From Historical Maps to LiDAR Data-Enhancing Landscape Ecological Research of Cultural Landscape Using Modern Remote Sensing Data Illustrated with Examples from Slovak Traditional Heritage Landscapes

Abstract

The study of cultural landscapes has a tradition stretching back several decades. During this time, methods have been developed based on the geographical data and technological capabilities available. However, with new data becoming available, new methodological and conceptual challenges arise in linking different types of landscape data. In our article, we attempt to address these challenges. These include historical maps and remote sensing data, such as aerial and satellite images and LiDAR data. We illustrate these using examples of traditional heritage landscapes in Slovakia. We critically evaluated the informational value of historical maps and their connection with remote sensing data. Our case studies focused on using LiDAR data to identify overgrowing processes, historical trackways, agricultural terraces, catchworks and tree vegetation in wood pastures. Digital technology provides new and more accurate data, as well as new ways of evaluating it, thereby enriching existing landscape ecological methods of cultural landscape research.

1. Introduction

Cultural heritage landscapes are multifaceted entities that incorporate the tangible and intangible interactions between human societies and their natural environments throughout history. Representing the visible imprint of human activity on the land, they range from ancient agricultural systems and medieval settlements to industrial complexes and sacred spaces. These landscapes embody aesthetic and ecological values, as well as deep historical, social and symbolic meanings that form part of our collective memory and identity [1,2].

The concept of cultural landscapes gained formal international recognition in the 1992 revision of the UNESCO World Heritage Convention, which recognised cultural landscapes as an essential category of World Heritage Sites [3].

The significance of cultural heritage landscapes is particularly pronounced in Europe, where thousands of years of human habitation have shaped diverse and layered landscapes [4]. From the terraced vineyards of the Douro Valley in Portugal to the pastoral meadows of the Carpathians and the relict industrial zones of the Ruhr in Germany, European landscapes reflect dynamic histories of adaptation, conflict, and coexistence between people and place. In Slovakia, the most important representatives of these landscapes are historic mining landscapes and traditionally farmed agricultural landscapes, which have been listed as UNESCO World Heritage Sites. While the mining landscape is characterised by the presence of technical works, water management systems, and tailings heaps associated with specific urbanisation [5], the hereditary agricultural landscape is characterised by a preserved structure of land use, a mosaic of fields and natural features reflecting traditional farming.

In academic literature, the study of cultural heritage landscapes is inherently interdisciplinary, combining perspectives from geography, archaeology, history, anthropology, ecology, architecture, and planning. As Sauer [6] and later geographers emphasised, cultural landscapes are not merely static objects of study but dynamic palimpsests that evolve through continuous negotiation between cultural practices and natural processes [7,8].

In recent decades, concerns over landscape degradation, urban expansion, climate change and the loss of traditional land use practices have intensified scholarly and policy interest in identifying, documenting and sustainably managing cultural heritage landscapes. Furthermore, the digital age has transformed the way in which these landscapes are studied and protected [9,10]. Advanced geospatial technologies, participatory mapping platforms and linked open data initiatives are now central to documenting and promoting cultural landscapes [11,12,13,14]. These tools support more inclusive heritage management strategies and allow for the better integration of cultural landscapes into broader territorial development and climate resilience policies.

Identifying and mapping cultural heritage landscapes is fundamental to preserving them. Modern approaches to changes in land cover and land use focus on very recent time horizons, using automated, semi-automated or manual interpretation of satellite images [15,16,17,18,19,20] with various geographical coverage. The main purpose of such surveys is to provide updated information on the spatial distribution of land-related phenomena, primarily for operational reasons and future landscape forecasting. However, cultural landscape development is a continuous, long-lasting process that combines natural conditions with the needs and abilities of human society. To anticipate the future, we must understand the past. When assessing cultural landscape development, the complex mechanisms that formed the natural environment in remote geological eras can be disregarded. The stabilisation of cultural, social, and production relationships in our cultural landscape occurred over several recent centuries. In Central Europe, the transition from a feudal to a capitalist system in the 18–19th centuries, and then from a capitalist to a socialist system and back to a free-market democratic system in the 20th century, had a significant impact on the cultural landscape. Land use has undergone significant changes during that time. These changes were mostly caused by technological development and changes in the political and property situation. An interval around the year 1950 was discovered within the Earth System science community to represent the most pronounced upward deflections of numerous global socio-economic indicators and trends (the growth of foreign investment, GDP, greenhouse gas emissions, population, rapid urbanisation, transportation, travel and tourism, the consumption of energy and water, deforestation rates, and many more). The deflection became known as the ‘Great Acceleration’ [21]. In recent decades, the main drivers of land use change, especially in post-socialist countries, have been the economic situation, social preferences, and market regulations. These drivers are often combined with unexpected natural disturbances, which have occurred more frequently and with greater intensity in recent years [22,23,24,25].

Combining preserved historical geographical data sources (such as historical maps) with modern methods and tools (GIS and remote sensing) enables relatively precise identification of the cultural landscape development of a certain territory. This approach has been widely adopted in recent landscape, ecological and geographical research (e.g., Refs. [26,27,28,29,30,31,32]). In the Central European landscape ecology school, research is based on a common scientific background and the availability of historical maps (e.g., Refs. [7,33,34,35,36,37,38,39,40,41,42,43,44]).

Recently, new remote sensing detection technologies have been developed, and the resolution and accuracy of existing technologies have been increased. These technologies are also becoming more accessible to users. Therefore, the new challenge for cultural landscape research is to incorporate these new data sources into existing traditional methodological approaches and exploit their potential for quantitative and qualitative advances.

The article provides an overview of proven approaches to landscape ecology research in cultural landscapes, using regional case studies as examples. The aim is to emphasise the importance of different types of data when identifying the various parameters of cultural and heritage landscapes. The informational value of historical data is critically evaluated, and it is demonstrated how new sources of up-to-date land surface data can extend and refine existing methods for identifying and assessing cultural landscape change using Slovak traditional heritage landscapes as an example (Figure 1).

2. Study Areas

Slovakia is a landlocked country located in Central Europe. It has a population of over 5.5 million. It covers an area of 49,033.47 km², 49.7% of which is agricultural land, 43% of which is forest, 5% of which is urban, and 2% of which is water. The average population density is 110 inhabitants per km². From a European perspective, Slovakia has a sub-mountainous to mountainous landscape. Highlands and mountains cover 60% of the territory and lowlands cover the remaining 40%. Altitudes range from 94 to 2655 m above sea level. According to the Köppen–Geiger climate classification, Slovakia falls under the continental climate zone (D), specifically Dfb (Humid Continental with Warm Summer). This means that Slovakia experiences four distinct seasons, including warm summers and cold, cloudy and humid winters. The average winter temperature is around −2 °C, but it can drop to −15 °C. Summers average 21 °C, but temperatures can exceed 30 °C. The average annual precipitation across the whole country is 743 mm, ranging from 500 mm in the south to 2000 mm in the highest mountains.

To illustrate the methods and procedures of cultural landscape research, we selected eight study areas that represent traditional agricultural heritage landscapes in Slovakia (Figure 1).

The Hriňová landscape is a traditional agricultural area characterised by dispersed settlements and sparsely scattered fields on heavily dissected uplands at an altitude ranging from 500 to 800 m.

The Sliač pasture–forest landscape is located near the town of Sliač in the Zvolen Basin on moderately dissected uplands at an altitude of 400 m.

The historical trackways under investigation at the Drahy site are related to Zvolen Castle, also known as the Deserted Castle, which was built in the medieval period and occupies an area of 4.7 ha at an altitude of 575 m. The castle was abandoned at the turn of the 14th and 15th centuries. Located in central Slovakia, the castle lies on the historical international trade route known as the ‘Magna Via’, which runs from central Hungary to Poland through the historical county of Zvolen.

