The Use of GIS Techniques for Land Use in a South Carpathian River Basin—Case Study: Pesceana River Basin, Romania

Abstract

This study is essential for medium- and long-term land-use management, as land-use patterns directly influence local economic and social development. Geographic Information System (GIS) techniques are fundamental tools for analyzing a wide range of geomorphological processes, including relief fragmentation density, relief energy, soil texture, slope gradient, and slope orientation. The present research focuses on the Pesceana river basin in the Southern Carpathians, Romania. It addresses three main objectives: (1) to analyze land-use dynamics derived from CORINE Land Cover (CLC) data between 1990 and 2018, along with the long-term distribution of the Normalized Difference Vegetation Index (NDVI) for the period 2000–2025; (2) to evaluate the basin’s natural potential byintegrating topographic data (contour lines and profiles) with relief fragmentation density, relief energy, vegetation cover, soil texture, slope gradient, aspect, the Stream Power Index (SPI), and the Topographic Wetness Index (TWI); and (3) to assess the spatial distribution of habitat types, characteristic plant associations, and soil properties obtained through field investigations. For the first two research objectives, ArcGIS v. 10.7.2 served as the main tool for geospatial processing. For the third, field data were essential for geolocating soil samples and defining vegetation types across the entire 247 km² area. The spatiotemporal analysis from 1990 to 2018 reveals a landscape in which deciduous forests clearly dominate; they expanded from an initial area of 80 km² in 1990 to over 90 km² in 2012–2018. This increase, together with agricultural expansion, is reflected in the NDVI values after 2000, which show a sharp increase in vegetation density. Interestingly, other categories—such as water bodies, natural grasslands, and industrial areas—barely changed, each consistently representing less than 1 km² throughout the study period. These findings emphasize the importance of land-use/land-cover (LULC) data within the applied GIS model, which enhances the spatial characterization of geomorphological processes—such as vegetation distribution, soil texture, slope morphology, and relief fragmentation density. This integration allows a realistic assessment of the physical–geographic, landscape, and pedological conditions of the river basin.

1. Introduction

Previous studies have shown that land-use/land-cover (LULC) patterns vary from one country to another [1,2], and are influenced by physical–geographical, and socio-economic factors such as urban sprawl, peri-urbanization, and depopulation [3,4,5,6,7,8]. Urbanization is an example of land intensification driven by increasing population density, resulting in a higher vertical profile in urban areas compared to rural settlements [9,10].

The accelerated pace of urbanization—often leading to the occupation of agricultural land—has devastating consequences for local and regional environmental degradation and changes, such as droughts, habitat loss, land-use changes, and land desertification [11,12,13,14]; in particular, desertification is a primary consequence of land degradation triggered by excessive human activity [15,16]. In addition, these phenomena affect the agricultural sector by depleting existing land resources and degrading fertile soils, thus increasing the susceptibility of local regions [17,18,19,20,21]. Ecosystem services (ESs) are highly vulnerable to land changes induced by human pressure [22], especially urbanization and intensive agriculture [23,24,25].

Modern 21st-century planning relies on geospatial data and analyses of LULC dynamics to identify trends or developments, and even to make medium- and long-term forecasts [26,27,28,29,30]. These data are most often used in conjunction with GIS, defined as “decision support systems that involve the integration of spatially referenced data into a problem-solving environment” [31]. The ultimate goal of a GIS is not simply to map reality but to use maps to solve real-world problems, such as those encountered in land-use planning or green infrastructure (GI) [32,33]. At the same time, the spatiotemporal representation in a given territory is the product of the processing, organization, and graphic translation of spatial elements, as perceived by the authors [34,35,36]. Digital Elevation Models (DEMs) are an essential source of data for representing the physical–geographic features of Earth’s surface [37]. These data serve as the primary basis for visualizing and analyzing terrain texture and classifying the relief of the study area [38], in this case, the Pesceana River Basin in Romania.

Remote-sensing and GIS techniques have emerged as essential, innovative, invaluable tools for assessing, monitoring, and investigating LULC changes within a well-defined geographical space, in this case, a river basin [39]. GIS and satellite-based analyses offer cost-effective approaches for the spatiotemporal assessment of a developing watershed, such as the Pesceana River in Vâlcea County, Romania, where traditional monitoring methods may struggle to keep pace with rapid digitalization [40,41,42,43,44,45,46]. By leveraging these tools, researchers can closely monitor the loss of vegetation cover, the decline of vineyards and orchards, agricultural expansion, urban sprawl, and changes in water surfaces with high spatiotemporal resolution [47,48,49,50].

The CLC database facilitates detailed spatiotemporal analyses at regional, national, and global levels, regardless of the territory’s size. This is because European land cover data are characterized by spatiotemporal continuity, allowing researchers to accurately identify different land-use categories [51]. LULC data have a wide range of applications, from monitoring land-use planning to tracking regional sustainable development [52] by observing changes in forest cover, cropland, and urbanization. It has been stated that LULC data can vary in terms of coverage, resolution, collection methods, and validity [53]. Furthermore, these data can be obtained using various research methods, such as field surveys, photogrammetry, drone-captured aerial imagery, and satellite images, before being classified according to the specific methodology adopted by the researcher [54].

Watersheds and inherited rural landscapes are among the most fragile natural resources, being significantly affected by LULC changes in recent decades [55,56,57]. Researching the impact of these shifts on hydrology is an important step in developing river basin management strategies, including efficient water resource planning [58] and measures to conserve plant associations within the Pesceana River Basin in Romania. Land-use (LU) dynamics in time and space represent an accelerating process, largely determined by anthropogenic activity, which contributes to the fragmentation of vegetation (forests, meadows, and pastures), land degradation and erosion, loss of soil fertility, landslides, and the depletion or loss of floristic and faunal biodiversity [59,60,61,62,63,64,65,66,67,68].

Some scientific studies have highlighted changes in land use and land cover as a fundamental factor that destructively affects biodiversity and ecosystems, in which the provision of goods and services is very important for human well-being [69,70,71]. In the Pesceana River Basin, Vâlcea County, Romania, the provision of goods and services is strictly linked not only to natural, terrestrial, and aquatic ecosystems but also to sustainable land use. Sisay et al. (2024) considered the assessment of LULC’s impact on ecosystem services as essential for understanding the implications of land-use change on human well-being, identifying key synergies between different ESs, monitoring changes in ecosystem health, and informing decisions on land-use planning and management and conservation efforts in the medium and long terms [72]. According to Marino et al. (2023), forested areas are linked to the provision of regulating services, while grazing areas are linked to the provision of ecosystem services (ESs) [73]. Increasing the intensity of human use of grasslands and forests not only degrades ecosystems over time and gradually reduces biodiversity in a watershed but also unintentionally erodes the edaphic factor, thereby weakening the value of ecosystem services, such as soil protection and water resource conservation [74,75]. Furthermore, investigating the impact of land-use change on uncertainty in ESs across different scenarios can help illustrate how human–environment interactions create uncertainty in these services, thereby promoting the rational and efficient development of spatial planning in a small- or medium-sized watershed [76].

According to Reyes-Cedeño et al. (2025), identifying optimal areas for reforestation is an extremely complex process that clearly requires integrating several environmental and spatial factors [77]. Thus, Qiu et al. (2019) and Dziedzic et al. (2022) emphasized the need to consider current land use, as built-up areas are largely unsuitable for planting trees, shrubs, and bushes [78,79]. The following physical, geographical, and geological conditions are important environmental or spatial factors to consider: precipitation, temperature, hydrographic network (rivers, lakes), slope exposure, relief type, and topography of the study area [80]. The most recent research in the field has identified areas for reforestation using the latest spatiotemporal modeling methods and digital data. Specifically, Cruz-Bello and Sotelo-Ruiz (2013) explored decision analysis models and multi-attribute optimization through a geospatial decision tool in central Mexico; a simple multi-attribute assessment technique (SMART) was used to classify the soil erosion, land use and cover, slope, and precipitation by programming with integers 0 and 1 to efficiently maximize environmental benefits [81]. Keleş et al. (2013) used remote sensing techniques and Landsat 7 ETM and IKONOS satellite data to estimate the main priority areas for reforestation in Turkey based on canopy closure, slope, and soil moisture [82].

