Modelling Water Erosion and Mass Movements (Wet) by Using GIS-Based Multi-Hazard Susceptibility Assessment Approaches: A Case Study—Kratovska Reka Catchment (North Macedonia)

Abstract

Kratovska Reka is a short (17.3 km) left tributary of Kriva Reka, whose watershed (68.5 km²) is located on the northwestern slopes of the Osogovo Mountains (North Macedonia). Due to the favorable natural conditions and anthropogenic factors, the Kratovska Reka catchment is under a high risk of natural hazards, especially water erosion and landslide occurrences. For this reason, the paper presents an approach of modelling of potential erosion and areas susceptible to the above-mentioned hydro-meteorological hazards in the Kratovska River catchment. Firstly, this study analyzed the main geographical features that contribute to intensive erosion processes in the area. Then, using the Gavrilović EPM erosion potential method, an average value of 0.56 was obtained for the erosion coefficient Z, indicating areas prone to high erosion risk. Furthermore, by using landslide susceptibility analysis (LSA), terrains susceptible to landslides were identified. The results shows that 1/3 of the catchment is very susceptible to mass movements in wet conditions (landslides). According to the combined multi-hazard model, 3.13% of the total area of the Kratovska River catchment is both at high risk of landslides and under severe erosion. The Kratovska River catchment is significantly endangered by the excessive water erosion processes (39.86%), especially on the steep valley sides, i.e., terrains that are completely exposed, under sparse vegetation, and open to the effects of distribution/concentration of the rainfall amounts throughout the year. Identifying locations with the highest erosion risk serves as the initial step in defining and implementing appropriate mitigation measures across local and regional scales, thus enhancing overall resilience to environmental challenges.

1. Introduction

Phenomena or events that degrade the quality of the environment and have a potential for endangering people, property, and infrastructure are called geohazards [1]. Based on the literature findings, geohazards have a significant impact on human health and economic development, leading to consequences such as human casualties [2,3,4,5], direct damage to infrastructure or private property [6,7], a negative impact on household income [8,9], alteration of consumer behavior among various income groups [10,11], and lower life satisfaction [12,13,14]. For the most part, geohazards are generated as a result of a combination of unfavorable natural factors and, in certain situations, they can also be caused by humans. Refs. [15,16], through examples, explain that geohazards have a negative impact on geosites.

Until a few decades ago, natural and geohazards were considered to be typically natural, i.e., occurring naturally. However, recently, it has been realized that anthropogenic factors have an increasing direct or indirect impact on their occurrence and frequency [17]. That is why the expression “anthropogenic–natural disasters” or “anthropo-geohazards” is used more frequently worldwide today. As noted by [18], during the last several decades, Europe has predominantly experienced hydro-meteorological hazards that frequently give rise to diverse risk factors, including environmental changes, population dynamics, infrastructure vulnerabilities, and socio-economic aspects [19,20,21].

In terms of climate-related risks, South-Eastern Europe (SEE) could be identified as one of the world’s critical areas [22]. This geographical area has been known for its numerous hydro-meteorological hazards throughout modern history [23]. These undesirable natural events often cause environmental, social and, therefore, economic damage, with long-term consequences, and can easily turn into disasters with catastrophically large effects. The severity of economic damage can be illustrated by the reported loss of over EUR 433 billion for the period 1980–2015 in the European Economic Area [24]. With an evidently increasing number of extreme weather events induced by climate change, hydro-meteorological hazards have emerged as a high-impact risk, especially in Europe [25,26,27]. Contemporary research in the field [24] has presented unfortunate facts related to hydro-meteorological hazards in Europe. All major hazardous events (heavy precipitation episodes/floods, storms, landslides/rockfalls, wildfires, droughts, extreme temperatures) have increased in frequency and/or intensity. Projected scenarios for the future are not optimistic, since even more natural hazard-related damage is expected for various sectors [25]. During these hazardous events, humans, and anthropogenic systems in general, are highly exposed and vulnerable [28]. Climate and weather conditions are not the only reason for this. As the global population continues to experience a rising trend (estimated to be 11 billion by the end of this century), (United Nations-UN), accompanied by urbanization of natural disaster-prone zones [29], society development and economic growth [30], the amplification of natural hazards impact is imminent. This socio-economic development has also led to tourism expansion, thus increasing its sensitivity to weather-related hazards [18].

In the last few decades, there has been great interest in analyzing and assessing the risk of natural hazards and geohazards. The huge interest is the result of a large number of casualties and the damage caused by these natural hazards around the world [31,32,33,34,35,36,37]. According to [38], the acceleration of the process of soil erosion can cause a potential problem for agriculture. Soil erosion may be doubled with an increase in slope and relative angle of terrain [39,40,41]. As pointed out by [42], the Western Balkans region is highly prone to water erosion processes and, therefore, the estimation of rainfall erosivity potential is essential for understanding the complex relationships between hydro-meteorological factors, soil erosion processes and main topographical features.

According to [2,43], this has led to the development of various simulation models of different accuracy and complexity, including empirical, stochastic, and deterministic models. Additionally, indicators based on biological, physical, social, and economic approaches have also been utilized. Empirical models are commonly used, particularly in countries where input data are often limited, and their validity is uncertain, such as Western Balkan and southeastern European countries. These models are preferred due to their simplicity and ease of use, having less input data and fewer computations, compared to more comprehensive models [43].

Ref. [43] conducted a study evaluating soil erosion and its spatial distribution using two empirical models, namely the erosion potential model (EPM), also known as the Gavrilović method, and the RUSLE model. The study focused on the Venetikos River catchment, the largest tributary of Aliakmonas River, located in northern Greece. Both models demonstrated a satisfactory simulation of the phenomenon, exhibiting acceptable precision and enabling the identification of areas most susceptible to erosion and land degradation. Although both models performed quite similarly and attributed underestimated results in comparison to the “actual” (measured) values of mean annual sediment discharge and yield, the EPM approach was used in the study to generate comparable data with neighboring countries where the same approach was applied [44,45,46].

The EPM model has been extensively implemented across the Balkans and in other countries worldwide, including Serbia, Croatia, Slovenia, Germany, Italy, Argentina, Belgium, and Greece [43,44,45,46,47,48]. Its application has provided reliable results for assessing water erosion severity, estimating mean annual soil loss and sediment yield, and implementing erosion control measures and torrent regulation at a regional scale.