The cadastral district of Horný Tisovník lies in Central Slovakia. Almost half of its agricultural land is managed in a traditional manner. This traditional agricultural landscape is characterised by a mosaic of small plots of land and scattered settlements. Agrarian plots on slopes are usually arranged in terraces. Historically, the cadastral district of Horný Tisovník was governed by two counties with different feudal economic systems, one agricultural and the other industrial.

The pasture and meadow catchworks are located in the Riečka catchment area, near the Poľana Protected Landscape Area, where the UNESCO-sponsored ‘Man and the Biosphere Programme’ is implemented. The region was first colonised by pastoralists in the 17th century, followed by peasants. Currently, the catchworks are only partly functional.

The last four cadastres demonstrate the evolution of traditional wood pastures. The village of Kameňany is located in the uplands at an altitude of 246 m. Its historical wood pasture has been converted into permanent meadows. Katarínska Huta is situated in the deeply dissected lower highlands at an altitude of 288 m, and the overgrown wood pasture here represents natural succession processes. Klátova Nová Ves is located in moderately dissected uplands at an altitude of 198 m. It is an example of a well-preserved wood pasture with clearly visible landscape structures. Hontianske Moravce is located in the uplands at an altitude of 147 m and has a preserved wood pasture.

3. Historical Maps

Historical maps represent a unique source of information on past landscapes, and when combined with aerial photographs and satellite images, they provide a vivid picture of the landscape and how it has been used over time. While remote sensing data is very precise and accurate, it only covers the last few decades of development. Historical maps, on the other hand, allow us to look back a few centuries. The 18th century was a period of significant change in Central Europe. It was a time when the feudal age ended and capitalism and the technical revolution began, marking the end of the old feudal rural landscape and the birth of the new pre-industrial landscape.

3.1. Sources of Historical Maps in Central Europe

The oldest maps depicting Slovakia were created in the 15th century [45]. However, they were very inaccurate and on a very small scale. Therefore, their potential application in landscape ecological research is limited. The first large-scale maps were created in the first half of the 18th century by mining cartographers, primarily depicting mining towns in central Slovakia and their surroundings. These maps primarily focused on mining objects, such as mines, towers, processing buildings, dump sites and water reservoirs. Other land areas or objects, such as settlements, fields and forests, were drawn only schematically, and the maps were inaccurate, especially in hilly and mountainous areas.

The first large-scale maps covering the whole Austrian Empire were those from the First Military Survey (1763–1785), with a scale of 1:28,800. This survey focused on all military-important objects and areas. These coloured maps (Figure 2a) show the following land uses: forests, shrubs, trees, bushes, grasslands, wetlands, fields, vineyards, buildings, settlements, gardens, water, rocks and roads. Relief was hatched according to the inclination of the slopes. Despite using the same methods and techniques, maps from different parts of the Empire varied in detail and accuracy [7,34,42].

Thanks to precise geodetic measurements, the maps from the Second Military Survey (1806–1869) are more accurate and informative than previous maps. The scale remained 1:28,800. The following land use forms could be identified on these coloured maps (Figure 2b): forests, forest edges, shrubs, trees and bushes, meadows, pastures, wetlands, fields, vineyards, buildings and settlements, gardens, water, rocks, roads and railways. Although relief was still hatched, the land use polygons were sharply outlined, enabling better definition of landscape features compared to the former mapping. Although these maps are relatively precise and accurate, they have the disadvantage of displaying multiple types of information (hatching, colour, borders and text) in the same place, which sometimes causes interpretation problems.

The main purpose of the Third Military Survey (1869–1887) was to provide the Austro-Hungarian army with precise, high-quality maps. The maps were produced at a scale of 1:25,000. Altitude was derived from sea level at Trieste, and relief was depicted for the first time using contour lines. These maps are much more accurate than previous ones. The same land use categories as in the Second Military Survey can be distinguished on these coloured maps. However, due to the turmoil following the collapse of the Austro-Hungarian Empire in 1918, only black-and-white map reproductions (at a scale of 1:25,000 or 1:75,000) are available [45] in the newly formed national states of Central Europe (Figure 2c).

In the 1950s, military topographical maps using the universal transverse Mercator projection were created (Figure 2d). The basic scale was 1:25,000, and the following land use categories could be identified on the coloured maps: forests, shrubs, trees and bushes, meadows, pastures, fields, water, permanent agriculture, gardens, buildings and settlements, rocks, and roads and railways. Due to their exact cartographic projection and new geodetic technology, these maps are highly precise spatially and can be easily processed in landscape ecological analyses. Although military maps were primarily intended for the army, they contain unique information on historical landscape structure, use and scenery that is rarely exploited nowadays.

The first cadastral maps of our territory were created between 1856 and 1867 at a scale of 1:2880. The following land use categories could be identified on the black-and-white cadastral maps: forests, meadows, pastures, fields, buildings, water and roads. Cadastral maps produced in the first half of the 20th century were mostly created at a scale of 1:5000, although some were created at a scale of 1:2880 or 1:2000 [45]. Due to their precise geodetic measurements and large scale, cadastral maps are highly accurate and provide detailed documentation of land use patterns, particularly historical ones.

Cadastral maps are a unique hybrid source of data, providing information not only on the location of objects on the Earth’s surface, but also on land use and the spatial structure of ownership. As is the case today, not all ownership or administrative boundaries recorded on a map have a physical manifestation in the landscape. Conversely, valid property rights can often explain the unusual location or positioning of roads or other features in the landscape.

3.2. Temporal, Spatial and Thematic Resolution of Historical Maps

The interpretation of historical maps depends mainly on their focus, level of detail, accuracy, and the cartographic method used to depict landscape phenomena [46]. These factors are particularly significant when comparing maps from different social or economic sectors and time periods. Therefore, the interpretation of historical maps must be objectively critical. To evaluate the possible localisation of spatial information shown on historical maps, an interpreter must combine their knowledge of the studied landscape (or the experiences of other authors who have conducted similar research) with their own expertise. For this reason, interpreters usually manually vectorise the spatial objects of interest on screen (as in Figure 3).

The spatial accuracy and reliability of historical maps generally decrease with age. To analyse historical maps in GIS, geographical coordinates must be assigned to points (landmarks) that have not changed location over the studied period and can be identified on both historical and present maps, such as churches, chapels, crossroads, bridges and mountain peaks. The number of points required depends on the transformation method used. Simpler methods only place and rotate a map within a coordinate system; more advanced methods can also deform it spatially. When applying a linear affine transformation, the root mean square (RMS) error of the transformed maps from the First Military Survey increases from 100 m in lowland areas to 500 m in mountainous regions (up to 800 m in narrow valleys). This is due to the cartographic techniques used for map tables, which are easily applied in open landscapes but difficult to apply in narrow mountain valleys [47,48,49,50,51]. However, as reported by Vessella [20], it is possible to achieve an RMS error of just a few metres when georeferencing the 19th-century cadastral map using a polynomial transformation.

Possible solutions to this problem include transforming a map in parts or vectorising polygons according to the relief and streams [29,31,35]. Maps from the 19th century are processed in the same way, but their spatial accuracy is significantly higher, thanks to their precise geodetic measurements (RMS less than 100 m). An alternative mapping method is retrospective mapping [52], also known as backdating [53,54,55]. In this method, the most recent maps are digitised first, and this vector layer is then used as a template for redrawing the older time layers. This approach avoids mapping unrealistic changes that could emerge from inaccuracies in older maps.

Due to differences in spatial information on various historical maps, phenomena must be grouped into comparable categories. The oldest and most generalised maps usually limit comparison [30,49,55]. The categories, as well as their quality and quantity, depend on the character of the landscape and the focus of the research.

Apart from the content of the patches (i.e., land use), their size, shape, and localisation (as expressed by the landscape indexes) also provide information about landscape processes in the sense of Forman and Godron [56]. Larger, more regular patches indicate more homogeneous natural conditions, while smaller, linear polygons indicate more heterogeneous conditions or the neighbourhood or transition of larger, homogeneous areas. As a general rule, the older the map, the less detailed it is. Several studies [34,44,57] have documented that the number of mapped patches on the oldest maps is three to four times lower than on maps from the mid-20th century (see Figure 3). When analysing the historical development of cultural landscapes, it is therefore necessary to critically evaluate the value and potential of historical maps in combination with remote sensing data. While these two data sources are fundamentally different, each offers its own advantages.