LULC and local and regional climate change can affect hydrological processes in river basins, leading to changes in flood frequency [83,84,85]. On the other hand, LULC changes due to anthropogenic activities irreversibly alter the characteristics of the basin and worsen rapid hydrological responses, resulting in destructive floods in low-lying areas [86,87,88]. Studies by Yang et al. (2018) and Hounkpè et al. (2019) highlighted the direct link between flood intensity and the spatiotemporal dynamics of land use through zoning ordinance [89,90], which is an effective tool that can be used for the appropriate and efficient management of floodplains through land-use planning, namely, flood level delineation and the application of appropriate land use and building codes to ensure the construction of structures resistant to severe floods [91,92]. The integration of spatial modeling and analysis in ArcGIS (version 10.7.2) with the capabilities of the CLC database helps address a gap in the existing literature. Although numerous studies have examined LULC dynamics in Romania [93,94,95,96,97], most have focused on large river basins or national-scale patterns, with relatively limited attention paid to smaller sub-basins such as the Pesceana River Basin. This gap has constrained our understanding of local-scale processes that drive spatial and ecological transformations.

We address these limitations by developing an integrated GIS framework that links multi-temporal land-use and land-cover data with a range of geomorphological indicators, including SPI, TWI, slope, and soil type. The aim was to track the spatial development of the catchment between 1990 and 2018, taking its natural potential into account. Each variable serves a specific diagnostic purpose: while SPI and TWI indicate erosion susceptibility and hydrological potential, respectively, slope and relief energy represent the primary physical constraints on land use. By refining these assessment models, the study offers a methodology that can be applied to other small and medium-sized Romanian catchments. Previous studies in the region have focused on investigating forest habitats and plant associations [98], while others have quantified invasive plant species within habitats of European interest [99]. Consequently, this study represents the first comprehensive investigation of land-use dynamics within the Pesceana River Basin, Romania.

The objectives of this research are: (1) to analyze land-use (LU) categories over an extended period (1990–2018); (2) to examine the Normalized Difference Vegetation Index (NDVI) distribution over a long temporal span (2000–2025); (3) to identify and map soil sampling points associated with specific vegetation types; and (4) to analyze the Digital Elevation Model (DEM) to derive the Stream Power Index (SPI) and the Topographic Wetness Index (TWI).

The results of this quantitative research are relevant to decision-makers, as the analyses contribute to natural resource conservation, the promotion of socioeconomic development in rural and urban communities, and long-term land-use planning. Furthermore, this study enhances the existing understanding of plant associations and soil types within the Pesceana River Basin, Vâlcea County, Romania.

2. Materials and Methods

2.1. Study Area

The Pesceana River Basin is located in the South-West Oltenia Development Region (Figure 1), in the south-eastern part of Vâlcea County. The map in Figure 1 provides detailed information on the location and characteristics of the Pesceana River Basin, highlighting geographical, administrative, and topographical aspects relevant to hydrographic studies or environmental analyses in this region of Romania. The representative administrative territorial units include Amărăști, Crețeni, Glăvile, Gușoeni, Lădești, Mitrofani, Pesceana, and Sutești.

The climate of the study area is temperate continental, influenced by its location within the Subcarpathians [100]. The multiannual average temperature is 10.40 °C, the average annual precipitation is 662 mm, and the relative humidity ranges between 75% and 78.7%.

The Pesceana River flows across the Subcarpathian hills and depressions for 45 km before flowing into the Olt River. From a hydrographic perspective, the basin features a well-developed network of numerous streams and valleys—direct tributaries of the Pesceana River—which cross the region from north to south. Its main tributaries are the Gușoeni, Verdea, Năvgrăpița, and Geamăna rivers. The Pesceana’s riverbed is wide and moderately meandering; consequently, during periods of low to moderate water levels, sandbanks form and divide the river into several branches.

Like all hilly springs, its sources and those of its tributaries have low discharge; thus, the Pesceana dries up during dry summers, becoming a semi-permanent stream [101]. Summer rains and sudden snowmelt abruptly increase the river’s discharge, sometimes causing minor damage, such as uprooting floodplain vegetation or flooding agricultural land in low-lying meadows.

Regarding the spatial distribution of rock types in the Pesceana River Basin (Figure 2), a uniform distribution of the Cândești–Frătești strata is noted, with large amounts of groundwater reserves, followed by gravel, sand, and clayey sand deposited in the main river beds; clays, sands, and marls with coal in the southern part of the basin; loess deposits in the south–southwest part of the basin forming the most fertile soils; and deluvial–proluvial deposits in the north, northeast, west–northwest, and south. These components are extremely important for understanding the processes that shape the relief, especially in the hilly and mountainous areas of the basin, as well as the extremely limited current deposits and banks in the south of the basin.

Demographic growth is a fundamental driver of intensifying anthropogenic impacts on the natural environment and governs land-use changes at global, national, regional, and local scales. Accordingly, a simple graph illustrating population trends provides a general framework for identifying the main stages of increased human influence on land use. In the study area, data from the National Institute of Statistics (NIS) [102] show a steady decrease in the population between 2005 and 2024 (Figure 3).

2.2. Data Sources and Methodology

The methodological framework integrates multi-source spatial data to analyze LU dynamics and natural potential in the Pesceana River Basin. CLC datasets (1990, 2012, 2018) and Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global DEM (30 m) were processed in ArcGIS 10.7.2 to derive thematic layers and the environmental indices.

Our spatial modeling approach was based on three core pillars: terrain morphology (including slope, aspect, and relief energy), hydrological variables—particularly SPI and TWI—and ecological indicators such as soil type and vegetation cover. We chose these specific parameters to better balance the physical constraints and opportunities that determine land-use patterns in the area. All spatial analyses were standardized in the Stereo 1970 projection. Field surveys were used to validate vegetation and soil maps and to record sampling points with Global Navigation Satellite System (GNSS). The integrated workflow emphasizes analytical interpretation over cartographic production, ensuring methodological transparency and reproducibility.

In this context, mapping LULC via GIS techniques is a standard approach widely utilized in the literature [103,104,105,106,107,108,109,110]. However, the process of mapping and quantifying land use could also be performed based on surveys of stakeholders [111], the local community [112], or visitors/tourists [113]. A study on ecosystem services in participatory land-use planning underlined the importance of the spatial dimension in territorial management; it emphasized that stakeholders primarily seek spatial solutions, as they are more interested in identifying ‘where to implement measures rather than understanding ‘why’ they are necessary [114]. Additionally, the grid square method (cartograms) was used to create maps highlighting the natural potential of the Pesceana River Basin.

To collect field data within the Pesceana River Basin, habitat boundaries and soil sampling points were georeferenced using a Garmin GPSMAP 64 GNSS unit set to the World Geodetic System (WGS) 84. The CLC database provides a comprehensive spatiotemporal representation of LULC at European, national, and local scales (

Table 1. LULC categories used in the Pesceana River Basin.

Current Number	Categories
1.	Discontinuous urban areas
2.	Industrial and commercial areas
3.	Non-irrigated arable land
4.	Vineyards
5.	Orchards
6.	Pastures
7.	Complex crops
8.	Land mainly occupied by agriculture
9.	Deciduous forests
10.	Water surfaces
11.	Natural grasslands
12.	Transitional shrubs

Source: authors’ processing of European Environment Agency data [116].

). This dataset offers high accuracy and resolutions well-suited to the objectives of this research [115].

The CLC data are particularly useful for assessing land changes over small or large areas, as well as over long periods [117]. Taking into account the spatiotemporal dimension of urban processes, such as urbanization, deurbanization, and urban restructuring, in addition to industrial processes such as natural resource exploitation [118], the CLC data provide a complex and detailed picture of the LULC dynamics in territorial statistical units; notably, for administrative units in the Pesceana River Basin and similar development regions. To create the CLC (1990, 2012, 2018) LU map for the Pesceana River Basin, we used the following databases: for representation of the relief (hillshade), we used the Global Digital Elevation Model (GDEM) (ASTER), version 2 [119]; and the LU data (CLC 1990, CLC 2012, CLC 2018) were downloaded from the Copernicus website [120]. The vegetation and soil texture map for the Pesceana River Basin was created by digitizing the vegetation and soil map of Romania [121,122]. The map of the spatial distribution of rock types was created by digitizing the 1:200,000 geological map produced by the Geological Institute of Romania (IGR) [123].

Reprojection of the data to Stereo 1970 was a mandatory step enabling the use of the data and the production of suggestive LU maps for the Pesceana River Basin [124,125,126,127]. Finally, ArcMap 10.7.2 was used to provide spatiotemporal representation and analysis via a GIS-based approach [128,129,130].

Landsat satellite images with a 30 m × 30 m spatial resolution were also used to cover the entire study area. For the year 12.06.2000, Landsat 7 Enhanced Thematic Mapper Plus (ETM+) imagery was employed; Landsat 8 Operational Land Imager/Thermal Infrared Sensor (OLI/TIRS) imagery was used for the year 14.08.2014; and Landsat 9 (OLI/TIRS) imagery was used for the year 03.07.2025. These satellite images were acquired from the United States Geological Survey (USGS) Earth Explorer platform [131], accessed on 15 June 2025.