Due to very favorable natural conditions as well as the human impact over the centuries, the territory of North Macedonia is highly exposed to various natural hazards [17,49], and geohazards [50]. Among the most dominant geohazards in this area are excessive erosion, torrents, landslides, rockfalls, floods, etc. [49]. Such phenomena, to a lesser extent, have also been recorded in the Kratovska Reka catchment, which is the case study in this paper. The main reasons for this are steep slopes and extensive areas covered with erodible crystalline rocks and tuffs, combined with high human impact and deforestation occurring over centuries. For those reasons, there are many sites in the catchment with severe erosion, especially along the banks and steep valley sides, with occasional threats to property and people’s lives. Landslides are frequent as well, especially those that cause material damage, and destruction of roads (for example, the Kratovo–-Probištip road), buildings, etc.

Only a few studies of geohazards in this area have previously been carried out [50,51,52,53,54]. However, most of these studies cover larger (regional to national) areas and only briefly mention the high erosivity and landslide susceptibility in the Kriva Reka catchment, without detailed analyses.

Because of the worldwide increased threat of geohazards, new methods are being developed to access or to perform a zonation of endangered areas, especially considering landslides, flashfloods, excess erosion, etc.

According to UNDRR, the multi-hazard concept refers to the selection of multiple major hazards that an area faces and the specific contexts in which hazardous events may occur simultaneously, cascadingly, or cumulatively over time, taking into account the potential interrelated effects.

Multi-hazard techniques are very important for the analysis of hazardous events [55,56,57,58,59,60]. Another approach could be achieved by implementing probabilistic and deterministic stochastic processes [61].

Accordingly, there is a need to develop models for the potential erosion and landslide susceptibility areas in the Kratovska Reka catchment. Therefore, the main objectives of this study are to: (1) perform multi-hazard assessment (hydro-meteorological hazards) in the study area by applying a modified erosion assessment model and joint systematic approaches; (2) produce hazard risk and susceptibility maps, and (3) identify places of greatest erosion risks as the starting point for defining and implementing suitable mitigation measures.

2. Data and Methodology

2.1. Study Area (Kratovska Reka Catchment, North Macedonia)

Kratovska Reka flows in the northeastern part of North Macedonia and passes through the city of Kratovo, where it is arranged with a concrete channel (built between 2015 and 2020) (Figure 1). The river source is on Osogovo Mountain at 1366 m (Lisec locality), while the inflow in Kriva Reka is at 412 m, thus, the total fall is 954 m. The length of the river is 17.3 km, and the catchment area covers 68.5 km².

The valley of Kratovska Reka is generally incised in volcanic rocks that dominate this area. The gorge-like valley has an average depth of 400–500 m and steep sides.

The lithology of the Kratovska Reka catchment is quite complex, represented mainly by Precambrian and Paleozoic crystalline rocks that are intruded and covered by younger intrusive and effusive magmatites [62] (Figure 2, left side). Overall, rock masses with high erodibility (tuffs, shists, clastic sediments, breccia, etc.) dominate, at 51.7%. They are easily susceptible to weathering, decomposition, and washing, which is the reason for the strong erosion processes in the Kratovska Reka catchment.

In the Kratovska Reka catchment, soil types that are the result of strong water–mechanical erosion and accumulation are widespread, especially on steep, hilly terrains with sparse vegetation. According to the National Soil Map at a scale of 200 k [63], the study area is primarily dominated by cambisols, leptosols, regosols, and vertisols (

Table 1. Main physical properties of the soils in the Kratovska Reka catchment (according to [64]).

WRB Classification of the Soil Type	Soil Texture Type	Clay Content in %	Structure	CaCO3 Content in %
Cambisol (brown forest soil)	Sandy loam or finer	8	ABC	0
Leptosol	Shallow soils lacking well-defined horizons	0	(A)-R	0
Regosol	Shallow, medium—to fine-textured	15	(A)-C	13.44%
Vertisol	Clay-rich soils	65	A-AC-C-(R)	0
Ranker	Lithomorphic soils	5	A-AC-C-R	0
Fluvisol	Heavy clays in the catchment area	10	(A)-(Ah)-C	0

).

These soil types are relatively shallow and predominantly cover volcanic rocks and schists in the underlying strata. However, it is important to acknowledge the limited availability of soil data and soil profiles for the catchment area, resulting in a soil map of poor quality for this region. Through field observations, it becomes apparent that the soils in the hilly–mountainous terrains are weakly developed and subject to significant erosion. In the narrow valley bottom of Kratovska Reka, downstream of Kratovo, fluvisols are also present. These soils are primarily sandy and contain a notable proportion of river sediments, such as gravel and pebbles.

In terms of topography, the catchment elevation ranges from 412 m to 1550 m, while 66.1% of the area is below 1000 m, where the human impact is strongest. However, the terrain slope has the highest impact on the erosion and landslide processes in the Kratovska Reka catchment. According to the 15 m DEM, 15.6% of the catchment has steep slopes (>30°), while the mean slope is 20.9°. The plains or terrain with slopes bellow 3° cover only 1.3%, showing that the entire area is highly inclined. Hillslopes generally have western, and south-western (adjacent) aspects, which favorably reflect on the erosion intensity. In the Kratovska Reka valley, it has been observed that the weathering of dacite–andesite rocks occur predominantly on the southern or western exposures, and much less on the northern sides.

Terrain relief is in the range of 34.3–243.1 m/km², with a mean value of 105.4 m/km². Terrain with an elevation difference of between 150–300 m/km² covers 8.3% of the catchment. This means that, the higher the topographic ruggedness index, the greater the danger of geohazards (erosion, landslides) (Figure 3).

According to [65], the climate is generally moderate–continental with an average annual temperature of 11.4 °C, ranging from a mean of 21.4 °C in August to 0.4 °C in January. The average annual precipitation is relatively low (728.4 mm), with the maximums in May (84.9 mm), and November (69.6 mm), and the minimum in August (35.1 mm). Of the total average annual amount, 57% falls in the vegetation period, 28% in the spring months, 22% in the summer, 24% in the autumn, and 26% in the winter. Of the total annual average number of days with precipitation (97), only 9% have a daily amount equal to or greater than 20.0 mm, reaching up to 110 mm/day [58]. During winter, the upper section of the catchment is typically covered in snow, which can swiftly melt as spring arrives. Intense rainfall, prolonged precipitation episodes, and rapid snowmelt contribute to significant overland flow and severe erosion in exposed or poorly protected areas. The main features of the climate are presented in Figure 4.

Closely connected with the previous factors, especially climate, is the hydrography of the area, with the most important hydrographic features presented in

Table 2. Characteristics of the river network in the Kratovska Reka catchment (P = area, N = number of tributaries, L = length of all tributaries, D = density of river network in km/km²), Hmm = average annual precipitation, Qsr = average flow, H-a m = mean height of the catchment, Hvl m = estuary altitude, ΔHs-v = mean height difference of the catchment, Lc = length of the catchment, Lw = length of the watershed, α^o av. = average drop of the watershed, AC = coefficient of asymmetry).