3.3. Analysis of Past Land Use and Its Implications

Overlaying temporal land use maps creates a spatial database of polygons containing information on past use; for example, the Historic Landscape Characterisation method [10]. Processing these attributes into contingency tables (using the area categories) creates a land use transformation matrix, indicating the amount of change and land use trends [20]. The localisation of maximum areas (e.g., in hectares) on the diagonal of the matrix indicates the relative stability of land use; a shift from the diagonal indicates the intensification or extensification of land use.

In general, two categories of land use change can be distinguished: (a) Areas without changes, representing stable landscape cores in the sense of; and (b) Areas with land use change (with variations in changes). Each land use change reflects a change in the physico-geographic and/or socio-economic conditions of the studied landscape [7,10,34]. Patches with permanent land use change indicate localities without strictly defined landscape or land use characteristics (e.g., transition between lowland and mountainous areas or agricultural and forest land use). Changes in land use may also indicate changes in natural conditions (e.g., the drainage of wet meadows followed by their use as arable land). Only a temporary change in land use, such as a rapid increase in the area of fields, could be caused by higher demand for food over a certain period of time. Some interpretations are valid for the whole landscape, while others reflect local phenomena, such as the tourism boom in the High Tatra Mountains in the 20th century and the construction of the Starina water supply reservoir [35].

Changes in land use and their direction may be expressed in terms of the intensity of land use change (e.g., Refs. [34,55,58]). Each land use form is assigned a land use intensity coefficient according to the amount of energy required for change. The intensity of land use change is then calculated as the difference in intensity coefficients between neighbouring time periods. Absolute intensity of land use change refers to the total amount of change, while relative intensity expresses its orientation towards intensification or extensification of use.

Land use is a concrete expression of human activity in space and time, accumulating a certain historical, economic, social and cultural potential, and representing an intersection of natural conditions, human knowledge and techniques [7]. Proposals for sustainable land use must be based on the analysis of past land use, which helps us to define a landscape’s potential [34,41]. Historical maps are an adequate source of information for landscape ecological research focused on land use development and the assessment of the relationship between natural conditions and their economic use, as well as for the prognosis and planning of sustainable land use. Historical development analyses can help identify risks and hazards faced by former inhabitants, such as floods, landslides and soil erosion, and apply this knowledge to present-day land use planning. The localisation and development of both settlements and communication lines are important factors affecting the accessibility of the landscape, which is reflected in land utilisation. Identifying patches with stable land use and analysing their natural conditions allows us to propose the most suitable, and therefore sustainable, future land use [10,34,35,59,60,61,62].

4. Remote Sensing Data

Technological advances in the 20th century have enabled us to track changes in landscape structure on much smaller timescales than was possible with historical maps [63]. In Slovakia, it has been possible to map changes in land use using remotely sensed data, such as aerial or satellite imagery, for the last 75 years [64]. Aerial and satellite imagery (and orthophotomaps derived from them) represent a fundamentally different source of information about the landscape compared to historical maps. While historical maps provide information about the landscape as interpreted by their creator and limited by the technological capabilities of mapping, aerial remote sensing imagery faithfully depicts the landscape as it was at the time of capture. Despite certain limitations associated with spatial resolution, spatial and thematic accuracy, and the possibility of overlapping clouds [65], remote sensing imagery offers significant advantages:

• The interpretation of mapped features lies entirely with the mapper, who selects the relevant information for their research from the real image of the landscape;
• Many remote sensing products provide significantly higher resolution and accuracy than historical maps, enabling the identification of objects and structures not depicted on maps [66];
• Large areas can be sensed at the same time;
• Satellite imagery, such as that from Landsat or Sentinel, is readily available and has a relatively short image repeat interval (every 16 or 5 days, respectively). This allows changes to be observed within a year [67].
• It is also possible to use automated landscape cover classification methods (Remote Sensing Image Classification Methods) [68,69], which are state of the art and provide massive parallel computing capability as well as a huge amount of online datasets [70];
• Since it is usually multispectral data, different spectral bands and combinations of spectral bands can be used, as well as the calculation of different indices or metrics, to classify the surface [65].

In recent years, orthophotos have also become readily available, and in Slovakia, for example, they have been regularly overwintered at three-year intervals since 2017. According to Zhu et al. [71], the short time intervals between images allow us to answer the following five questions: when the change happened; where the change happened; what the change was; how the change happened; and why the change happened.

4.1. Satellite Images and the Data Derived from Them

To demonstrate the application of satellite data, we present the identification of overgrown areas in Slovakia between 1990 and 2006 (Figure 4). Corine Land Cover (CLC) layers from the respective years were used for the analysis. CLC 1990 and CLC 2006 were derived from Landsat satellites (30 m resolution) through visual interpretation as part of the European Environment Agency’s Copernicus project (www.copernicus.eu, accessed on 20 March 2025). Since 1990, the Corine Land Cover Mapping of Europe project has developed a comprehensive classification of land cover classes, as well as a methodology for monitoring and assessing land cover change [72]. Land cover change was identified by interpreting changes in CLC categories between 1990 and 2006 [73], thereby highlighting agricultural areas where the intensity of use had decreased to the point of abandonment. This was reflected in the subsequent overgrowth of the area, leading to a change in the original landscape structure.

Overgrowth was defined as a change in one of the following Corine Land Cover 1990 categories: non-irrigated arable land (code 211), vineyards (221), fruit trees and berry plantations (222), pastures (231), complex cultivation patterns (242), agricultural mosaics with significant natural vegetation (243), and natural grassland (321). These are areas that may have been overgrown by 2006 into transitional woodland shrub (324); forests (311, 312 and 313); or agricultural mosaics with significant natural vegetation (243).

A similar transformation table (a matrix of flows) based on the Corine Land Cover approach has been used in several studies to assess changes in land cover [53,74,75]. We also assessed the dependence of overgrown areas on selected natural conditions, such as distance from settlements, elevation, slope, climatic conditions, soil conditions and topography [73].

Based on these results, we found that, in 2006, less than 18% of underwater agricultural areas were in various stages of secondary succession (see Figure 4). From 1990 onwards, grasslands (meadows and pastures) were the most overgrown land cover type, occupying less than 66% of overgrown land between 1990 and 2006. This was followed by predominantly agricultural areas with a significant proportion of natural vegetation (25%), and non-irrigated arable land (6%). The remaining categories had a total of 3% of the remaining vegetation. The proportion of overgrown areas increased with distance from settlements and in sites with less favourable conditions for agricultural production, such as cooler sites at higher altitudes, steeper slopes and more rugged topography [73].

However, it should be noted that the area of overgrown land is probably larger than it appears. Resolution of CLC (minimum mapping unit 25 ha and minimum change 5 ha) limits a more detailed assessment. For example, traditionally farmed agricultural landscapes, characterised by small alternating patches, are mapped in the Corine Land Cover methodology as a whole, as Complex cultivation patterns (code 242) or Agricultural mosaics with significant natural vegetation (243). More recent Copernicus products, such as the high resolution layers and the CLC Backbone [76], allow for more detailed mapping; however, their resolution is at 20 or 10 m. In addition, the direct use of freely available satellite images, such as LANDSAT (30 m, or 15 m by using the pansharpening function) and Sentinel (10 m), only allows the identification and mapping of basic land use structures and changes of more than 5 ha. Some typical features of agricultural landscape heritage, such as terraces, cannot then be identified. Image resolution is important for identifying different objects on Earth’s surface [77]. Finer resolution enables the detailed mapping of individual features, such as buildings, trees and fields, as well as linear features, such as boundaries, roads and streams. This is essential for mapping cultural landscapes. This is because it is the spatial structure and arrangement of elements in the landscape that create its characteristic image and character, which are fundamental attributes of a heritage cultural landscape [78,79]. For the identification of more detailed structures, remote sensing imagery with finer resolution is needed (see Figure 5).