The selection of images was based on strict methodological criteria that were designed to ensure temporal and spectral comparability of the data. Specifically, the following image properties were considered: (1) similar spatial resolution of the sensors (30 m); (2) a cloud cover of less than 5%; and (3) from the summer period, corresponding to the maximum photosynthetic activity and high chlorophyll content of the vegetation. It was not necessary to apply an additional correction for cloud cover, as the selected images contained less than 5% clouds, which did not influence the NDVI values. To identify and evaluate green areas, the Normalized Difference Vegetation Index (NDVI) was calculated for each of the three years analyzed (2000, 2014, and 2025) using the spectral bands corresponding to the red (Red) and near-infrared (NIR) ranges. The NDVI allows for highlighting the spatial distribution and dynamics of vegetation, as well as comparing its condition over time.

Three NDVI layers were generated for the years 2000, 2014, and 2025 using two bands of the satellite images.

The NDVI was used to quantify the extent and condition of vegetation cover within the Pesceana River Basin, Vâlcea County, Romania.

Assessing ecosystem dynamics requires a clear understanding of the variability of green vegetation in a given area [132,133]. The Normalized Difference Vegetation Index (NDVI) is a standard tool for this purpose. By utilizing the contrast between the red (RED) and near-infrared (NIR) spectral regions of the electromagnetic spectrum, the NDVI allows for the acquisition of important data on both vegetation density and the overall health of plants [134].

Mathematically, the index is the normalized ratio of NIR and RED reflectance—a relationship that many researchers use to quantify surface vegetation [135]. Its widespread use in remote sensing is based on its reliability in spatiotemporal mapping and vegetation cover assessment [136,137]. In our analysis of the Pesceana River basin, we used the NDVI to assess both the developmental stage and density of the local canopy. Healthy vegetation generally yields positive values, with higher values reflecting greater vitality and density [138]. To ensure the reliability of our results, the NDVI was calculated using Equation (1) according to the established protocols described in [139,140]:

$$\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}= {\displaystyle \frac{\mathrm{N}\mathrm{I}\mathrm{R}-\mathrm{R}\mathrm{e}\mathrm{d}}{\mathrm{N}\mathrm{I}\mathrm{R}+\mathrm{R}\mathrm{e}\mathrm{d}}}$$

(1)

where NIR describes the near-infrared band, and Red indicates the red band. The NDVI formula was adjusted to correspond to the spectral bands specific to the satellite image, such that the calculation could be performed in the Raster Calculator. Each image was processed separately because the spectral bands corresponding to the Red and NIR domains differ between the Landsat 7, 8, and 9 sensors due to changes and optimizations in the new platforms. The results are NDVI rasters (for each year, separately) with continuous values between −1 and +1, which were subsequently classified (Reclassify tool, see the classes below). The NDVI values obtained for the analyzed years were grouped into vegetation density classes according to the thresholds used in the specialized literature [141,142]. The classification aimed to facilitate the spatial interpretation of the state and distribution of vegetation, as well as the comparison of its evolution over time.

Thus, the following NDVI classes were defined:

−1–0 and 0–0.1: areas without vegetation or with negligible vegetation (water, bare soil, built-up areas);

0.1–0.2: very sparse vegetation;

0.2–0.4: sparse vegetation;

0.4–0.6: medium-density vegetation;

0.6–0.8: dense vegetation;

0.8–1.0: very dense vegetation.

This classification scheme allows for highlighting the degree of plant cover and the spatio-temporal variations in photosynthetic biomass, which are frequently used in remote sensing and environmental analysis studies.

Before the supervised classification, Landsat satellite images from 2000, 2014, and 2025 were radiometrically corrected. The supervised classification was performed using the following main steps: (1) creating a shapefile to define the classes of interest (e.g., built-up area, agricultural land, etc.) using the Image Classification tool; (2) digitizing the polygons for each class, which constituted the training set; and (3) applying the Maximum Likelihood Classification algorithm to perform the supervised classification. A cloud mask was not necessary because the selected images had a cloud cover of less than 5%, which did not influence the analysis results. Further information on the entire detailed procedure for preprocessing Landsat satellite images, the classification algorithms used, and the accuracy assessment can be found in Sisay et al. (2023) [143].

To assess the accuracy of the supervised classification, validation was performed with the Kappa coefficient using independent control points. The Kappa coefficient measures the agreement between the obtained classification and the reference data while considering the probability of chance agreement. Following Landis and Koch (1977), a Kappa coefficient score >80% indicates extremely strong agreement in the land classification, while a score of 40% indicates weak agreement between the reference and classified data [144]. The Kappa value obtained was over 85%, indicating very good classification accuracy and high reliability of the land-use change analysis results. Land-use changes for 1990, 2012, and 2018 were detected using the post-classification Change Detection method, with supervised classification unnecessary since the CLC data were already classified.

The changes were calculated using the following formula:

Increase/decrease area percentage = (new value − old value)/old value) × 100

In other words, for the change between 1990 and 2012, the formula was

Increase/decrease area percentage = (2012 area − 1990 area)/1990 area) × 100

For the change between 2012 and 2018, the formula was

Increase/decrease area percentage = (2018 area − 2012 area)/2012 area) × 100

To create a map showing how each land category changed, we used the Intersect tool. This tool identifies the overlap of polygons between two or more layers (layers/shapes), allowing for the calculation of areas.

When processing the Landsat images, the rasters were reprojected in the Stereo 70 system to ensure spatial coherence with the other layers in the project. During this operation, reprojection errors may occur, which manifest as small pixel shifts or spatial distortions between layers. To minimize their impact on the NDVI analysis and supervised classification, raster alignment checks and additional corrections were applied, thus ensuring adequate spatial concordance between all layers used in the Pesceana River Basin analysis.

After additional corrections, the Root Mean Square Error (RMSE) was calculated, which was less than two pixels, equivalent to less than 60 m, indicating a correct reprojection. The RMSE calculation was performed as follows: (1) The reprojection error of the Landsat images in the Stereo 70 system was quantified by comparing the positions of known Ground Control Points (GCPs) before and after the reprojection. (2) Control point selection: Several distinct, visible, and easily identifiable points were selected in the Landsat image (e.g., road corners, buildings, intersections). The same points were also identified in the images in the Stereo 70 reference system. (3) Error measurement: For each point, the difference in coordinates between the reprojected position and the real position was calculated. The total reprojection error was calculated as the RMSE, which combined the deviations along the X- and Y-axes for all points [145]:

$$\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}=\sqrt{{\displaystyle \frac{1}{\mathrm{n}}}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}[{\left({\mathrm{E}}_{\mathrm{X},\mathrm{i}}\right)}^2+{\left({\mathrm{E}}_{\mathrm{y},\mathrm{i}}\right)}^2]$$

where E_x,i represents the deviation along the x-axis (the difference in coordinates), E_y,i represents the deviation along the y-axis (difference in coordinates), and n is the number of control points. The result is expressed in the coordinate system’s measurement unit (in our case, meters).

Topographic indicators, such as the SPI (Stream Power Index) and TWI (Topographic Wetness Index), calculated from the ASTER GDEM DEM (30 m), are suitable for characterizing hydrological conditions and water flow in the Pesceana River Basin. These indices are used to identify areas with erosion potential (SPI) or to evaluate relative soil moisture and water accumulation (TWI). The SPI and TWI are based on calculating the slope and flow accumulation for each cell of the DEM.

The SPI is defined as follows [146]:

$$\mathrm{S}\mathrm{P}\mathrm{I}={\mathrm{b}}^2=(\mathrm{A}\mathrm{s}\times \tan \mathsf{\beta })$$

where As represents the specific water catchment area, and $\beta$ is the local slope gradient measured in degrees.

The general formula for the TWI is as follows [146,147,148]:

$$\mathrm{T}\mathrm{W}\mathrm{I}=\mathrm{l}\mathrm{n}{\displaystyle \frac{\mathrm{a}}{\mathrm{t}\mathrm{a}\mathrm{n}\mathsf{\beta }}}$$

where a is the cumulative area in the upslope that drains through a point (per unit length of the contour), and tan$\mathsf{\beta }$ is the local slope of the ground surface (in radians).

In GIS and hydrological analysis, flow accumulation is a raster that indicates the number of cells or contributing areas that flow through each pixel of the DEM. Flow accumulation does not have a simple formula; it is calculated using a flow-direction algorithm (e.g., the D8 algorithm). The flow accumulation raster receives a value representing how much water would theoretically flow through that pixel from all upstream cells; that is, it indicates the drained area that contributes to the flow through each cell. High values represent valleys, streams, and main streams, while low values represent ridges, steep slopes, and areas with rapid flow and little accumulation.