Watershed	P km2	N	L km	N/P	D	Hmm	Qsr m3	H-a m	Hvl m	ΔHs-v	Lc km	Lw km	αo av.	AC
Kratovska Reka	68.5	20	98.4	0.5	2.0	701	0.6	879	406	473	17.3	17.4	21.0	1.5

.

The presence or absence of vegetation cover is a crucial factor contributing to erosion processes in the investigated area. According to the [66] CORINE Land Cover 2018 (CLC2018) map, the largest area of the catchment is covered by discontinuous urban fabric (36.3%). Non-irrigated arable land occupies 20.7%, pastures 15.9%, and complex cultivation patterns 15.7% of the total area. Forests in the area are predominantly situated at altitudes above 1000 m, with broad-leaved forest, mixed forest, transitional woodland–shrub, and bare soils below. The CLC2018 model shows a significant reduction in forest areas compared to CLC2000, probably due to logging.

The Kratovska Reka catchment completely belongs to the municipality of Kratovo. According to the last census of 2021 [67], the municipality of Kratovo has 7545 inhabitants, which is significantly less than the population from the census in 2002 (10,441 inhabitants). Thus, this is a typical depopulation area. In this sense, in the past centuries, there has been a much higher human impact on the landscape, not only by the larger population (concentrated in the hilly villages), but also by extensive agricultural activities.

2.2. Methodology for Erosion Assessment

For erosion assessment, the erosion potential model (EPM), also known as the Gavrilović method, is widely used in the majority of southeastern European countries, both at regional and national scales. This model was developed in the 1960s, based on erosion field research conducted in the Morava River catchment area (Serbia). The model includes erosion mapping, estimation of sediment quantities, and classification of torrential flooding [68]. As reported by [69] this method is convenient for areas with limited data derived from previous erosion research, but it does not extensively delve into the underlying physics of the erosion processes.

In North Macedonia, similarly to other countries in the region, the estimation of average erosion potential and sediment yield is commonly accomplished using the EPM [70]. The method is based on the following equation:

Wy = T ∙ H ∙ 3.14 ∙ sqrtZ³ ∙ f

where: W is the average annual amount of erosive material in m³; T is a temperature coefficient in the form: T = (0.1 ∙ t + 0.1) 0.5, where t is the annual mean air temperature; H indicates annual precipitation in mm; Z is the erosion coefficient ranging from 0.1 to 1.5 and above; and f is the studied area in km². Among these factors, the coefficient Z has the highest importance in combining rock erodibility/erosion resistance (Y), land cover index (Xa), index of visible erosion processes (φ), and mean slope of the catchment (J) in the ratio:

Z = Y ∙ Xa ∙ (φ + sqrtJ^0.5)

According to the available data, in our GIS-based approach, appropriate digital layers are prepared, and the calculation of the erosion potential model (EPM) (e.g., [71,72], see Figure 5) involves the following procedures:

• Processing Landsat 8 satellite images (

  Table 3. Different spectral channels of Landsat 8 satellite images.

  Band	Wavelength (φm)	Resolution (m)
  Band 1	0.43–0.45	30
  Band 2	0.45–0.51	30
  Band 3	0.53–0.59	30
  Band 4	0.64–0.67	30
  Band 5	0.85–0.88	30
  Band 6	1.57–1.65	30
  Band 7	2.11–2.29	30

  ) from 2010 and 2020 (cloud-free images are used from the months of May (vegetation period), August (dry period) and October (the fall)), which includes supervised classification into six classes (populated areas, water bodies, surface mines, forests, low vegetation, industry, and transportation). This step results in obtaining vector maps of land use, used for the parameter Xa;
• Treatment of land maps based on the available soil folders, considering the recent situation of soil types and supplementing them with real-time information derived from the processing of Landsat 8 satellite images. This includes supervised classification and the separation of surface open pits, which are considered as areas of disturbed soil layers. This step provides the parameter Y;
• Determination of visible erosion processes through the analysis of orthophoto images, contributing to the calculation of the erosion coefficient φ;
• Creation of a digital terrain model (DTM) according to [73], where DTM serves as the basis for calculating the slope factor J_a.

Z values typically range between 0 (no erosion) and 1.5 (higher amount of erosion). Unlike traditional cartographic tools, the GIS approach in erosion potential model (EPM) implementation relies heavily on deriving most of its model parameters from the digital elevation model and satellite images [71,74,75,76,77,78,79].

Thus, for the Y coefficient, a previously prepared and digitized 100 k geological and soil map was used and then rasterized to 15 m. Notably, these maps were the most detailed available for the survey area and were also considered highly accurate. The given raster values correspond to the rock erodibility and erosion resistance values, according to the method proposed by Gavrilović [68]. Generally, the values range from 0.1 (very resistant rocks) to 2.0 (non-resistant rocks) [50]. However, due to the challenges associated with accurately estimating erodibility correlations [80], the fitting of the values is accomplished through a process known as double rooting in the form: Y = sqrt (Y1). The land cover index, X*a, is prepared from the CORINE Land Cover model (CLC2018), with values ranging from 0.1 (dense forests) to 1.0 (bare rocks). Relevant values are incorporated into the CORINE Land Cover (CLC) classes based on the suggested values from the original model [50].

In order to determine the value of the coefficient φ for visible erosion processes, the traditional model’s subjective assessment is replaced with the utilization of Landsat 8 band 4 (b4-red). This involves dividing the grayscale values (ranging from 0 to 255) by 255. This is because this specific band consists of 255 shades of grey, where lower values indicate areas without visible erosion processes, while values closer to 255 indicate areas with excessive sheet, rill, and gully erosion. However, higher values may also indicate the presence of light anthropogenic structures, uncovered rocks, waste disposal sites etc. [50,70,79]. To address this issue, a correction is applied using the slope (a-in degree) in the form: φ = ((b4/255)∗log(a + 1)), which leads to significantly more accurate values for the coefficient φ.

The slope factor (J) is computed using the available 15 m digital elevation model (DEM) as a raster layer representing the slope angle in radians (a = a/57.3). Following these steps, the GIS-calibrated coefficient Z is calculated using the equation:

Z = sqrt(Y)*φ*((X*a + φ)*log(a +1) + sqrt(a/57.3)).