4.2. More Precise Orthophoto and LiDAR Data

Orthophotos ranging from panchromatic to the latest 4-channel are readily available in Slovakia. The first full-area aerial photography of Slovakia was carried out in the mid-20th century (1949–1950). This panchromatic photographic imagery had a resolution of 0.5 m, providing a unique view of Slovakia’s landscape before industrialisation and collectivisation following 1950. In the following decades, the armed forces carried out aerial photography, but only of selected parts of the territory. Together with the initial imagery, this was not made available to the public until 1990. Full-colour aerial imagery with a resolution of 0.5 m (0.25 m since 2008 and 0.2 m per pixel since 2014) has only been produced commercially at three-yearly intervals since 2003. Since 2017, the Geodetic and Cartographic Institute in Bratislava and the National Forestry Centre in Zvolen have been the contractors for the freely available images. The images are taken in three-year cycles; the most recent cycle produced 4-channel images (e.g., enabling the calculation of the NDVI vegetation index) with a resolution of 0.15 m [80]. As previously mentioned, linking this information source to historical maps is not straightforward due to the differing resolution and display principles of content. While it is possible to link functional attributes in terms of land use when studying the evolution of land use, it is not possible to directly compare the structural attributes of individual landscape features. As Olah [34] and Solár [57] have shown, the number of identifiable patches from aerial photography can be up to ten times higher. Nevertheless, it is possible to retrospectively attribute younger structural characteristics from aerial photographs to functional patches identified on historical maps. The ability to identify small objects, such as trees, was employed to map overgrowth in the traditional pastoral landscape of Podpoľanie (Gallayová [81]), where the intensity of traditional pastoral use has decreased over the decades. The level of detail in the image enables much more precise mapping of overgrowth processes than CLC options allow. However, a finer resolution may not always be optimal [65]. CLC, in turn, enabled us to gain an understanding of overgrowth across the entire Slovakian territory in a relatively short time, which would have taken considerably longer using orthophotos. However, if our aim is to map processes in as much detail as possible, the level of detail in the data source is important. Gallayová [81] reports that field verification showed woody vegetation below 1 m was not identifiable in the 2003 imagery with a pixel resolution of 1 m, whereas images from 2018 with a pixel resolution of 25 cm allow identification of this vegetation (Figure 6).

Among the latest sources of remote sensing data freely available for the whole of Slovakia are classified point clouds of the airborne laser scanning by the Office of Geodesy, Cartography and Cadastre of the Slovak Republic (OGCCSR,). Since 2017, the territory of Slovakia has been periodically scanned at three-year intervals, resulting in a classified point cloud with an elevation accuracy of up to 0.05 m, a positional accuracy of up to 0.16 m and an average number of 20 points per sq. m. The classification of points allows for the distinction of vegetation (low, medium, high) from buildings, bridges, water and ground surface.

Methods of Assessing Vegetation Change

LiDAR data enable us to assess vegetation changes in detail, including their height, tree density, and canopy shape (Figure 7). We can create a detailed digital terrain model from which we can identify historic roads, logging pits, drainage ditches and terraces, and measure their parameters (such as length, width and height). We can also map the extent to which pasture has been overgrown by woody plants in detail. To identify overgrown areas, we used, for comparison, the visual interpretation method by Gallayová [81] and two simple calculations in the ArcGIS environment (Figure 8):

• Method I: visually estimate the rate of overgrowth in a 50 × 50 m grid (Figure 8b)
• Method II: points classified as medium or high vegetation were filtered from the point cloud (LAS Dataset/Filters). We converted the selected points to a raster with a 0.5 m resolution; thus, the raster was created to identify overgrown and non-overgrown 0.5 m squares in the area. Then use block statistics to determine the proportion of overgrown squares within a defined block size of 5 × 5 m (Figure 8c).
• Method III: Alternatively, overgrowth was expressed as tree density, which was counted as the proportion of points classified as medium or high vegetation out of all points in the cloud above the area of the square 1 × 1 m resolution (see Figure 8d).

Figure 8 and Figure 9 show that these methods produce different results. Method II corresponds to the results of visually determining (Method I) the degree of overgrowth, with some differences being caused by the LiDAR data, which was obtained during the leafless season. This method is much quicker than visual interpretation and enables the reliable identification of individual trees in shaded areas. Method III, which identifies the proportion of vegetation points out of all points, provides a detailed view of gaps in forest cover, but underestimates the representation of the category with tree cover above 75%. However, it provides a more detailed view of the degree of overgrowth of grassy areas. The most suitable method depends on the purpose of the assessment.

5. Identification of Historical Landforms

The distinct character of Slovakia, coupled with its considerable landscape diversity, is intricately intertwined with the historical development of its landscape structures. The issue of historical landscape structures has long been a subject of interest across various scientific disciplines and professions. Research in Slovakia has commenced with the conservative protection of listed buildings. Consequently, the active protection of not only protected objects, but also those without legislative protection—including historic landscape structures—was promoted [82].

As was elucidated in preceding sections of this article, the subject of landscape cover and its traditional and historical applications is well understood in Slovakia and internationally. However, less well-known, yet equally worthy of attention, are the historical forms of relief, which often determine the spatial arrangement of the landscape and, in some cases, its overall characteristic appearance. The concept of historical relief forms can be understood as a simple reference to the past, inscribed upon the landscape in a manner analogous to the wrinkles present on a face. The interpretation of these references is facilitated by the morphology of the terrain, specifically the microscale landforms. Our research has spanned a period of over ten years, with a focus on the study of various historical landforms. This paper outlines the research methods used to study the following types of historical landform: transportation landforms in the context of settlements and fortifications, agricultural landforms, and mining landforms (briefly).

In 2014, the initial paper was published that addressed the subject of transport landforms within the context of settlement [83]. The mapping employed in all of the presented research was specific to the forest environment in which the historic roads are located, requiring specific procedures. The presence of mixed deciduous–coniferous stands can impede the identification and precise location of preserved historic roads when using conventional aerial photography. Therefore, innovative LiDAR methods were employed in the research.

A collaborative investigation was undertaken with archaeologists to ascertain the correlation between transportation landforms and the settlement itself. The present study identified the conditions and factors that influenced the location of pre-medieval and medieval settlements, as well as the links in the landscape through which they interacted by roads, historical roads. The routing of historical trackways was unstable and closely associated with the development of settlements—the nodes of the country. The region underwent significant changes due to the shifting of natural landscapes and the gradual expansion of the settlement network. In the late middle ages, towns that were granted the privilege of storing merchandise, collecting road and water tolls, and exercising other rights on behalf of a monarch played a significant role. As Slámová et al. [84] demonstrate, the aforementioned provisions were responsible for the intermittent redirection of transport routes.

Utilising the paradigm of agricultural landforms, it delineates a progression in the innovation of the methodologies employed. The emergence of terraces in Slovakia has been traced back to two distinct periods. Firstly, during the Wallachian colonisation of foothill and mountain areas between the 14th and 16th centuries, and secondly, during the 17–19th centuries, when settlement was established in areas deemed unsuitable for agriculture due to the presence of infertile soils and challenging climatic and terrain conditions. The implementation of multifunctional economic land-use strategies resulted in the creation of a diverse landscape by the close of the 19th century. As Slámová et al. [82] demonstrate, small-plot arable land and small-scale agricultural plots were the norm. The terrace structures, characterised by their steep slopes, exhibited a mitigating effect on the process of erosion. The landscape exhibited a high degree of biodiversity, characterised by a diverse array of landforms and a wide spectrum of agricultural products. Subsequent to the communist regime, private land ownership was considerably curtailed, and the subsequent intensification of agriculture in the 1970s resulted in the further extensive destruction of traditional agricultural practices. Presently, terraces that have been in use for a considerable time are to be found only in economically marginal agricultural areas in Slovakia, and even there, they are only present on a localised basis [85].