At 30 m resolution, ASTER GDEM and SRTM provide sufficient granularity for the Pesceana basin, such that local variations in slope and flow accumulation are not very sensitive to DEM resolution. The reason for this lack of sensitivity is that the basin has moderate topography, with no complex microrelief, which is only captured at finer resolutions (<10 m). In general, for the TWI and SPI, sensitivity increases in steep areas or with very fine-resolution DEMs, where local variations in slope significantly affect flow accumulation.

For small- or medium-sized basins with moderate topography, local topographic variations captured at 30 m are sufficiently representative of indicators, such as the TWI and SPI, without generating drastic changes in values when moving to slightly finer or lower resolutions (e.g., 20–60 m). In such cases, the spatial variability of the slope or upstream area is relatively stable at the 30 m scale, which explains the low sensitivity observed in practice. The choice of the ASTER GDEM to analyze the Pesceana River Basin is justified from several points of view. Although SRTM is a recognized and robust source for digital elevation models, ASTER GDEM offers uniform global coverage, a 30 m resolution sufficient for medium-scale hydrological studies, and free accessibility. ASTER GDEM is also as effective as SRTM at capturing essential details of hydrological flow and topography, with minimal differences in results for indicators such as the SPI and TWI. In addition, recent improvements in data quality and excellent compatibility with GIS software 10.7.2. make ASTER GDEM a valid and efficient choice for analyzing the Pesceana River Basin. ASTER GDEM is suitable for medium-scale hydrological studies, as it can capture relevant topographic details at this resolution. SRTM tends to be more detailed in flat areas with fewer topographic artifacts. ASTER GDEM and SRTM can present measurement/topography errors in mountainous/steep areas. ASTER GDEM is suitable for medium- and large-scale analyses, especially in cases where global coverage is important. It performs very well for hydrological flow studies or TWI/SPI calculation. SRTM is robust for flow analyses in flat areas and has a good reputation in global hydrology studies. However, in mountainous areas, it can be less accurate than ASTER GDEM.

The SPI raster reclassified into four risk classes was intersected with the habitat samples (the point shapefile with the habitat samples). Using the Spatial Analyst Tool (Extraction—Extract Values to Points) in ArcGIS, we quantified the SPI class corresponding to each observation point (sample) in each habitat type. This allowed us to assess the exposure of the habitats to erosion risk. In the calculations file, we calculated percentages reflecting the proportion of points (samples) in each SPI class in relation to the total number of points (samples) in that habitat. At the same time, we also calculated the total number of samples in each risk class. We used the habitat sample shapefile and not the IDW raster to extract SPI values because the points represent actual observations and allow for an accurate assessment of the exposure of each habitat type to erosion risk. The IDW rasters represent only the weighted spatial influence, not the actual habitat, and using them directly would have led to a generalization that does not reflect the field data.

To represent the presence of different habitat types, each habitat was coded as a binary variable, corresponding to columns h1, h2, h3, h4, h5, and h6. Each column contains values 1 or 0, indicating the presence (1) or absence (0) of the respective habitat at the observation points. This binary coding allowed the application of the IDW method to interpolate the influence of the habitats in space. For each habitat, IDW was applied separately, thus generating six continuous rasters, one for each habitat type. Each pixel of the resulting raster receives a value between 0 and 1, corresponding to the influence of the respective habitat depending on the proximity to the points with value 1. This allows the relative assessment of the dominance of the habitats, without claiming that the rasters represent exact ecological boundaries or probabilities of presence. After generating the six IDW rasters, they were combined into a final raster (categorical type), which reflects the dominance of the habitat at each observation point. This was achieved by assigning each pixel to the habitat corresponding to the maximum influence value among the six rasters. The result represents the habitat with the strongest influence in the respective area.

Similar to the case of habitat types, we performed IDW interpolation for soil samples to represent the presence of different types of environments (anthropogenic, natural, semi-natural). Each type of environment was coded as a binary variable, corresponding to p1, p2, and p3. Each column contains values 1 or 0 (presence or absence); through IDW, three continuous rasters were created, which were subsequently joined, according to the methodology described previously.

Contour lines (isohypses), which connect points of equal elevation, were extracted from the DEM. These lines were subsequently used to determine water storage capacity and estimate excavation volumes. Relief fragmentation density was calculated using the grid square method (cartograms). This process involves a series of steps: fragmentation density can be determined by extracting centroid points from each grid square and interpolating between them.

Another DEM processing method involves applying the grid square method to calculate relief energy, which represents the vertical component of relief fragmentation. A cartogram is a thematic map that displays areas with homogeneous properties, represented by polygons shaded in proportion to their attribute values [149]. The method of squares consists of superimposing a regular grid of squares over the study area, with each square representing a spatial unit of analysis. Inside each square, synthetic values of geomorphometric parameters are calculated (e.g., total length of valleys, maximum elevation difference).

Based on the DEM, two complex processes were carried out to obtain the slope gradient and slope orientation within the Pesceana River Basin. The slope gradient represents the inclination of the land surface relative to the horizontal plane, generally calculated in degrees (but it can also be expressed in percentage or permillage).

The density of relief fragmentation expresses the degree of surface dissection by the valley network and is defined as the ratio between the total length of the valleys and the area of the analyzed unit [150,151]. Methodology in ArcMap: Preparing the hydrographic network: Topology check, redesign if necessary; creating the square grid (ArcToolbox—Data Management Tools—Sampling—Create Fishnet—cell size: 1000 m, geometry type: polygon)—subsequently, the grid is cut according to the study area (Clip tool); intersecting the hydrographic network with the grid (Analysis Tools—Overlay—Intersect, input: Fishnet + hydrographic network)—result: river segments in squares; calculation of the length of the hydrographic network in each square (add new field (Length_km)—Calculate geometry—Length (Kilometers), then Spatial Join (Target: Fishnet, Join: intersected network, Statistics: SUM for summing the river segments in each square, separately); calculation of the fragmentation density (add new field (Df)—Field Calculator—SUM_Length_km/Area of squares); cartographic representation: Symbology—Graduated Colors, with classification into six categories. The relief energy expresses the vertical amplitude of the relief in a spatial unit (in our case, in a square) [152]. Methodology in ArcMap: DEM preparation (Spatial Analyst Tools—Hydrology—Fill); calculation of the local maximum altitude (Spatial Analyst Tools—Neighborhood—Focal Statistics, parameters: Input Raster—DEM, statistics type: Maximum, Neighborhood: Rectangle, 1000 × 1000 m); calculation of the local minimum altitude (Spatial Analyst Tools—Neighborhood—Focal Statistics, parameters: Input Raster—DEM, Statistics Type: Minimum, Neighborhood: Rectangle, 1000 × 1000 m); calculation of the relief energy (Spatial Analyst Tools—Map Algebra—Raster calculator: DEM max − DEM min); cartographic representation: Symbology—Graduated Colors, with classification into eight categories.

Figure 4 summarizes the conceptual workflow of the proposed GIS-based framework. The diagram illustrates the transition from spatial data acquisition and processing to integrated spatial outputs and their interpretation for decision-support purposes. Rather than functioning as a predictive model, the framework supports territorial prioritization and risk-aware planning at the local administrative level.

3. Results

We present the research results in sequential order in this section, according to the proposed research objectives.

3.1. Land-Use Distribution in the Pesceana River Basin

The evolution of LU within the Pesceana River Basin, Vâlcea County, Romania, was assessed for the period 1990–2018. Based on the systematic processing of land cover data, the 1990 map displayed 12 (LULC) classes (Figure 5a), while the maps for 2012 and 2018 each displayed 10 classes (Figure 5b,c).

According to the (CLC) database, the LULC categories identified in the study area include the following: discontinuous urban areas, industrial and commercial areas, non-irrigated arable land, vineyards, orchards, pastures, complex cultivation patterns, land mainly occupied by agriculture, deciduous forests, water surfaces, natural grasslands, and transitional woodland-shrub areas. Among these, deciduous forests are the dominant land-use category in the Pesceana River Basin.

From approximately 85 km² in 1990, the area covered by deciduous forests increased to more than 95 km² in both 2012 and 2018 (

Table 2. Surface km² of land-use/land-cover categories in the Pesceana River Basin.