The main climate parameters in the model, such as the air temperature coefficient (T) and average annual precipitation (H), are determined using ERA5 average monthly data for precipitation and air temperature, as well as MODIS land surface temperature (LST). These data are accessed and exported through the Google Earth Engine (GEE) script, covering the period from 1 January 2000 to 31 December 2022. The resulting raster models (grids) of average air temperatures and precipitation for the Kratovska Reka catchment serve as the basis for the T and H coefficients in the erosion potential model (EPM). Simultaneously, the accuracy of both models is verified using real measurement data from the Kratovo meteorological station, demonstrating a high level of accuracy, at 98.5%.

With the availability of air temperature and precipitation models, along with an erosion coefficient (Z) model, the mean annual erosion loss (Wy) for the entire catchment is calculated.

The GIS-based EPM approach has been previously tested in various regions and watersheds in North Macedonia [70,81], as well as in other countries [44,80,82,83,84,85,86,87,88,89,90]. Evaluation of EPM accuracy is generally performed by comparing the results with measuring the quantity the sediment deposition in the reservoirs [85,88,90], demonstrating strong correlations with both, measured and observed data.

2.3. Methodology for Landslide Susceptibility Assessment (LSA)

Modelling and mapping landslide-prone areas on a regional scale is a very complex task, because of many natural and anthropogenic factors related to landslide processes. For that reason, the results of landslide susceptibility assessment (LSA) from previous research on similar sized (test) areas were also included [49,91,92]. Furthermore, the work of [93] highlights the importance of evaluating six key triggering factors that significantly impact landslide activity. These factors include lithology, slope angle, land cover, terrain curvature, distance from rivers, and distance from roads.

The identification of influencing factors was the basis of several methods of LSA [94]. In a related study [95], these factors were categorized into two main groups, internal and external factors. The internal category encompasses mechanisms occurring within the mass that lead to a decrease in its shear strength below the external forces exerted on the mass by its surroundings, thereby inducing failure. On the other hand, the external category comprises mechanisms external to the mass that surpass its internal shear strength, consequently resulting in failure. According to [96], depending on the specific characteristics of the study area, it is essential to include at least three factors in GIS analysis, topography, lithology, and land use. These factors play a crucial role in assessing and modelling various aspects of the study area’s multi-hazard susceptibility to specific processes. According to [97], the most frequently encountered conditioning factors are lithological units, tectonic features, slope angle, proximity to road or drainage networks, land cover, and rainfall distribution (concentration). Nonetheless, it is worth noting that there may be other factors that could be equally influential in certain contexts, even though they may not be as commonly considered.

In this research, six causative factors were taken into consideration: lithology, slope, plan curvature, land use, distance from streams, and distance from roads (

Table 4. Spatial data layers used in this study.

Factor-Layer	Source	Data Type
Slope, Planar curvature	15-m DEM	Grid (GeoTIFF)
Lithology	Digital geologic map	Grid (GeoTIFF)
Land use	CLC-2018	Grid (GeoTIFF)
Distance from stream	Topographic vector database	Line (shp)
Distance from roads	OSM roads database	Line (shp)

; Figure 6). The selection of these parameters was based on the nature of the study area, the scale of analysis, and data availability [50]. For the slope and plan curvature data, a 15 m digital elevation model (DEM) of the entire catchment was used, and SAGA GIS software was employed to derive these variables. The lithology information was obtained from a digitalized and rasterized geological map of the town of Kratovo, at a scale of 100 k. This lithology map was then prepared within the Kratovska Reka catchment using QGIS software. It included nine lithological units, ranging from Precambrian schist to Cenozoic solid volcanic formations, Miocene dacitic ignimbrites, Pliocene andesites, tuffs, and Quaternary clastic sediments. The land use layer was created based on the CORINE Land Cover (CLC2018) classification hierarchy. Distance from streams was determined using the topographic river network (25 k) provided by the State Agency of Cadaster. Distance from roads was obtained from the OpenStreetMap (OSM) road network in vector (.shp) format. In the following steps, five buffer zones were established for both roads and streams, with increments of 20 m. These buffer zones were then converted into raster format. Finally, all the layers, including lithology, slope, plan curvature, land use, distance from streams, and distance from roads, were converted into a raster grid format with cell sizes of 15 × 15 m.

Following the previous steps, the next crucial stage involves selecting an appropriate LS (landslide susceptibility) method. In cases where the study area is extensive and lacks a comprehensive landslide inventory, statistical analysis of landslides and frequency ratio methods are considered highly effective. This approach relies on examining the correlation between the spatial distribution of landslides and each conditioning parameter. By assessing the relationship between landslides and these parameters, the LS method can be determined and implemented accordingly [98]. In this study, the relationship between landslide occurrence and the selected conditioning factors (such as slope, lithology, land cover, etc.) is assessed through the calculation of the landslide susceptibility index (LSI). The LSI method involves calculating the index for each category within the chosen factors. This allows for the estimation of the susceptibility of each category to landslides based on their respective characteristics. According to [50], the analysis of LSA (landslide sensitivity analysis) is employed to assess the significance of different variables or changes in relation to landslide occurrences. Weighting factors are determined to evaluate the influence of each variable. These weighting factors compare the calculated density of landslides with the overall density of landslides in the study area. This comparison helps determine the relative importance of each variable in relation to landslide susceptibility [99]. It can be expressed as:

$${\mathrm{W}}_{\mathrm{ij}}=1000\left({\mathrm{f}}_{\mathrm{ij}}-\mathrm{f}\right)=1000\left(\frac{{\mathrm{A}}_{\mathrm{ij}}{}^{*}}{{\mathrm{A}}_{\mathrm{ij}}}\cdot \frac{{\mathrm{A}}^{*}}{\mathrm{A}}\right)$$

where: W_ij has high importance for a certain class i of the parameter j; f_ij is the landslide density within class i of parameter j; f is landslide density tho on the whole map; A_ij* is a landslide surface in a certain class i of parameter j; A_ij is the surface of a certain class i of the parameter j; A* is the total area of the landslide on the entire map; A is the total area of the entire map. In the subsequent step, the weights assigned to each variable are combined in an equation to generate a resultant landslide susceptibility index (LSI) map for the study area. By summing up the weights obtained for each variable, the LSI map provides an integrated representation of landslide susceptibility, incorporating the influence of multiple factors. This map helps identify areas with higher or lower susceptibility to landslides, based on the combined effects of the analyzed variables. The weight values of factors used for the LSA model are presented in

Table 5. The weight values of factors used for the LSA model.