Terraces represent a defining feature of cultural landscapes and serve as a testament to the technological expertise and economic development of a specific historical era. These elements are ubiquitous in Slovakia’s traditional cultural landscapes. The Landscape Atlas of the Slovak Republic classifies these types as traditional meadow-pastoral landscapes, landscapes with traditional scattered settlements, and landscapes with traditional vineyards. Such structures are characteristically present within archetypes of traditional agricultural landscapes, such as vineyards and orchards. They are most frequently classified as historical structures of agricultural landscapes (HSAL) [82,85].

Catchworks are comparatively uncharted phenomena when contrasted with terraces, and the first study to be published in Slovakia was in 2015. Catchworks are defined as water gutters found in mountain meadows, which are typically constructed at varying elevations. These wetlands are also referred to as catch-meadows or downward-floating water meadows. Catchworks are defined as structures designed to collect water from a river or its tributaries. Subsequently, the water is conveyed through artificial channels or gutters until it reaches a point where it is expected to flow downhill through the grass. At this juncture, the gutter wall is interrupted, and the flow of water from the gutter to the meadow is regulated using turf hatches. The prevailing force that governs the distribution of water is gravity, and the operational costs are frequently negligible [86]. Catchworks are typically located in hay meadows or pastures and are specifically engineered to convey water from streams to mountain slopes [87].

5.1. Methods Used to Identify Landforms

All of the research studies presented were elaborated using Quantum GIS (QGIS Geographic Information System Version 3.42, Open Source Geospatial Foundation. Available online: https://www.qgis.org (accessed on 22 May 2025) and the System for Automated Geoscientific Analyses (SAGA, Available online: https://saga-gis.sourceforge.io/en/index.html (accessed on 22 May 2025)). GIS. Access to topographic data on forests, urban structures, watercourses, historical roads and height points, as well as digital models of relief (DMR3.5 and DMR5.0), was facilitated via the ZBGIS (Basic Database for a Geographic Information System) Geoportal public website, which is operated by OGCCSR. Web maps of ZBGIS (https://geoportal.gov.sk/gallery/datasets/detail/8d11d6d5-71d0-4a83-a097-f5b9280068c2, accessed on 30 March 2024) were provided in accordance with the OGC WMS (Open Geospatial Consortium Web Map Service) 1.3.0, 1.1.1 standards.

In some of our earlier studies, we employed LiDAR scans developed as part of the “Decision Support in Forest and Country” project at the Centre of Excellence (CEX ITMS 26220120069) and procured from EUROSENSE Ltd, Wemmel, Belgium. Other data was provided by the Department of Forest Management and Geodesy at the Technical University in Zvolen [88]. In addition, other LiDAR products were obtained from the ZBGIS Geoportal. It was made available to the public via web servers operated by OGCCSR in our recent research [84]. Access to historical maps of the First Military Survey, Second Military Survey and Third Military Survey was facilitated via web map servers operated by the Slovak Environmental Agency and the Ministry of the Environment of the Slovak Republic.

At the inception of the research, historical maps, topographic maps, orthophotomaps and objects from the terrain determined by the global navigation satellite system (GNSS) hiking instrument were utilised [83]. The positioning of the mapped objects in the field was facilitated by utilising a tourist GNSS device, namely the Garmin 60csx (Garmin Ltd., Olathe, KS, USA).

5.1.1. Recognising Transportation Landforms

The mapping and positioning of historical roads and groups of holloways was conducted during the field research from 2013 to 2015. In the context of linear objects, such as roads, a range of potential deviations from the measurements was assigned, and buffer zones delineating mapped historical roads were established (3 m, 5 m, 10 m and 30 m). All access points from the countryside, entries to the forest and historical forest roads were marked, as these positions are connected to nearby villages, small settlements and the nearby town. This indicates the regional and international context of the investigated area. A comparison was made between trackways marked with GNSS and roads identified on historical military maps. The length and number of roads, as well as the network density of current and historical forest roads, were evaluated [83].

LiDAR data was collected using a Riegl LMS-Q680i scanner at an altitude of 600–900 m. The scanning angle was set at 60° (Field of View), the average overlap was 40%, and the frequency was 122 Hz (SR). The acquired point clouds have a density of approximately 5 points per square meter. The identification of historic roads was achieved through the processing of ‘raw’ data using a classification filter method, subsequently exported to the LAS (laser) 1.2 format with the support of RIEGL Laser Measurement Systems GmbH and Terrasolid Ltd. [85]. The data obtained from the utilisation of aerial digital photogrammetry was acquired through the employment of an UltraCam Xp, which possesses a geometric resolution of 10 cm on the surface of an overlapping bound, thereby ensuring the provision of 60% and 80% cross-over coverage, respectively. A photogrammetric project was created in the Trimble Inpho software environment (Trimble Geospatial. Office software Trimble Inpho. Available online: https://geospatial.trimble.com/en/products/software/trimble-inpho (accessed on 22 May 2025) to produce a DTM and a DSM. The precision of the aerotriangulation output is contingent upon the accuracy of the DSM and terrain surface [88].

Cross-section profiles of the holloways in the Drahy area were created, and vertical profiles were constructed using the QGIS ProfileTool plugin (QGIS Version 3.42 Python Plugins Repository. Profile Tool. Available online: https://plugins.qgis.org/plugins/profiletool/ (accessed on 22 May 2025). The depth of the holloways was evaluated, as were the distances of the vertices at the highest and lowest positions of the cross-section profiles [88]. The reference data set provides a comprehensive overview of the measured spatial points on the longitudinal and transverse profiles of historic roads in the two areas, under deciduous and mixed stands. Reference points numbered 267 on the road profiles were measured using GNSS via the Slovak permanent observation service. The longitudinal and spatial profiles were focused using a geodetic dual-frequency GNSS receiver (Topcon Hiper GGD, Topcon Corporation, Tokio, Japan). The length and angular relations to each point/item were determined by means of a universal geodetic surveying station (Topcon GPT 9003M, Topcon Corporation, Tokio, Japan). The Slovak permanent observation service provides correction data in real time, enabling the determination of points with a positional accuracy of 0.02 m and a height accuracy of 0.05 m.

In 2023, the scope of the research expanded, owing to the utilisation of a more extensive array of datasets and contemporary technologies. The research involved the processing and utilisation of data from the LiDAR DMR 5.0 dataset, which is publicly accessible and has a pixel resolution of 1 m².

The historical trackways were modelled using the least cost path (LCP) algorithm (QGIS Python Plugins Repository Least-Cost Path. Available online: https://plugins.qgis.org/plugins/leastcostpath/ (accessed on 22 May 2025)), which was employed to compute paths in a DTM. This facilitates the identification of the shortest paths between a single start point and one or more endpoints represented in vector layers. The LCP is guaranteed to be the most economical route relative to the cost units defined by the original cost raster that enters the weighted-distance function. In the early medieval period, the primary mode of transportation was by horse- or ox-drawn carts. The critical slope parameter restricted the feasibility of wheeled transportation, thereby rendering switchbacks a more effective mode of ascent and descent in comparison to direct uphill or downhill routes. A cost function for wheeled vehicles traversing critical slopes was utilised in order to calculate the cost raster. The critical slope value was set to 12% [84]. The initial points were established at the entrances of the historical trackways to the forests, which had been mapped using GNSS. The final point was the gatehouse of the Deserted Castle, Upper Castle. The length of each individual path from the initial to the final point was calculated.

Slope-oriented historical trackways are, in fact, microtopographic hydrological features. The detection of historical trackways in a DTM was achieved through the implementation of two distinct methodologies. Method I entailed the hydrological analysis of slope-oriented sunken linear landforms, while Method II involved the analysis of slope-oriented and contour-oriented sunken linear landforms. The first method involves the extraction of a channel network and the subsequent calculation of the wetness index from the DTM. The objective of Method II is to detect the microtopography of sunken linear landforms by employing a combination of various visualisation techniques.