Current Number	Categories	Surface (km2)
		1990	2012	2018
1.	Discontinuous urban areas	19.04	14.11	14.11
2.	Industrial and commercial areas	0.03	0.03	0.03
3.	Non-irrigated arable land	15.77	17.07	17.37
4.	Vineyards	42.20	7.89	7.64
5.	Orchards	7.82	5.93	5.93
6.	Pastures	8.98	15.44	15.14
7.	Complex crops	38.70	28.67	28.67
8.	Land mainly occupied by agriculture	19.62	55.19	55.44
9.	Deciduous forests	85.12	96.64	96.64
10.	Water surfaces	0.02	0.02	0.02
11.	Natural grasslands	0.40	-	-
12.	Transitional shrubs	3.29	-	-

Source: authors’ processing of The Copernicus Global Land Service data [120].

). Likewise, land mainly occupied by agriculture expanded from about 19 km² in 1990 to over 35 km² in 2012 and 2018 (Table 2). These changes suggest that the population within the basin remains heavily dependent on agricultural activities, including crop cultivation and livestock farming.

The land-use distribution in 1990 shows that deciduous forests were found in the north-west of the Pesceana River Basin, complex crops were mostly found in the north-east, and urban areas were dispersed throughout the territory—mainly along the river courses (Figure 5a). The lands occupied by crops were mainly in the west–southwest, the south, and along river courses, as water availability is the main factor in the development of agricultural practices. These crops serve not only as a source of food for the local population but also as a source of income.

The area of land mainly occupied by agriculture in the northern, central, and southern parts of the Pesceana River Basin increased in 2012 (Figure 5b). This increase was also confirmed by statistical data from the NIS for the two territorial administrative units: Crețeni and Glăvile. Crețeni recorded agricultural areas of 1613 and 1715 hectares in 1990 and 2012, respectively. Moreover, Glăvile recorded agricultural areas of 2293 and 2739 hectares in 1990 and 2012, respectively [69]. This increase in area was mainly due to the mechanization of agriculture after 1990 and to the pedoclimatic conditions of these areas, which are favorable to agriculture. Compared with 1990, natural meadows and transitional shrubs had completely disappeared by 2012.

The successful application and use of smart technologies—such as GNSS, drones, and irrigation water control—further increased the area of land occupied by crops in 2018 (Figure 5c), when compared to that in 1990. The use of satellite images to optimize agricultural crop technology, soil fertilization, pest protection, and weather warnings is among the most widely used practices that have shown yield returns in the development of agricultural crops in recent periods.

The land-use changes in the Pesceana River Basin (Figure 6) for the period 1990–2012 were decreases of −34.30 km² in vineyards, −4.92 km² in discontinuous urban areas, −10.03 km² in complex crops, −3.29 km² in transitional shrubs, and −1.89 km² in orchards. Increases were recorded for agricultural land (35.57 km²), followed by deciduous forests (11.51 km²) and pastures (6.45 km²). For the period 2012–2018, increases were recorded for deciduous forests (2.83 km²) and orchards (2.32 km²), and decreases were recorded for the following land categories: discontinuous urban areas, complex crops, non-irrigated arable land, vineyards, and pastures.

The matrix of land-use and land-cover changes was quantified by highlighting the increases and decreases in the areas of each land-use class according to the National Institute of Statistics (NIS) (

Table 3. Matrix of land-use and land-cover changes in the Pesceana River Basin.

Land-Use and Land-Cover Classes	Area 1990 (Hectares)	Area 2014 (Hectares)	Difference (Hectares) (1990–2014) *
Agricultural	15.687	15.615	−72
Arable	5.817	6.306	+489
Pastures	7.061	7.458	+397
Haylands	718	376	−342
Vineyards and wine nurseries	1.011	795	−216
Orchards and fruit nurseries	1.080	680	−400
Non-agricultural land	:	10.634	10.634
Forests and other forest vegetation	:	8.321	8.321
Occupied with water, ponds	:	356	356
Occupied with constructions	:	912	912
Roads and railways	:	517	517
Degraded and unproductive land	:	528	528

Source: authors’ processing of NIS data [102]. : Missing data. * Values represent the difference between the increased and decreased areas during 1990–2014 at the level of each administrative-territorial unit and each class, illustrating the increase or decrease in areas.

).

Four of the twelve land-use and land-cover classes recorded decreases in total area in 2014 compared with 1990. The rest of the classes (namely, two measured and six with positive values due to the lack of statistical data for 1990) show increases in area during the analyzed time interval. The increase in the pastures land-use category in the Pesceana River Basin represents an important factor in local economic development through sheep breeding and the provision of traditional products such as milk, cheese, and wool for visitors [153]. For the local habitat, sheep dogs are also important for guarding sheep against carnivores [153,154].

3.2. NDVI Distribution in the Pesceana River Basin

Analysis of the 2000 NDVI values across the six vegetation categories in the Pesceana River Basin (Figure 7a) identified 80.63 km² of very dense vegetation, representing 33% of the total area. Medium-density vegetation covered 76.54 km² (32%), followed by dense vegetation at approximately 55.26 km² (23%). The smallest extent was recorded for sparse vegetation (30.15 km² or 12%), while the ‘lack of vegetation’ and ‘very sparse vegetation’ classes both occupied 0 km².

By 2014, these vegetation classes exhibited notable shifts. The largest category remained very dense vegetation, which expanded to cover 125.21 km² (52% of the total area). Dense vegetation increased to 103.24 km² (42.5%), whereas medium-density vegetation decreased substantially, occupying only 13.30 km² (5.4%) (Figure 7b). A comparison of the 2025 data reveals significant changes in vegetation density since 2014. Particularly striking is the expansion of very dense vegetation by 27.10 km², representing a growth of approximately 11% (Figure 7c). This upward trend contrasts sharply with the “dense vegetation” category, which saw a decline of 42.34 km²—a decrease of about 17% over the same period. Simultaneously, sparse vegetation, which still constitutes a smaller proportion of the catchment area, increased from 0.8 km² in 2014 to 3.6 km² in 2025.

The analysis of land-use patterns in the Pesceana River basin (Figure 8) was based on a multi-temporal dataset comprising Landsat 7 (12 June 2000), Landsat 8 (14 August 2014), and Landsat 9 (3 July 2025) images. Our results show that built-up areas expanded by approximately 19 km² in 2000, but this trend reversed, with a slight decrease observed between 2014 and 2025. While agricultural land reached its peak in 2000 with an additional 133 km², it remained relatively stable between 2014 and 2025. Particularly noteworthy is the development of forest areas: these exceeded 90 km² in the period 2014–2025, a significant increase that probably indicates active reforestation or natural regeneration. Conversely, in 2000, forested areas decreased by more than 10 km² compared to the 2014–2025 period.

Grassland vegetation was reduced to 0 km² in both 2014 and 2025, suggesting either conversion to agricultural land or replacement through afforestation activities. Water surfaces remained constant throughout the analyzed period, indicating no significant hydrological alterations.

To illustrate more recent vegetation trends, we have included preliminary NDVI data for 2025. These results show a qualitative correlation between the increase in very dense vegetation and the forest expansion already documented in our LULC maps. However, these observations remain preliminary; future research will prioritize statistical trend analyses to more precisely quantify how NDVI values and forest cover changes correlate over time.

3.3. Analysis of the Natural Potential of the Pesceana River Basin

Regarding the hypsometry of the Pesceana River Basin, it ranges from 136 m to 485 m. In Figure 9, the relief of the study area is represented using contour lines. Environmental factors such as temperature and atmospheric precipitation are closely related to altitude, positively or negatively influencing land cover [155,156,157,158,159]. In the Pesceana River Basin, the areas with the lowest altitude may remain largely intended for agriculture due to their suitability for development. On the other hand, areas with high altitude may remain better protected from anthropogenic activity.

The topographic profile highlights decreases in altitude over long distances, dropping to 280 m over the first 9000 m, while over the next 30–100 m, the altitude registers significant variations, with sudden decreases.

Relief fragmentation density within the Pesceana River Basin is predominantly characterized by very low values (below 1.5 km/km²), which cover the largest portion of the study area (Figure 10). This trend indicates generally low terrain fragmentation—a condition that favors agricultural development, the establishment and expansion of settlements, and the construction of critical technical and municipal infrastructure, including water supply, sewage systems, and transportation networks.

Alternatively, fragmentation density can be determined using the isoline method, by extracting centroid points from each grid square and interpolating between them. This analysis confirms that the Pesceana River Basin has generally low fragmentation, which favors agricultural development, urban expansion, and the extension of technical and municipal infrastructure (e.g., water and sewage networks, communication routes).

Regarding relief energy, values below 20 m/km² are confined to a very small area in the southern part of the basin (Figure 11). In contrast, the highest values (exceeding 100 m/km²) are extensively distributed across the north, northwest, west, northeast, and east–southeast. This distribution is primarily driven by the intense relief fragmentation caused by the basin’s main river systems.