Factor	Value	Factor	Value
Lithology		Land coverage
Clastic sediments	5	Broad-leaved forest	0.5
Tuffs	4	Mixed forest	0.5
Schists	3	Transitional woodland–shrub	2
Dacitic ignimbrites	2	Pastures	4
Andesites	1	Complex cultivation patterns	1
Slopes		Land principally occupied by agriculture	1
0–10°	2	Sparsely vegetated areas	1
10–20°	6	Non-irrigated arable land	1.5
12–30°	10	Urban fabric	1.5
30–40°	8	Bare rocks	4
40–50°	4	Streams
Convergence (curvature)		0–20 m	2
Concave	5	>20 m	0
Flat	1.5	Roads
Convex	1.5	0–20 m	2
		>20 m	0

.

In accordance with [1], a similar approach to the previous method is applied for reclassifying the LSI values into distinct susceptibility zones and performing map validation. The total weighted value assigned to each factor is as follows: 30 for slope, 15 for lithology, 10 for land cover, 8 for curvature, and 2 for stream and road distance, resulting in a cumulative total of 67. These weights indicate the relative importance of each factor in triggering landslides within the catchment. The ratio of these weights is 15:7:5:4:1:1, illustrating that slope is the most influential landslide-causal factor, followed by lithology, land cover, planar curvature, and distance to streams and roads, respectively.

Landslide susceptibility is determined by summing the values of all six parameters. This is performed by aggregating the values for each individual grid cell across all six digital layers. The resulting model is then classified into five distinct classes using natural breaks classification in SAGA GIS 9.0.0 and QGIS 3.30.1-’s-Hertogenbosch software. These classes represent different levels of landslide susceptibility, ranging from very low to very high, as outlined in Table 10. The entire process, including the steps, procedures, and quantitative approaches employed in this study, are summarized in the workflow (Figure 7). Furthermore, each key element of the methodology is described in detail, providing a comprehensive understanding of the study’s methodology and analysis.

3. Results

3.1. Erosion Modelling

To calculate the total amount of eroded material, with the help of SAGA GIS, through the previously obtained coefficient of erosion Z (or risk of erosion), an appropriate map was prepared for the Kratovska Reka catchment (Figure 8).

The obtained results in the form of the erosion coefficient (Z) show a significant presence of areas with medium, high, and very high risk (values greater than 0.4), which is 28.79 km² or almost 50% of the Kratovska Reka catchment (

Table 6. Areas in the Kratovska Reka catchment subject to the risk of erosion (according to Z coefficient values).

Class	km2	in %
0–0.4	17.71	25.82
0.4–0.8	31.13	45.39
0.8–1.2	12.91	18.83
1.2–1.6	4.36	6.35
>1.6	2.47	3.61
Total	68.58	100.00

). These areas experience significant soil erosion and sediment transport, even during moderate rainfall events. The land undergoes strong washing and erosion processes, leading to the production, transportation, and accumulation of deposited material. This behavior is particularly pronounced during intense rainfall with rates exceeding 0.5 mm/min or prolonged periods of heavy rainfall episodes. The eroded material accumulates on the cultivated areas, roads, etc. Otherwise, the mean value of the erosion coefficient Z in the Kratovska Reka catchment is 0.56 (

Table 7. Erosion coefficient Z and average annual production of erosive sediment for the Kratovska Reka catchment (P = catchment area; P acc = accumulation area; X*a = coefficient on neatness on the catchment; Koef. Z = coefficient of erosion; W m³ year = total erosion potential; W m³/km² = specific erosion of a single surface).

Watershed	P km2	P acc. km2	X*a	Coef. Z	W m3/year	W m3/km2
Kratovska Reka	68.58	1.16	0.43	0.56	72.709	1063

). Significant factors for such high values of the erosion coefficient are the erodible geological substrate (volcanic tuffs, breccias, slates, Eocene, Miocene, and Pliocene sediments), the bareness–absence of vegetation, and the steep terrain slopes.

Based on the findings of the study, it is evident that areas with the highest erosion potential, as indicated by the Z coefficient values, are primarily located along the steep valley sides of the Kratovska Reka and its tributaries. These areas, particularly those lacking adequate vegetation cover or having only sparse grass vegetation, are particularly vulnerable to erosion processes. Due to the absence of protective vegetation, these exposed areas are more susceptible to erosion and the associated risks (Figure 9).

The model outputs reveal that the average annual sediment production in the study area amounts to 72.709 m³, which is a notably high value (Table 7). The most erosive regions within the catchment are identified as the steep valley sides of the Kratovska Reka and its tributaries. These areas, particularly when devoid of vegetation or covered with sparse grassy vegetation, lack protection against the erosive impact of rainfall. Consequently, they are highly susceptible to erosion processes, leading to the substantial sediment production observed in the model results.

Out of the entire watershed area, a significant portion of 17.35% is classified as being at a very high risk of erosion, surpassing 2000 m³/km²/year (equivalent to a soil loss of 2 mm per year). This represents a considerable magnitude of erosion. In these high-risk areas, excessive erosion takes place, leading to various erosive landforms, loss of valuable fertile land, and the accumulation of a substantial amount of sediment in riverbeds. Specifically, in the Kratovska Reka valley, there is a distinct belt with significantly high values of specific erosion, ranging from approximately 1500 to 3500 m³/km²/year (

Table 8. Areas in the Kratovska Reka catchment according to the amount of erosion.

Class	km2	in %
0–500	16.46	24.01
500–1000	21.39	31.19
1000–1500	11.88	17.33
1500–2000	6.94	10.12
>2000	11.90	17.35
Total	68.58	100.00

). Conversely, in the southeastern, higher parts of the catchment, towards Zletovska Reka, the potential erosion is within normal, natural values below 500 m³/km²/year, thanks to the presence of well-forested areas. To mitigate the degradation and prevent further loss of natural resources, particularly soil and water, it is crucial to implement appropriate preventive conservation measures and activities in areas where the intensity of erosion exceeds 1000 m³/km²/year. These measures are essential for preserving the integrity of the ecosystem and safeguarding the valuable resources present in the region.

In addition to the calculation of eroded material, it is important also to determine the proportion of sediments that reach to the main river (Kriva Reka), as not all sediments exit the Kratovska Reka catchment. Therefore, according to the second part of the EPM approach, the sediment delivery ratio (Ru) is estimated by the following equation

Ru = 4(O*D)*0.5/(L + 10)

where: O = the length of the catchment border (km), and D = the difference between medium altitude and catchment outlet altitude (km). Specific sediment yield (SSY) is estimated as

SSY = W*Ru

According to the preceding equation, the average annual transfer of eroded sediment from Kratovska Reka to the estuary of Kriva Reka amounts to 691 m³/km²/year (

Table 9. Average annual transfer (discharge) of eroded sediment from the Kratovska Reka catchment (Hav = mean height of the catchment; Hmin = estuary altitude; O = length of the catchment border (km); Hw = mean height difference of the watershed; L = length of the watershed (km); Ru = retention coefficient; W = average annual sediment production).