The extraction of microscale topographic features is facilitated by an algorithm that employs the local relief model derived from a high-detail DTM. The implementation of logical and arithmetic operations, classification, visibility analysis, overlaying procedures and image fusion to a single composite raster resulted in the creation of a useful product that visualised historical holloways. The following DTM raster derivatives were fused: The sky view factor (SVF Documentation QGIS 2.6. Sky View Factor. Available online: https://download.qgis.org/qgisdata/QGIS-Documentation-2.6/live/html/en/docs/user_manual/processing_algs/saga/terrain_analysis_lighting/skyviewfactor.html (accessed on 22 May 2025)) was calculated, independent of the orientation or shape of micro-landforms against Hillshade. The local downslope curvature and the sharpen filter were also applied. The filter function was able to highlight terrain depressions and remove redundant data. Subsequently, a binary image of microtopographic features was created using the geodesic morphological reconstruction module (SAGA-GIS Tool Library Documentation (v6.2.0 SAGA-GIS Tool Library Documentation (v6.2.0). Tool Geodesic Morphological Reconstruction. Available online: https://saga-gis.sourceforge.io/saga_tool_doc/6.2.0/grid_filter_12.html (accessed on 22 May 2025)). Finally, the raster product was smoothed using the generic filter (majority) from the EnMAP-Box Processing Toolbox (Open-source plugin for QGIS EnMAP-Box Processing Toolbox. Available online: https://www.enmap.org/data_tools/enmapbox/ (accessed on 22 May 2025) in order to reduce the presence of what has been termed “the salt-and-pepper effects” in a binary image [86]. The LCP model is a valuable tool for identifying the most efficient trackways, taking into account both travel costs and the slope parameter. In order to validate this hypothesis, buffer zones measuring 50 m were created alongside the GNSS tracks and LCP. The microtopographic features present within the buffers constituting the validation dataset were subsequently vectorised (values above or equal to 1 were utilised for this purpose). The vectorised microtopography of sunken linear landforms was then intersected with the GNSS and LCP vectors. Subsequently, the overlap within the buffer was calculated (Figure 10).

5.1.2. Recognising Agricultural Terraces

The research was conducted utilising topographic maps, with vector data being obtained through manual digitisation. These data were then verified in situ [85,89]. In the cadastral area of Horný Tisovník, agricultural terraces were vectorised from the topographic base maps of the Slovak Republic at a scale of 1:10,000. The presence of agricultural terraces was identified as a distinctive feature in topographic maps, categorised as “artificial anthropogenic shapes defined by an edge”. A meticulous examination was conducted to ascertain the orientation of these terraces, ensuring that they were aligned with the contour direction of the slope. It was further noted that the presence of these terraces along roads was deemed to be indicative of their relationship to the contour lines, as roads are likely to have been constructed through or alongside these terraces. The assessment of the number of terraces was conducted using the following metrics: terrace length (in kilometres) and terrace density (in kilometres per square kilometre).

The researchers compared and evaluated historical and current land use on the terraces in relation to natural conditions and the productive potential of the land for agriculture. The agricultural landscape of the Horný Tisovník cadastral area is distinguished not by fertile soil, but by an extensive system of agricultural terraces that has persisted in the area for a minimum of 250–300 years. This knowledge of the environmental history of Horný Tisovník enables the determination of the optimal future management of the landscape and its land use (Figure 11).

5.1.3. Recognising Agricultural Catchworks

The mapping of catchworks, springs and waterlogged areas was conducted in situ by means of a Leica GS05 GNSS receiver (Leica Geosystems, St. Gallen, Switzerland). The catchworks were identified as longitudinal depressions arranged according to contour lines on the slopes. The DTM3.5 model, with a 10-m grid, was utilised to calculate the altitude above sea level (500–550 m; 550–600 m; 600–650 m; 650–700 m; 700–750 m) and the slope thresholds were set at 0°, 1°, 3°, 7°, 12°, 17°, and 25°. Furthermore, the functionality of the catchworks was evaluated based on the presence of water or plants indicating groundwater, and these were divided into functional, partially functional, and non-functional categories. The field verification of biotopes was conducted in the catchworks [90,91].

In 2021, a methodological innovation for identifying catchworks using a DTM was published [92]. Catchworks can be identified as linear landforms in pixel-based derivatives of a DTM. However, identification is challenging due to the fact that these shallow, narrow, concave linear landforms generally possess edges that are not sharply defined, thus hindering automatic, semi-automatic, or manual detection. Since 2017, the OGCCSR has been engaged in a project for a new, highly accurate LiDAR system (DMR5.0) to cover the entire territory of the Slovak Republic. The estimated date for completing LiDAR coverage of the entire territory of Slovakia was 2023, and the study area had not yet been scanned. Therefore, an ESRI TIN (Triangulated Irregular Network) in the ESRI TIN format was procured from EUROSENSE Ltd. (2017). The ESRI TIN format is a proprietary geospatial vector format developed by ESRI for ArcGIS products. EUROSENSE Ltd. specialists were responsible for the creation of the TIN, which was derived from a series of aerial photographs captured between 2017 and 2019. The resolution of these photographs was measured at 25 cm per pixel. The creation of a digital model was undertaken using the photogrammetric method, with the relevant software, Socet Set^®, having been developed and published by BAE Systems. The accuracy of TIN RMSEz and RMSExy was found to be no greater than 0.6 m in built-up areas, transport corridors in valleys alongside streams, and sites devoid of woody vegetation. The size of catchworks varied in the range of 0.1–0.5 m depth and 0.5–1.0 m width. The model was then subjected to further processing, resulting in the automatic generation of a grid of elevation points. Subsequently, the network of points was manually supplemented with break lines in the 3D environment where terrain edges were visible. Concurrently, the network of points underwent manual editing.

The points were inserted or adjusted to ensure optimal alignment with the terrain. However, a regular grid is required for pixel-based analysis of local morphometric variables in order to detect the microtopography that delineates the edges of shallow channels. Consequently, a DTM was created, utilising a raster with regularly gridded data (Grid) from the original terrain information system (TIN) provided by EUROSENSE Ltd. The grid, with a size of 4.61 m/pixel, was automatically generated. The grid was subsequently filtered using the SAGA Simple Filter tool to accentuate terrain discontinuities. Curvature is a common tool used to identify landforms related to channelised and hill–slope processes in detail. Profile curvature is defined as the parallelism to slope, representing the second derivative of elevation with respect to distance along the line of maximum slope. Positive values are indicative of increasing slope steepness in the downslope direction, which in turn leads to accelerated water flow. Negative values are indicative of a convex surface, which results in a reduction in the velocity of water flowing through the system. The absence of any numerical value indicates that the gradient is linear. The edges of catchworks were assumed to be indicated in each grid cell as the maximum upwardly concave values of profile curvature in that cell. The selection of these grid cells was achieved through the reclassification of the raster as “maximum” and “other values” by employing a raster calculator (Figure 12).

5.2. Identified Landforms

5.2.1. Historical Trackways

High-resolution DMRs are developed using publicly available LiDAR products such as DMR5.0. These are very useful for deriving the exact and precise microtopography of sunken linear landforms under the dense canopy of broad-leaved forests. This is demonstrated by overlaying microtopographic feature segments in the validation dataset with real GNSS tracks and LCPs, which indicate the most efficient paths for medieval travellers. Overlapping segments determined the exact position of potentially preferred trackways, with a total vertical accuracy of 0.25 m or less and a horizontal accuracy of 0.50 m or less. Superimposing the segments of both the GNSS tracks (41,871 m) and the LCPs (20,931 m) represented 39% of the total length [84].