Another method to process a DEM involves using the method of square modules (cartograms) to obtain the relief energy, which represents the vertical relief fragmentation.

The vegetation of the Pesceana River Basin constitutes a critical life-support system for local human communities [160], integrating multiple components of the natural environment, including climate, hydrology, edaphic factors, fauna, and flora. Currently, natural and semi-natural vegetation occupies a limited portion of the basin, largely due to anthropogenic pressures that have significantly reduced its extent in favor of agricultural expansion. River meadows and low hill terraces exhibit a fragmented distribution as a result of cultivation, leading to notable shifts in floristic composition and community dynamics.

A defining characteristic of these changes is the widespread prevalence of ruderal and segetal species, which constitute highly ruderalized habitats. This dominance of ruderal vegetation is not accidental, but rather reflects intensive agricultural use and ongoing anthropogenic influence on formerly natural or near-natural plant communities.

As shown in Figure 12, the vegetation in the Pesceana River basin is far from uniform. It is characterized by hilly terrain, fluctuating elevations, and proximity to the river. The presence of mixed forests of oak, hornbeam, and gorun indicates a predominantly temperate climate, but the presence of thermophilic oak species suggests microclimates exposed to higher temperatures. Along the riverbeds, meadow vegetation remains a crucial factor for the health of aquatic habitats. This structural complexity not only highlights the ecological diversity but also shows how the local flora adapts to the specific stresses of the respective subzones due to a constant interplay of biological and geographical factors.

The soil distribution in the Pesceana River basin directly reflects the interactions between local topography and climate over time. For example, loamy and clayey soils favor forest ecosystems and certain crops that can tolerate low drainage. Sandy-clayey soils and more heterogeneous soil profiles, on the other hand, provide the necessary versatility for a wider range of agricultural uses. This soil mosaic not only defines the agricultural boundaries of the basin but also determines its capacity to maintain biodiversity in different subzones. As shown in Figure 13, these soil types are classified according to their particle composition and structure. Understanding these nuances is more than a purely taxonomic exercise; it is essential for assessing soil fertility and drainage potential. Mapping each soil type according to its specific texture and geographical location allows for a better evaluation of its suitability for intensive forestry or diversified agriculture.

The slope gradient distribution for the Pesceana River Basin indicates moderate values for approximately half of the area (30–40°) (Figure 14), while surfaces with low inclination (below 10°) were generally located in the southern area and on the river courses; in these areas, the land is almost flat, thus favoring not only the location and expansion of buildings but also the development of agricultural crops.

Two complex spatial analyses were performed using the DEM to determine slope gradient and aspect within the Pesceana River Basin. Slope aspect (orientation) is crucial for studying the influence of solar radiation and insolation on geomorphological processes. In the study area, western and southwestern exposures predominate, characterizing the slopes as semi-dry and semi-warm (Figure 14). Notably, southern slopes—which receive the maximum solar radiation—exhibit a higher forest line compared to northern slopes. In contrast, northern and northeastern exposures receive the least solar radiation, making them the coldest and wettest areas in the basin.

The DEM also facilitated the calculation of the SPI and TWI. The SPI reflects the erosive power of surface runoff by integrating slope and catchment area [161].

Negative SPI values identify areas with potential for sediment deposition. In the Pesceana River Basin, these values represent the smallest area, with minimums reaching −13; this indicates high flood susceptibility in the southern part of the basin and along the main river channel (Figure 15). Conversely, high SPI values correlate with steep slopes where erosion is intensified, increasing the risk of land degradation. A maximum value of 10 was recorded across the majority of the Pesceana River Basin.

The TWI is an indicator that reflects the equilibrium moisture—that is, the spatial dispersion of soil moisture or the areas where water tends to accumulate—and is also called the Compound Topographic Index (CTI) [162,163]. In this study, the SPI values were reclassified using the quantile method to delineate between very low-, low-, moderate-, and high-risk areas over the entire study area (Figure 16). The SPI values were often strongly skewed, with a few extreme-risk areas and many low- or moderate-risk areas. Quantile classification allows the data to be divided into classes with an equal number of observations, ensuring a proportional representation on the map and allowing for comparison of relative risk areas without extreme values dominating the visualization.

This approach allows for risk comparisons between areas but introduces some interpretative peculiarities:

• Sample overlaps with high-risk areas: Samples located in high-SPI classes correspond to locations with high combinations of slope and specific area. This correlation suggests that the SPI may be a useful predictor for identifying areas with high erosion potential.
• Non-overlapping areas often occur in valleys: In deep valleys, the slope tangent may be very small, and the SPI value may be moderate even though the specific area is large. Thus, the SPI may underestimate the risk of concentrated erosion in these areas. In valley areas, although the drainage area is large due to runoff convergence, the slope angle is small, and its tangent tends to small values. As a result, the contribution of the slope tangent limits the final SPI, ensuring it remains moderate or low.
• Methodological limitations and error sources: DEM resolution—a lower-resolution DEM may omit the local microrelief and small slope variations, affecting the SPI calculation accuracy. GPS accuracy of the samples—positioning errors may cause points to be placed in different SPI classes. Factors not included—the SPI does not consider the soil texture, vegetation cover, or anthropogenic practices, which significantly influence real erosion.

The SPI provides a topographic estimate of erosion potential, but its interpretation should be performed with caution. Overlaying samples with SPI risk areas confirms the relevance of topography, but areas with an apparent low risk, such as valleys, may in fact be vulnerable to concentrated erosion. Thus, the SPI must be complemented with field data and geomorphological considerations for an integrated assessment of erosion risk.

The TWI values generally ranged from 2 to 26 (Figure 17) and can help identify areas that may be affected by flooding and are highly susceptible to the formation of marshy surfaces.

TWI values of 26 are correlated with a high water accumulation potential due to the low slope gradients in the respective area; these values were recorded in the direction of flow of the Pesceana River, mainly on the main tributaries located in the south of the basin and partially in the north (Figure 17). The predominance of values ranging from 2 to 10 was observed throughout most of the basin, reflecting the prevalence of high slopes that do not allow the accumulation of water reserves for human consumption, animal husbandry, and agricultural use.

Overlaying the SPI and TWI indices reveals that agricultural expansion is concentrated in areas with both low slope and low TWI. This spatial overlap likely reflects a preference for specific hydrological and topographic conditions conducive to agriculture. While these initial patterns are suggestive, we intend to go beyond qualitative observations in further research by applying rigorous statistical tests and correlation analyses to quantitatively define these relationships.

3.4. Field Sampling in the Pesceana River Basin

Figure 18 illustrates the soil sampling points in the Pesceana River Basin, with each point associated with a specific vegetation type and a degree of naturalness (natural or anthropized). While most of the samples came from semi-natural grasslands and forests, some areas were influenced by human activities. The study area includes various habitats, from mixed forests and oak associations to meadows with varied plant species, reflecting the local biodiversity and differences in anthropogenic influence. These soil sampling points were distributed throughout the Pesceana River Basin and were chosen to reflect the diversity of soil types and vegetation near the localities in the region.

Most habitat samples (approx. 57%) are located in low or very low risk areas; almost 16% are in medium risk areas, and approx. 27% of the samples are in high-risk areas (Figure 19). Most habitat samples are located in valley areas, especially in the floodplains associated with the hydrographic network. This distribution may reflect the geomorphological and hydrological conditions favorable to the development of the analyzed habitats. Valleys are characterized by low slopes, the accumulation of fine sediments, and relative geomorphological stability, which favor the establishment and maintenance of habitats. These aspects are among the reasons the sampling effort is concentrated in these sectors.

The habitat types in the Pesceana River Basin were found to include mixed forests and diverse plant associations (Figure 20), such as Aegopodio podagrariae–Alnetum glutinosae (alder wetlands), Carpino–Fagetum (beech and hornbeam forests), Poetum pratensis (natural meadows), Potentillo micranthae–Quercetum dalechampii (oak forest associations), Quercetum frainetto–cerris (birch and oak forests), Scripetum sylvatici (marsh areas), and semi-natural meadows. The plant associations belong to the following Natura 2000 habitat types: 91E0* Alluvial forests with Alnus glutinosa and Fraxinus excelsior (Alno-Padion, Alnion incanae, Salicion albae); 9130 Asperulo-Fagetum beech forests; 6440 Alluvial meadows of river valleys of the Cnidion dubii; 91Y0 Dacian oak-hornbeam forests; 91M0 Pannonian-Balkanic turkey oak-sessile oak forests; 6430 Hydrophilous tall herb fringe communities of plains and of the montane to alpine levels [164].