Watershed	Hav	Hmin	O	Hw	L	O/L	Ru	W*Ru	Erosion Rate
Kratovska Reka	879	406	42.13	0.473	17.39	2.42	0.65	47.261	691

). This value represents the average quantity of sediment transported from the Kratovska Reka watershed to the downstream area of Kriva Reka on a yearly basis.

In addition to identifying potential erosion areas, it is also important to identify areas with high sediment deposition. To accomplish this, we utilized the Topographic Wetness Index (TWI) in SAGA GIS, which effectively highlights valley bottoms. In the case of the Kratovska Reka catchment, the TWI values range from 2.8 to 22.1, with values above 12 being most suitable for determining deposition areas (this was also confirmed through detailed field surveys). By applying this criterion, we calculated an accumulation or deposition area of 1.16 km². The majority of the deposition area is located in the Kratovska Reka valley, specifically downstream of the town of Kratovo (see Figure 10). The significant portion of eroded material, which amounts to an average of 25,448 m³/year, is primarily accumulated within this deposition area, particularly during flood events, on top of the fluvisols. The sediment carried by these floods consists of a variety of particle sizes, ranging from pebbles and gravel to sand, silt, and clay (mud), thus leading to alterations in the soil structure. The distribution of sediment types on the floodplain is determined by the energy of the water flow. In close proximity to the river, only the larger-sized material, such as sand, can deposit, due to the strong current. In areas where the flow is relatively lower, silt and fine sand tend to deposit predominantly.

3.2. Landslide Susceptibility Modelling

According to the probability map of the occurrence of landslides in the Kratovska Reka catchment, created with the help of the LSI + AHP natural breaks classification and the prepared

Table 10. Landslide susceptibility areas in the Kratovska Reka catchment according to the natural breaks classification.

Class	km2	%
1	1.05	1.53
2	7.86	11.47
3	22.75	33.17
4	14.28	20.82
5	22.65	33.03
Total	68.58	100.00

, we can determine that class 3 is represented at the highest level (33.17%), with an average probability of occurrence of landslides, on the steep valley sides of the Kratovska Reka and the source parts of its left and valley sides of the right tributaries (Figure 11).

Subsequently, the second and a similar percentage representation (33.03%) is recorded by class 5, with a very high (also the highest) probability of landslides, and with the greatest intensity in the valley of the Kratovska Reka, downstream of the town of Kratovo, and the valleys of the left tributaries. With 20.82%, class 4 is represented, with a high intensity of occurrence of landslides, with the highest representation along the valley and source sides of the right tributaries of the Kratovska Reka, as well as in the lower (western) part of the catchment area, more precisely, downstream at the confluence in Kriva Reka. Class 2, with low intensity of probability of occurrence of landslides, is represented (11.47%) in a very small area, specifically in the catchment of the left tributary of the Kratovska Reka, Latišnica, and around the volcanic mounds in the southern part of the catchment. The first susceptibility class (class 1), representing the smallest and nearly negligible portion (1.53%) of the total probability, corresponds to areas with a very low intensity of landslide occurrence. These areas are predominantly located on the volcanic mounds in the southern part of the study area. The low probability of landslides in this class suggests that these areas are relatively stable and less prone to landslide events, compared to other parts of the region.

A crucial step in LSA is model validation. To achieve this, a comparison was made between the locations of landslides recorded during a field trip (a total of 31) and the LSA zonation. The results of this comparison are defined in

Table 11. Location of the recorded landslides in relation to the LSI classes.

Class	No	%
1	0	0.0
2	1	3.2
3	5	16.1
4	6	19.4
5	19	61.3
Total	31	100.0

.

Thus, out of the total of 31 landslides, 19 (or 61.3%) are located in the very high LSI zone. When combined with the high LSI zone (class 4), the overlap accounts for 80.7% of the recorded landslides, indicating a very high level of accuracy for the implemented model.

In order to assess the overall performance of the LSA model for the study area, additional validation analysis was conducted using the receiver operating characteristics (ROC) curve and the area under the curve (AUC) value [43]. The AUC value characterizes the quality of the probabilistic model by indicating its reliability in predicting the occurrence or non-occurrence of events. A good fit is represented by AUC values ranging from 0.5 to 1, while values below 0.5 indicate a random fit [100]. For a more comprehensive evaluation of the model, in addition to the 31 recorded true-positive landslides (with a value of 1) in the validation dataset, a number of false-positive landslides (with a value of 0) were also selected. It is generally recommended to have 2–3 times more false-positive landslides than true landslides for proper validation [101]. In our case, 90 false-positive landslides were randomly selected as sampling points from the digital elevation model (DEM) using SAGA GIS, followed by careful inspection. It is important to note that the selection of false-positive landslides (0) should be performed cautiously to avoid inadequate or confusing results [54].

The ROC curve and AUC value in this study were calculated using SPSS statistical software and are presented in Figure 12. Based on the calculated data, the AUC value is determined to be 0.856, indicating a good level of accuracy for the model employed.

3.3. Multi-Hazard (Combined Erosion and Landslides) Modelling

According to the study’s objective, a geohazard map was prepared by combining erosion and landslide susceptibility maps (Figure 13). Using SAGA GIS software, the areas with high erosion potential were overlapped with the areas highly susceptible to landslides. In this way, the areas heavily endangered by both hazards (multi-hazard) are identified.

The analysis conducted using a special procedure in SAGA GIS reveals that a portion of the Kratovska Reka catchment is at risk from multiple hazards, namely landslides and excessive erosion. This area, referred to as the “total at risk” in the study, encompasses approximately 3.13% of the total catchment area (

Table 12. Areas with a high risk of erosion, landslides, and multi-hazard in the Kratovska Reka catchment.

Geohazards	Area
	km2	%
Landslides	22.65	33.03
Excessive erosion	6.83	9.96
Multi-hazard	2.15	3.13
Total at risk	27.33	39.86
Total	68.58	100

). These multi-hazard zones are predominantly located along the valley sides of Kratovska (Tabačka) Reka, extending from the village of Gorno Kratovo, downstream to the city of Kratovo. The terrain in this area is characterized by deforestation, bare ground, steep slopes, and the presence of weakly resistant rocks. It is important to note that there is also a potential danger to the surrounding houses situated in these high-risk zones.