5.2.2. Historical Agricultural Terraces

We observed clear differences in the distribution of terraces within historical territories that were governed by different feudal economic systems and had different natural conditions. The length of the terraces was significantly higher in Modrý Kameň County (9.14 km/km²) than in Divín County (4.40 km/km²). Compared to Divín County, Modrý Kameň County has a lower average altitude, milder slopes, and closer watercourses. Generally, positive correlations influencing terrace distribution exist among ascending distance from watercourses, rising altitude, slope steepness, and units of natural potential vegetation. These are represented by submontane beech forests at lower altitudes and beech and fir forests at higher altitudes. Negative correlations indicate that Carpathian oak–hornbeam forests and current agricultural land use are found near watercourses and on gentle slopes, but not at higher altitudes. The negative correlation between descending watercourse distance and steep slopes shows that the steep slopes of the valley foothills close to watercourses were not used for agriculture [89] (Figure 11).

5.2.3. Historical Agricultural Catchworks

Four situations arose during the vectorisation of catchworks: (1) It was impossible to identify catchworks using adjusted orthomosaics, the Normalised Green Red Difference Index (NGRDI) or buffer zones of break lines. These catchworks only came to light during the final field survey. (2) The buffer zones of break lines helped to delineate catchworks where the contrast of green hues in the adjusted orthomosaic was insufficient and no higher biomass was indicated by the NGRDI. (3) Catchworks were indicated in the adjusted orthomosaics and through NGRDI calculations, but their position was outside the break line buffer zones. (4) Full data coincidence was observed—catchworks were found inside the buffer zones of break lines and corresponded with high NGRDI values. Profile curvature indicated the presence of some catchworks, but the limited data density in this survey was the reason for the lower coincidence values between the maximum profile curvature and the catchworks identified in the field (67%), as well as between the catchworks and the break lines created by EUROSENSE Ltd. 2017. The length of catchworks mapped in the field (1939 m) increased by 2877 m (60%) to 4816.30 m through the delineation of potential catchworks from the rasterised TIN model. These catchworks need to be confirmed in the field [92].

Figure 11. Agricultural terraces in Horný Tisovník: (a) The study area’s location in Slovakia’s traditional agricultural landscapes and the terraces visible in the digital relief model; (b) Different types of terraces analysed in the maps and mapped in the field; (c) Historical maps used to identify the different types of terraces. Source: Adopted from the results of the study published by Slámová et al. [89] and Slámová and Hronček [93]. Source maps: https://geoportal.gov.sk/; https://gis.nlcsk.org/islhp/; http://geo.enviroportal.sk/atlassr/ (accessed on 15 January 2016).

Figure 11. Agricultural terraces in Horný Tisovník: (a) The study area’s location in Slovakia’s traditional agricultural landscapes and the terraces visible in the digital relief model; (b) Different types of terraces analysed in the maps and mapped in the field; (c) Historical maps used to identify the different types of terraces. Source: Adopted from the results of the study published by Slámová et al. [89] and Slámová and Hronček [93]. Source maps: https://geoportal.gov.sk/; https://gis.nlcsk.org/islhp/; http://geo.enviroportal.sk/atlassr/ (accessed on 15 January 2016).

Figure 12. Catchworks in Hriňová: (a) The position of the study area in traditional agricultural landscapes in Slovakia; (b) The maximum profile curvature calculated and compared with catchworks identified using GNSS in the field and with breaklines (rasterised for the purpose of this study); (c) A downward-oriented topographic profile crossing catchworks and different land uses in its segments also includes Natura 2000 (Lk1, Lk5) and nationally significant biotopes (Lk6). Source: Adopted from the results of the study published by Slámová et al. [91,92].

Figure 12. Catchworks in Hriňová: (a) The position of the study area in traditional agricultural landscapes in Slovakia; (b) The maximum profile curvature calculated and compared with catchworks identified using GNSS in the field and with breaklines (rasterised for the purpose of this study); (c) A downward-oriented topographic profile crossing catchworks and different land uses in its segments also includes Natura 2000 (Lk1, Lk5) and nationally significant biotopes (Lk6). Source: Adopted from the results of the study published by Slámová et al. [91,92].

6. Mapping of Woody Pastures

Wood pastures are specific ecosystems that play diverse ecological, agricultural and socio-economic roles. They are grazed habitats characterised by solitary woody plants [94]. Not only do wood pastures provide space and food for livestock, they also create an ecosystem that enhances biodiversity and provides multiple benefits, or ‘ecosystem services’, through their specific combination of elements (perennial grasslands combined with overstory trees that are often pollarded and senescent). Even intensively managed wood pastures provide a variety of benefits and amenities. As well as grazing, they provide various tree products such as firewood, fruits, forage and litter, as well as wild edible plants, mushrooms, honey and space for pollinators [95]. Traditional multi-purpose landscape management has created a mosaic of habitats of different scales across Europe. Recent studies [96,97] have reported a wide range of ecosystem services associated with wood pastures. Local stakeholders valued productive (provisioning) ecosystem services more, while regional stakeholders valued regulatory and supporting ecosystem services more.

These habitats were traditionally part of Europe’s agrarian landscape [98]. They are a set of habitats that differ greatly from region to region, with specific management and structural characteristics, and a strong trend of ecological degradation due to management conversion. As recently as the middle of the last century, these habitats were still relatively abundant in Slovakia [99]. However, changes in land use, coupled with the disappearance of traditional management practices and socio-economic changes, have led to shifts in their spatiotemporal distribution. The absence of appropriate management and subsequent processes of secondary succession has led to their gradual degradation and eventual extinction. Only a few dozen wood pasture sites remain in Slovakia, varying in functionality (Figure 13 and Figure 14), while others have been severely degraded or converted to another type of land use (Figure 15).

The prevailing focus at present is on the restoration of the functions of wood pastures as a component of agroforestry systems. In addition to the aforementioned benefits, this traditional agricultural practice is recognised as an effective means of combating the climate crisis, both in terms of mitigating its effects and adapting to its manifestations [100]. However, they also significantly strengthen the overall resilience of the landscape [101].

A logical outcome of this trend is the increased attention of the scientific community to wood pastures, their research, and also the assessment of their revitalisation options [102]. The determination of the potential for the restoration of wood pastures in the present study was based on three factors: (i) traditional knowledge of the historical presence of silvopastoral ecosystems, (ii) current conditions and distribution, and (iii) defining site presence and selected landscape-ecological conditions in order to identify the potential for the restoration of wood pastures. The decision-making process regarding the future utilisation of wood pastures for agroforestry purposes is significantly influenced by the available data concerning their historical distribution.

6.1. Methods of Wood Pasture Identification Using Satellite Imagery

The identification of relic wood pastures was facilitated through the utilisation of historical records or cartographic documents, or a combination of both. The identification of relic wood pastures was further facilitated through the consideration of features such as the presence of aged (senescent—veteran trees), trees exhibiting indications of historical grazing (e.g., overall tree density, structure and species composition), open or semi-open growth patterns, patchy wood supply, irregular site boundaries, patchiness with frequent clearings and areas with sporadically spaced trees. In the context of typically structured dense forests, the presence of open-growing senescent trees is an unmistakable indication of the area’s historical association with wood pasture or open/rare forest ecosystems.

The historical distribution of wood pastures was assessed on the basis of aerial photographs taken in the 1940s and 1950s. This was the first occasion on which the territory of Slovakia was comprehensively photographed from the air. This particular dataset offers a unique insight into the traditional organisation of the landscape immediately preceding the onset of the large-scale collectivisation and transformation of land management that characterised the communist era.

The research was conducted in the Banská Bystrica and Košice self-governing regions with the objective of identifying intact silvopastoral sites. Wood pastures were identified visually on the basis of characteristic textural features in both areas (16.2 km²), which indicated well-preserved sites. The features in question primarily comprised open tree formations, solitary trees (especially veteran individuals), well-defined boundaries of land use forms, macrosigns of active grazing (especially in the form of eroded trails), and the absence of obvious transition zones into closed-canopy forest.

The presence of core (veteran) trees was observed within each wood pasture. The cartographic input was derived from aerial photographs captured during the 1940s and 1950s. The aerial photographs from the 1950s (black and white, 0.5 m resolution, ©GEODIS Slovakia, Military Topographic Institute Banská Bystrica) and satellite images (©EUROSENSE, GEODIS Slovakia) were utilised as input images. The materials were processed through a WMS server in ArcMap 10.3 (ESRI, Redlands, CA, USA), which was used for all procedures applied in this study.