In the Pesceana River Basin, the variety of habitat types and plant associations reflects its important ecological diversity, with a combination of forest areas, grasslands, and wetlands, each with unique ecological characteristics and species composition.

• Aegopodio podagrariae–Alnetum glutinosae [165]. This is a specific association of wetlands and riparian areas, dominated by black alder (Alnus glutinosa), which forms meadow forests. This habitat plays a crucial role in regulating soil moisture and preventing riverbank erosion.
• Carpino–Fagetum [165]. Beech (Fagus sylvatica) and hornbeam (Carpinus betulus) forests are found in lower mountain and hilly areas, being typical habitats for well-drained and moist—but not saturated—soils. These forests represent habitats with high biodiversity, providing shelter for a wide range of species of trees, shrubs, and understory plants, as well as for numerous animal species.
• Poetum pratensis [165]. These semi-natural grasslands, rich in species of grasses and wildflowers, are dominated by the grass Poa pratensis. Meadows play a key role in maintaining biodiversity, providing food resources and habitat for pollinating insects and other small organisms, and act as a source of food for domestic animals in the case of grazing areas.
• Potentillo micranthae–Quercetum dalechampii [165]. This association includes mixed oak forests (Quercus dalechampii) with various understory plant species such as Potentilla micrantha.
• Quercetum frainetto–cerris [165]. The plant community of oak (Quercus frainetto) and holm oak (Quercus cerris), characterized by its resilience to drought stress, is particularly adapted to drier soils and warmer climates. These woodlands shape hilly landscapes; in addition to their physical presence, they serve as important refuges for biodiversity in regions where water availability is increasingly seasonal or limited.
• Scripetum sylvatici [165]. This is an association specific to swamp or peatland areas, with plants such as Scirpus sylvaticus (forest sedge) and other species adapted to water-saturated soils.
• Mixed forest. These forests contain a combination of deciduous and coniferous species, and are frequently found in transitional areas between grasslands and pure deciduous or coniferous forests.

4. Discussion

Spatiotemporal analysis is a powerful tool for translating theoretical data into best practices, decision-making processes, and strategic land-use planning [166,167]. Spatial analysis enables the correlation of existing natural resources with land-use patterns within physically delimited territories. This approach facilitates the production of maps that integrate climatology, hydrology, soil fertility, and LULC dynamics.

Leveraging GIS techniques, including remote sensing and Landsat 7, 8, and 9 satellite imagery, provides a deeper understanding of the spatiotemporal dimensions of the Pesceana River Basin [168,169,170,171,172,173,174,175,176]. Satellite imagery can provide extremely useful information in a cost-effective and timely manner [177]. For example, in the Pesceana River Basin, spatiotemporal analysis of NDVI data derived from high temporal resolution satellite imagery such as Landsat 7, Landsat 8, and Landsat 9 is an effective tool for monitoring or quantifying vegetation cover dynamics [178,179,180].

LULC maps (e.g., those from the CLC database) present various land cover types, such as deciduous forests, grasslands, wetlands, urban areas, industrial areas, natural grasslands, land mainly occupied by agriculture, and non-irrigated arable land. The maps produced in this study can serve various purposes, such as for land-use management [181], natural disaster management and urban planning [182,183,184], ecological and precision agriculture [185], forestry [186], the technical–urban development of administrative territorial units and construction infrastructure [187], and addressing climate change issues at regional and local levels [188].

Geographical factors also play a critical role in urban expansion and LU dynamics [189], including altitude [190,191,192,193], slope [194,195,196,197], and proximity to water bodies [198,199,200,201,202,203,204]. Beyond routine mapping, land-use analysis must consider specific geomorphological features [205] and a wide range of environmental risks. These risks—from flooding to subsidence, rockfalls, and land instability [206,207,208]—pose significant challenges to the safe use of land.

Ultimately, land use is not merely a spatial category but a fundamental pillar of socioeconomic progress. Its management is inextricably linked to local environmental conditions and the success of long-term sustainable development strategies [209,210], requiring a delicate balance between temporal planning and landscape resilience. Romania, with its diverse physiographic and landscape characteristics covering approximately 240,000 km² [211,212], exhibits a complex LU structure, with a balance between agricultural, forestry, urban, and environmentally protected areas [213,214]. This spatial-temporal configuration is influenced by relief diversity, natural resources, population density, and policies at local, regional, national, and European levels [215,216,217,218,219,220,221,222]. Political changes at these scales have also affected land-use policy, as evidenced by the Central Bărăgan Plain, where long-term changes are evident in the fragmentation of cultivated land [223].

According to specialized studies, physical–geographical, landscape, geomorphological, climatic, hydrographic, pedological, and anthropogenic factors play a significant role in determining land degradation susceptibility across Romania [212,224]. The spatiotemporal distribution of LULC types is dominated by agricultural lands, which cover approximately 61% of the country and are primarily concentrated on plains and low-altitude plateaus. In contrast, meadows and natural forests are predominantly found in hilly and higher mountain regions [225]. Nearly 40% of Romania’s territory is arable, a figure of particular relevance given the extensive exposure of degraded lands, which are highly fragile and vulnerable to further environmental pressures [212,226].

Specialized studies also confirm the viability of the results obtained in this research. Srejić et al. [227] stated that the Velika Morava basin is characterized by intensive agriculture, which further intensifies soil erosion in the Velika Morava river basin; this is also true in this study, where agricultural areas have increased after 1990. Shukla et al. (2023) used GIS spatial data that included DEM, soil data, watercourse network, and land use [228], which were demonstrated in this study. Based on the DEM, two complex processes were performed regarding slope and slope orientation in the Pesceana river basin. Jujea et al. (2023), in the case of the upper Bistrița basin [229], the last decade has been characterized by a diversity of changes between different land cover categories, which was demonstrated due to the implementation of technology and remote sensing by increasing the accuracy of the satellite images used in this study, such as Landsat 7, 8 and 9. Similar research on the impact of artificial surfaces on the environment is negative [230], as they are directly responsible for the artificial landscape, as in the present study, where the discontinuous urban and industrial areas are directly responsible for the degradation of the ecosystem. Moreover, the results of this study align with the research carried out by Ulloa-Torrealba et al. [231], which indicates a Kappa coefficient score of over 80%, indicating a very good accuracy of the land-use classification and a high viability of the results of the land-use change analysis in the Pesceana River Basin, Romania.

Based on the NDVI data presented in Figure 7a–c, several observations can be made. In 2000, approximately 87% of the basin was covered by medium- to very-dense vegetation, with only small areas classified as sparse or non-vegetated. By 2014, very dense and dense vegetation expanded to cover 94.15% of the area, indicating a substantial increase in vegetation density and a corresponding reduction in medium and sparse vegetation categories. Between 2014 and 2025, a significant redistribution of vegetation density classes occurred. While the “dense vegetation” category decreased slightly, both the “medium” and “very dense vegetation” categories increased. Areas without vegetation or with only sparse vegetation cover, on the other hand, remained negligible. Overall, these patterns indicate a stable, basin-wide dominance of dense vegetation cover, which persisted throughout the entire study period.

By integrating LULC maps with NDVI dynamics and terrain indices—in particular SPI, TWI, and slope—this study provides a framework for local authorities to make informed decisions. These data layers are essential for prioritizing conservation areas and managing the ongoing expansion of agricultural land. Ultimately, such information enables land-use interventions that are not only efficient but also sustainable in the long term.

The expansion or persistence of forest cover contributes to enhanced water retention capacity, reduced surface runoff, and improved slope stability, which are particularly relevant in the context of flood prevention and erosion control. Conversely, the intensification of agricultural land in low-slope areas may increase surface runoff and sediment transport if not managed appropriately. By integrating vegetation, topography, and hydrological indices, the proposed framework enables a spatially explicit interpretation of these ecosystem service dynamics, which is relevant for civil protection planning and natural risk management.

The results of the field research highlight six types of habitats of European conservation interest in the Pesceana River Basin (Figure 21), and of these, the most important for conservation are the Alnus glutinosa alluvial forests (vegetative assemblage Aegopodio podagrariae–Alnetum glutinosae), alder and hornbeam forests (vegetative assemblage Quercetum frainetto–cerris), and mixed forests of alder and beech with hornbeam (vegetative assemblage Carpino betuli–Fagetum sylvaticae). As such, we propose including the following areas in the protection regime to preserve the forest habitats of community interest: (1) Dealurile Crețenilor, (2) Făgetele de la Pesceana, and (3) Arinișurile de pe Valea Pescenei—Vâlcea.