4. Discussion

Soil degradation has a significant impact on soil productivity, leading to reduced agricultural yields and ecosystem functionality [44]. To effectively manage soil erosion and assist decision-makers in implementing appropriate remediation measures and mitigation strategies, it is crucial to monitor and assess the system to gather accurate and reliable information on soil erosion under current climate and land use conditions [102]. This information serves as a foundation for understanding the extent of soil degradation and identifying areas that require intervention to prevent further erosion and restore soil health. By obtaining comprehensive data on soil erosion dynamics, policymakers and land managers can make informed decisions and develop targeted strategies for sustainable land management practices.

According to the obtained results of the EPM, it follows that Kratovska Reka has an erosive catchment area. The slope of the initial relief has the biggest impact on the Kratovska Reka catchment and the fluvial processes in it. The annual transfer of eroded sediment from Kratovska Reka to the estuary in Kriva Reka averages 691 m³/km²/year. The average annual sediment production is 72.709 m³. The values of the erosion coefficient (Z) show a significant presence of terrains with medium, high, and very high risk (values greater than 0.4), which comprises 28.79 km² or almost 50% of the Kratovska Reka catchment. Otherwise, the average value of the erosion coefficient Z in the catchment is 0.56. Of the total area of the Kratovska Reka catchment, as much as 17.35% is under a very high risk of erosion, which exceeds 2000 m³/km²/year (a soil layer of 2 mm per year), which is a huge value. Excessive water erosion occurs here, which causes the creation of numerous erosive relief phenomena, the loss of fertile areas, and the filling of riverbeds with a large amount of sedimentary material. The specific erosion rate in the Kratovska Reka catchment is estimated to be 1286.5 m³/km²/year. The high specific erosion rate suggests that the investigated catchment experiences significant erosion rates, leading to the loss of soil and sediment deposition in the river system.

In comparison to the entire Kriva Reka catchment, the Kratovska Reka catchment exhibits higher erosivity, with a coefficient Z of 0.56 compared to 0.51. Additionally, the average annual erosion rate for the Kratovska Reka catchment is estimated to be 1063 m³/km²/year, whereas the Kriva Reka catchment has an erosion rate of 805 m³/km²/year. These results indicate that the average annual erosion in the Kratovska Reka catchment is nearly twice as high as the average specific erosion rate for the entire territory of North Macedonia, which is reported as 681 m³/km²/year according to [50]. Furthermore, a comparison was made with a similar-sized study area, the municipality of Štrpce in southern Serbia. In a study [103], it was found that 55.20% of the Kratovska Reka catchment exhibits low erosion rates, whereas the majority of the Štrpce municipality (50.03%) displays low erosion rates. Similar observations have been made by studies conducted by [38,45,104], which indicate that the intensity of soil erosion is closely linked to dynamic interactions among natural conditions, demographic and settlement indicators, and changes in land use.

This study considered six factors as preconditions for landslide occurrence, namely slope, lithology, land cover, plan curvature, distance from streams, and distance from roads. These factors were converted into raster format and standardized to a cell size of 15 m × 15 m. The results revealed that a significant portion of the catchment (1/3) falls into the class with a very high probability of landslides, particularly in the downstream valley of the Kratovska Reka, below the city of Kratovo, and in the valleys of its left tributaries. The analysis also identified the steep valley sides of Kratovska Reka and the upper reaches of its right tributaries as areas with the highest probability of landslide occurrence. Conversely, the higher mountainous regions, characterized by lower slopes and composed of solid and compact volcanic rocks, exhibit a lower probability of landslides. Overall, the high-risk class, indicating a significant likelihood of landslides, encompasses approximately 33.03% of the Kratovska Reka catchment.

At the national level, the regions in North Macedonia with high and very high landslide susceptibility are predominantly found in hilly terrain, mountain foothills, valley bottoms, gorges, depressions, and basins [54]. These areas are typically characterized by the presence of Neogene lacustrine sediments. In terms of the entire country, a significant portion (18.9% according to the LSI model, 33.4% according to the LSI + AHP model, and 40.1% according to the AHP model) falls into high and very high landslide susceptibility zones, as verified by the landslide dataset [54]. In comparison, the Kratovska Reka catchment exhibits a higher susceptibility to landslides, when compared to the findings of [54]. Results provided by this study indicate that the Kratovska Reka catchment is particularly prone to landslides, highlighting the need for appropriate measures and strategies to mitigate the associated risks in the area.

In order to enhance the mitigation measures and approaches, several aspects should be taken into consideration. Firstly, there should be a significant improvement in the inventory of landslide data, ensuring its accuracy and completeness. Additionally, the spatial resolution of the precondition data layers should be enhanced to capture more detailed information. Improving the weighting of factors is another important aspect, as it can contribute to a more accurate assessment of landslide susceptibility. This may involve refining the weights assigned to different factors based on their relative importance and influence on landslide occurrence. Furthermore, the inclusion of additional precondition factors can enhance the model’s predictive capabilities. Factors such as the topographic wetness index (TWI), stream power index (SPI), normal difference vegetation index (NDVI), geological structural elements, precipitation intensities, and other indicators can provide valuable insights into the susceptibility of an area to landslides (mass movements-wet).

Moreover, it is recommended that alternative model validation approaches, such as those proposed by [105,106,107], are explored. This can help ensure the robustness and reliability of the susceptibility models. Overall, a comprehensive assessment of total susceptibility to natural hazards can provide valuable information for identifying vulnerable areas and implementing measures to protect the environment, as well as natural and cultural heritage [79,108,109,110].

The terrain characteristics of the Kratovska Reka catchment in North Macedonia, including steep slopes and surface disintegration of rocks, make it susceptible to various geohazards (see Figure 13). Among these hazards, excessive water erosion and landslides are particularly prominent. To assess and understand these risks, accurate GIS-based models have been developed to determine the potential occurrence of such hazards. These models indicate that a significant portion of the Kratovo municipality within the catchment area is at risk of experiencing these hazards in the upcoming multi-year period. The results hold significant importance for the local community and various institutions involved in environmental protection and emergency management. By providing an overview of the most potentially endangered locations in the area, these results serve as a valuable resource for decision-making and planning purposes [103]. They help stakeholders identify areas that require special attention and prioritize efforts to mitigate risks, protect communities, and manage emergencies effectively.

The places with the highest erosion potential and the riskiest parts for the probability of landslides are on the steep valley sides of the Kratovska Reka and its tributaries, and the source parts of the right tributaries, especially in places that are completely exposed or are under sparse grassy vegetation and unprotected from the erosive effect of rain. Precisely because of the above, the bare and steep valley sides of the Kratovska Reka and its tributaries, with an excessive amount of precipitation, can contribute to an increase in the intensity of erosion and the occurrence of landslides.

