respective areas, to prevent and combat the effects of these phenomena destructive.

Figure 2 .
Figure 2. Research methodologyAlthough the specialized literature presents different methods of delimiting grasslands(Simon et al., 2017; Simon et al., 2020), in the case of this study, their spatialization and representation was done according to the Corine Land Cover (CLC) database, 2018 edition.In order to identify the risk of flooding, the areas of grasslands in vector format were extracted from the map showing the risk of flooding in the entire region.In order to establish the "rank" of flood vulnerability, the grasslands were classified into five classes: (1) grasslands with no or very low risk of flooding; (2) grasslands with reduced flood risk; (3) grasslands with moderate flood risk; (4) high flood risk grasslands and (5) very high flood risk grasslands.

Figure 3 .
Figure 3. Distribution of grasslands in the study area (processing from: CLC, 2023; Geospatial, 2023) ; Moisuc et al., 2000), and from the point of view of biodiversity conservation, they are grasslands with High Natural Value and great diversity of species (Hoancea et al., 2017; Cojocariu et al., 2018; Cojocariu et al., 2019).

Figure 4 .
Figure 4.The land use map and the precipitation map of the study area (processing from: CLC, 2023; Harris, 2022) Rainfall is one of the most important factors in the production of floods.The higher the amount of rainfall the higher the surface runoff and implicitly the risk of flooding (Blistanova et al., 2016).The precipitation map was generated based on CRU TS v4.07 Data Variables: PRE data, for the year 2022 (Harris et al., 2022) and reclassified into five classes, according to figure 4. Altitude is another major factor in the occurrence of floods.Flat areas, located at low altitudes, are susceptible to liquid runoff accumulation, compared to areas located at high altitudes, on steep terrain (Kazakis et al., 2015; Cabrera et al., 2020; Ogato et al., 2020).For the study area, the elevation map was obtained based on a Digital Elevation Model (DEM) with a spatial resolution of 25 m.To participate as a factor in the flood vulnerability analysis, the elevation map was reclassified into five classes, from 53 m, up to 2473 m, according to figure 5.The slope is one of the most important factors in flood risk assessment because it determines the amount, duration and speed of water runoff(Hagos et al., 2022): on land with small slopes, the speed of water runoff is lower and the stagnation period is longer , these being more vulnerable to floods compared to lands with steep slopes(Bapalu, 2006; Rimba et al., 2017; Singh et al., 2020).For the study area, the slope map, expressed in percentages, was generated from the DEM, in the ArcGIS software, later being reclassified into five classes, depending on the degree of participation in the evaluation of the territory's susceptibility to flooding (fig.5).

Figure 6 .
Figure 6.Drainage density map, road distance map and river distance map

Figure 7 .
Figure 7. Flood vulnerability map of the study area

Figure 8 .
Figure 8. Classification of grasslands according to vulnerability to floods