The proposals aim to preserve valuable forest formations from a phytosociological, ecological, and biodiversity point of view, contributing to maintaining the integrity of natural habitats and to the objectives of the Natura 2000 network. To maintain the integrity of the areas proposed for protection, it is recommended to implement an adaptive management plan, which should include periodic biodiversity monitoring, invasive alien plant species control, and forestry exploitation limits near the core area. Implementing a protection status for these areas would contribute to achieving the objectives of the European Union Biodiversity Strategy 2030, which provides for the protection of at least 30% of the terrestrial surface. At the same time, maintaining these forest habitats in a good state of conservation offers ecosystem benefits such as regulating the hydrological regime and conserving local genetic resources.

The exploitation of the fairly high fertility potential of the soils within the Pesceana river basin necessarily requires intervention with hydro-improvement works aimed, firstly, at protecting against floods caused by river waters and, secondly, at eliminating excess moisture caused by other factors. Thus, based on the study, we propose the following remedial policies to local and regional authorities: damming of areas that become economically profitable; dams to be sized appropriately to ensure against winter and autumn floods; drainage of the dammed areas to avoid excess moisture through drainage channels and pumping stations; irrigation of the lands within the dammed areas, given the deficient rainfall regime in the south of the basin; construction of reversible pumping stations for both drainage and irrigation; organization of the protected territory to allow for mechanized, intensive agriculture with higher economic efficiency.

Appropriate land-use and ecosystem management measures: deforestation should be prohibited or at least limited/controlled through local reforestation; prohibition of uncontrolled construction on conservation habitats and arable land; agricultural crops should not expand to the detriment of deforestation; construction of dams against floods that can affect not only conservation habitats, cereal crop areas, but also the population.

A full cost–benefit analysis was beyond the immediate scope of this research; however, the GIS datasets we developed serve as a necessary basis for future economic modeling. Beyond simple data, these resources allow local authorities to project investment needs and assess environmental gains against socio-economic constraints. The spatial results provide a preliminary framework for evaluating the potential economic implications of land-use change. For example, areas characterized by high erosion susceptibility (SPI) may imply increased costs related to soil degradation, infrastructure maintenance, or flood damage, whereas zones suitable for reforestation or sustainable agricultural practices may generate long-term benefits through ecosystem services such as water regulation and soil conservation. In this sense, the proposed GIS-based framework can support local authorities in prioritizing investments by spatially identifying areas where preventive measures may reduce future environmental and economic costs. In essence, this GIS framework serves as a bridge, moving from simple land-use monitoring to practical, actionable planning. By combining spatial data with environmental variables, the methodology provides a more robust approach to risk reduction and sustainable land management at the local level.

The present study adopts a basin-scale, integrative approach that combines land-use dynamics with terrain-based and hydrological indices, as well as field-derived habitat data. This multi-layer integration allows for a spatial interpretation of environmental constraints and management priorities in small river basins, which are often underrepresented in large-scale analyses. The added value of the approach lies not in the novelty of individual methods, but in their combined application within a coherent spatial decision-support framework tailored to local-scale planning needs.

5. Conclusions

Although the study area is not officially designated as a Natura 2000 site or as a member of the Romanian Protected Areas Network (RONPA), our field surveys revealed several habitats of significant conservation value. Of particular importance are the priority habitats identified under codes 91E0* and 6240* (Annex I of the Habitats Directive). The presence of these specific zones suggests that parts of the Pesceana River basin meet the criteria for future designation as a Natura 2000 site, highlighting an ecological potential that has not yet been formally recognized [232].

The maps show a sustained decline in deciduous forest area between 1990 and 2018 based on long-term spatial trends. This forest decline is largely attributable to deforestation and the conversion of forest areas to agricultural land. Over the same period of almost three decades, complex cropping systems and diverse agricultural land use steadily increased, indicating an intensification of agriculture throughout the catchment area. Although discontinuous urban expansion and industrial growth were also evident, they lagged behind the more comprehensive shift toward agriculture. These dynamics underscore a profound socio-economic transformation in which urban-industrial development and agriculture increasingly encroached upon the catchment’s natural forest areas.

Ultimately, the Pesceana River basin functions as a mosaic of diverse habitats—from wetlands and wet forests to dry grasslands—that are crucial for regional ecological stability. This biodiversity not only provides habitat for local flora and fauna but also acts as a natural buffer against soil erosion and plays a vital role in local water regulation. To demonstrate the originality and performance of the study, we integrated several spatial layers: soil sampling points, vegetation associations, and a four-category SPI classification for erosion risk. This process resulted in a series of maps that capture the natural potential and land-use vulnerability of the Pesceana River Basin in great detail. By directly correlating these SPI values with local soil and vegetation data, the model identifies erosion risks from high to very low—effectively transforming descriptive GIS monitoring into a replicable tool. Ultimately, this shifts the focus from simple mapping to providing actionable spatial intelligence needed for local governance and sustainable development.

While the present analysis provides a solid quantitative foundation, we acknowledge that relying solely on spatial and numerical datasets has its limitations. To capture the full complexity of land-use dynamics, future studies would benefit from integrating these objective findings with qualitative, sociodemographic insights. Collecting primary data through fieldwork or structured questionnaires would allow for a more nuanced perspective on the human factors driving land-use changes—aspects that may not be fully represented by quantitative modeling alone. A second limitation concerns the current statistical field data; future research will include additional parameters, such as temperature, humidity, and pH, which are important for developing floristic and faunal biodiversity.

Planned future research directions include (1) multi-scenario predictions, for example, the Land Change Modeler (LCM) for the year 2030, which predicts the change in the number of land types based on the Markov transition matrix, where the spatio-temporal distribution is based on the transition probability [233]; (2) Markov modeling, which is most often used in monitoring, simulating changes, and LULC trends, but can also be used to predict the magnitude of land-use change and the stability of future land development [234,235]; and (3) investigating the land-use transition matrix. This is important for analyzing the dynamic transformation of land types within a river basin and a country, as it quantifies transitions between land-use types over a given period [236,237].

The Stream Power Index (SPI) is primarily controlled by slope and discharge flow energy. In valley areas, the slope is low, the specific flow energy is low, and the dominant processes are sediment transport and deposition rather than active erosion [238,239,240]. Therefore, these sectors exhibit low SPI values. Because a significant portion of our soil sampling sites is situated within these low-gradient zones, the majority of the samples correspond to the ‘low’ and ‘very low’ SPI classes.

This study contributes to spatial analysis by introducing an integrated GIS-based framework that synthesizes terrain and hydrological indices (SPI, TWI, slope, and relief energy) to evaluate natural potential in small-scale basins. In contrast to previous descriptive mapping efforts, this approach prioritizes the analytical interpretation of spatial relationships. The result is a replicable model specifically designed for environmental assessment and strategic land-use planning. This approach allows for the relative visualization of habitat distribution and influence, without claiming that the final raster provides exact ecological boundaries or probabilities of presence. The advantage of the IDW method in this context is that it produces a spatial generalization of habitat presence, taking into account proximity to observation points, which is appropriate when the main goal is to identify areas with dominance of a particular habitat. It is important to emphasize that the method has limitations. IDW does not account for ecological processes such as geomorphology, hydrology, or human impact; it is sensitive to the distribution of observation points, and the final values can be affected by arbitrarily chosen parameters, such as weighting or the search radius. At the same time, the fact that a habitat dominates a pixel does not exclude the presence of other habitats in the same location; it only indicates the strongest relative influence. In conclusion, the use of IDW on binary columns h1–h6 generated continuous habitat influence surfaces, which were subsequently synthesized into a dominance raster, providing a visual representation of the areas where each habitat has the strongest influence.

Policy recommendations for authorities: (1) aligning the Sustainable Development Goals (SDGs) with the application of zoning regulations that protect critical ecosystems, while ensuring responsible development in the medium and long term, (2) using GIS and remote sensing technology that can improve the productivity and sustainability of appropriate land-use management, monitoring land-use changes, but also predicting the impact of land use on the environment in the Pesceana River Basin, (3) protecting soil quality and preventing its alteration through water ponds in the low-lying southern areas of the basin, such that the soils can support sustainable food, (4) the need for a batter management plan at the county and regional level against puddles, water retention, water volume that depends on the flow regime in different seasons, (5) organizing training courses for urban planners, cadastres on sustainable land-use practices, (6) developing local partnerships between the public and private sectors to implement sustainable land-use projects and finance agricultural crops, and (7) involving universities in internships on agricultural farms. Thus, by recommending these policies, we aim to ensure that medium- and long-term land-use management serves as a link between effectiveness, sustainability, and the responsibility of local and regional communities for the resilience of economic development within the Pesceana River Basin.