Within the Kratovska Reka catchment area, 3.13% of the total area is categorized as multi-hazard zones, meaning they are prone to both landslides and excessive water erosion. These areas are primarily located along the valley sides of the Kratovska (Tabačka) Reka, spanning from the village of Gorno Kratovo, downstream to the town of Kratovo. The terrain in these zones is characterized by deforestation, lack of vegetation cover, steep slopes, and the presence of rocks with low resistance to erosion. These factors contribute to the heightened vulnerability of these areas to multiple geohazards.

In the Kratovska Reka catchment, it is crucial to comprehensively assess the erosion situation, both remotely and in the field, in order to define and implement appropriate anti-erosion measures to mitigate the intensity of erosion and soil loss. It is important to prioritize the dominant hazards, in order to implement targeted preventive measures. Considering the specific conditions and physical characteristics of the investigated area, measures can be tailored to address the most significant hazards, such as landslides and excessive water erosion. By focusing on these dominant hazards, it is possible to develop and implement effective preventive strategies that mitigate the risks associated with soil loss and protect the environment, natural resources, and infrastructure in the catchment.

The standardization and implementation of multi-hazard methodologies would be a significant step towards enhancing the qualitative monitoring and identification of geohazards at the local and regional scale in North Macedonia. This approach would emphasize the importance of risk assessment and management programs, which are currently under development in this part of southeast Europe. Furthermore, future studies could expand on this work by conducting broader research that covers the entire territory of North Macedonia.

The maps and calculations derived from this research could serve as a foundation for a comprehensive inventory of hydro-meteorological hazardous events. Such an inventory would provide valuable knowledge for establishing appropriate management and mitigation guidelines and measures, not only within the studied area but also at regional and national levels in the region [38]. This would greatly contribute to improving the overall understanding and response to geohazards in North Macedonia, leading to more effective risk management and enhanced resilience to natural disasters induced by climate change.

The research included a crucial aspect of validating the results obtained from the implemented approach. The most effective validation tool for the erosion potential model (EPM) is comparing the model results with the average annual sedimentation in reservoirs that rivers flow into, typically measured using echo sonars. Unfortunately, there are no such reservoirs in the Kratovska Reka catchment or its larger recipient, the main river of Kriva Reka. Another option for future investigations is comparing the model results with measured sediment load, but the only gauge station nearby is located on the main river, Kriva Reka, near the village of Trnovac. However, a similar GIS-based EPM approach implemented by [50] on the Kriva Reka has demonstrated that the model produces highly accurate results when comparing sediment transport through the river profile. To further enhance the accuracy of the model, the installation of a gauge station with sediment transport measurements near the outlet of Kratovska Reka would provide a direct opportunity for result comparison and validation. Until then, it is considered that the model’s comparison with sedimentation data for the entire Kriva Reka catchment, as well as the heuristic approach involving opinions from three additional national erosion experts regarding soil erosion quantity based on the national maps, demonstrates acceptable accuracy of the obtained results.

Validating the landslide susceptibility model is relatively easier, as it requires at least a basic landslide inventory. In this regard, the positions of the 31 landslides recorded during the field survey were compared with the model’s zonation, resulting in an 80.7% overlap with the high and very high susceptibility zones.

The ROC curve and AUC value in this study were also calculated. Based on the calculated data, the AUC value was determined to be 0.856, indicating a good level of accuracy for the model employed. Additional validation can be conducted by analyzing the Kappa values. Furthermore, it is crucial to emphasize the necessity for further research, particularly related to the validation of results using precise LIDAR measurements and other advanced technologies to assess whether erosion and landslide risks are increasing or decreasing in this part of North Macedonia.

5. Conclusions

The analysis of territories that are highly susceptible to geohazards plays a crucial role in facilitating the adoption of appropriate preventive measures and effective environmental management actions by local, provincial, and state services [103]. The primary objective of a multi-hazard map is to identify and delineate hazardous areas, thereby aiding in activities aimed at reducing the hazards and mitigating future damage [55,56]. This becomes particularly significant when considering the impact of climate change, which is associated with increased landslide activity and excessive erosion. The approach presented in the paper has the potential to generate reliable multi-hazard maps at the national level and provide valuable information for regional planning and decision-making processes. By utilizing such maps, stakeholders can make informed choices and take appropriate measures to minimize the risks associated with geohazards and enhance overall resilience to environmental challenges.

The Kratovska Reka catchment, with its distinctive physical characteristics, geographical position, and rich geodiversity, offers an ideal setting for conducting a comprehensive analysis of geohazards on a regional basis. Utilizing Geographic Information Systems (GIS), relevant data pertaining to the entire research area were used to analyze and assess the extent of excessive and water erosion, susceptibility to landslides, and identify areas prone to multi-hazards. By leveraging the capabilities of GIS, a holistic understanding of the interaction between the geohazard and landscape within the case study was achieved. Creating a cadastre of areas endangered by geohazards in the Kratovska Reka catchment can indeed be a valuable future goal. Such a cadastre could serve as an essential tool for sustainable spatial and urban planning, thus enabling decision-makers to incorporate hazard mitigation measures into development plans, in order to reduce unfavorable effects on natural resources, the local economy, and socio-demographic processes.

Landslides have been identified as the predominant geohazard in the Kratovska Reka catchment, accounting for approximately 33.03% of the total area vulnerability. Additionally, the analysis of excessive erosion revealed a mean erosion coefficient of 0.56, indicating a high level of erosion within the analyzed catchment. Nearly 50% of the area is affected by the highlighted erosion processes. By integrating the analysis of these two geohazards using GIS techniques and employing methods such as natural breaks, a multi-hazard model combining water erosion and mass movements (wet) was derived for regional scale analysis. This model indicates that 3.13% of the catchment is highly or very highly susceptible to the combined effects of water erosion and landslides. The overall risk associated with these geohazards, including the multi-hazard susceptibility, amounts to 39.86% of the catchment area. These findings emphasize the importance of addressing both landslides and excessive erosion in the management and mitigation of geohazards within the Kratovska Reka catchment in North Macedonia.

In this area, which is a part of the Kratovo-Zletovo paleovolcanic area, combining the obtained results with field research is crucial for effectively preserving and protecting the geodiversity values. The results of this work contribute to the growing evidence of erosion risk in southeast Europe [22], shedding light on the specific challenges and vulnerabilities within the Kratovska Reka catchment in North Macedonia. The identification and assessment of erosion risk in this part of the region provide valuable information for understanding the broader patterns and dynamics of erosion processes in southeast Europe [25,26,27]. The findings can be used as a foundation for further research and refinements in erosion risk assessment methodologies, including the incorporation of additional factors, validation techniques, and improved modelling approaches [55,56,57,58,59,60].