6.2. Mapping Results

A total of 61 sites of wood pastures were identified, which exhibited varying degrees of disturbance by succession. Within these areas, the presence of vigorous core trees was identified at 36 sites, a finding that is crucial for the realisation of potential restoration in a relatively short period of time. These sites were identified as the primary candidates for restoration, with the restoration of historic conditions being a relatively unchallenging process. This restoration would necessitate only a minimal management intervention, involving the removal of woody debris and the reintroduction of a traditional grazing regime. The restoration potential of the remaining sites is much more limited and would require considerably more effort and time, precisely because of the absence of mature core trees in the typical spatial matrix, which is one of the basic structural and functional features of wood pastures [102]. A thorough analysis of the environmental variables revealed that the selected wood pastures were not established randomly, but rather were situated within a specific landscape context, characterised by altitude, soil quality, and distance from settlements. This knowledge could assist in the prioritisation of sites for the restoration of abandoned silvopastoral ecosystems, with a focus on feasibility, and the selection of those sites exhibiting the greatest potential for successful restoration.

Orthophotomaps and satellite imagery are considered a valuable source of information for the assessment of landscape cover and historical land use change. However, their application in the analysis of wood pastures is constrained by several limitations. The primary weakness of the method is its two-dimensional nature, which precludes reliable capture of the vertical structure of vegetation. This hinders the ability to distinguish between different developmental stages of succession or to identify solitary trees in the grass matrix. The seasonality of imagery, the dependence of such imagery on meteorological conditions (e.g., cloud cover), the limited resolution of satellite systems and shadow effects in sloping terrain can also present a challenge.

6.3. Use of LiDAR Data in Mapping of Wood Pastures

Wood pastures are defined as a unique and valuable ecosystem, combining elements of forest and agricultural landscapes. These ecosystems are distinguished by a sparse stand structure, which facilitates the presence of a diverse array of plant and animal species, thereby resulting in a high degree of biodiversity. Moreover, wood pastures are of historical significance, having played an important role in traditional management in the past, providing a variety of ecological and economic services to rural communities [103]. Conversely, accurate mapping of these structures is extremely challenging, especially given their mosaic spatial heterogeneity. In this context, airborne laser scanning, also known as LiDAR, offers a very powerful tool that allows detailed analysis of the vegetation and terrain structures of these specific ecosystems.

LiDAR technology provides three-dimensional data with high spatial and elevation accuracy, thus facilitating detailed analysis of the vertical structure of stands, identification of individual trees, quantification of vegetation density and recognition of subtle landforms even under vegetation cover. Consequently, it enhances and considerably augments the precision and informative value of conventional image data in evaluating the condition, structure and regeneration potential of wood pastures.

In addition to the detailed analysis of vegetation structure and micro-relief, LiDAR technology also provides significant potential for the identification and assessment of wood pastures in a wider spatial context. The merits of LiDAR data, particularly its high spatial accuracy, capacity to penetrate vegetation cover, and ability to discern subtle variations in elevation, enable more reliable discrimination of open and mosaic structures characteristic of silvopastoral ecosystems. The identification of preserved wood pastures, where solitary and veteran trees, as well as traces of historical management (e.g., grazing, erosion trails, or remnants of old land use boundaries) need to be identified, is facilitated by these features.

The integration of LiDAR data with historical aerial and satellite imagery, as well as environmental variables, facilitates a more comprehensive reconstruction of landscape evolution and a more accurate identification of sites with high restoration potential. In future studies, the combination of visual interpretation (based on textural and structural features), historical image sources, and current LiDAR models is likely to prove particularly effective in identifying sites of wood pastures with varying degrees of succession.

Furthermore, the quantitative analysis of structural characteristics (e.g., stand density, average tree height, vertical heterogeneity) at different sites is enabled by LiDAR data, which is crucial for the process of site stratification and the subsequent prioritisation of restoration measures. Combined with geographic information on elevation, slope orientation, availability and quality of soil resources, it is possible to identify landscape contexts that have historically supported the establishment and maintenance of wood pastures. This knowledge is of inestimable value when planning the restoration of abandoned or degraded silvopastoral systems, as it enables the targeting of management interventions to the most promising sites.

In the future, the potential of LiDAR technology extends to the realm of long-term monitoring of wood pasture dynamics. This encompasses applications such as succession monitoring, management effectiveness assessment, and the identification of novel emergent vegetation forms. Consequently, this technology signifies a pivotal instrument for research and mapping of wood pastures, as well as for their effective conservation, restoration and sustainable utilisation under changing environmental and social conditions.

LiDAR technology provides a point cloud, which facilitates the acquisition of a three-dimensional image of the landscape. It is evident that, in consideration of the aforementioned points, the creation of digital terrain models (DMR), digital surface models (DSM), and specific vegetation height models, such as the canopy height model (CHM), is a feasible undertaking. LiDAR data facilitates the analysis of various characteristics of wood pastures, including the location and number of individual trees, vegetation height, and stand density. In addition, the hydrological characteristics of the environment can be determined [104]. Such an approach facilitates a comprehensive and detailed assessment of the structure of these forest stands, which is crucial for the management and conservation of these valuable ecosystems.

6.4. Methods of Wood Pasture Analyses Using LiDAR Technology

For the purpose of our analysis, which focused on the wood pastures in the Hontianské Moravce area, we used LiDAR data provided by OGCCSR through the ZBGIS portal. The data comprised a point cloud in LAS format, which was processed in the ArcGIS Desktop 10.3.1 environment. The acquired data were then processed further into derived models, including DEM and DMP, which were found to be crucial for the analysis of terrain and vegetation features (Figure 16).

In the initial phase of LiDAR data processing, the location of individual trees within the designated area was ascertained. For the purpose of this study, CHM was employed, which was derived by subtracting DMR from DMP. This methodology facilitates the extraction of vegetation height data and the identification of individual tree tops based on the detection of local extrema (local maxima). This approach is widely utilised in forestry and ecological applications that employ data acquired by airborne laser scanning [105,106]. The employment of this methodology enabled the determination of the number of trees in the study area, a parameter that is imperative for the analysis of stand structure (Figure 17 and Figure 18).

The subsequent stage of the research was to analyse the vertical structure of the vegetation by classifying the height values from the CHM. The distinction between various vegetation types was made on the basis of disparate height thresholds, encompassing grasslands, shrub formations and tree stands. This methodological approach facilitated the acquisition of detailed information on the structure of the wood pastures, which is imperative for the evaluation of biodiversity and ecological characteristics of the landscape. This type of analysis is frequently employed in vegetation structure studies and ecosystem service assessments [107].

In addition, shaded landform models (DMPs and DMRs) were utilised, providing a visual analytical tool for the examination of micro-relief landforms. Shading (hillshading) facilitated the delineation of minute terrain features with ecological pertinence, including elevations, depressions, and diverse geomorphological forms that are frequently the consequence of historical management or natural processes in wood pastures [108]. This information has enabled us to develop a more profound comprehension of the manner in which terrain conditions, in conjunction with vegetation formations, influence biodiversity.

7. Discussion

7.1. Using Historical Maps in Landscape Ecological Research

Historical maps are a unique source of information about past landscapes, how they were organised and how they were used over the last 250–300 years. They depict landscapes that no longer exist as they were viewed by their creators. They provide reliable information on the functional characteristics of landscapes, as well as the relationship between human activity and land use. Their main advantage lies in their ability to offer an unparalleled insight into the past. Disadvantages include thematic limitations, relatively long periods between mapping, and the impossibility of updating or refining them.

Using historical maps for landscape ecological research can present several problems. Firstly, historical maps that are usable for research have only been available since the 18th century, so there is a temporal issue. Secondly, thematic distinction is an issue, as the informative value of a map depends on its original purpose (e.g., economic, military or land maps). The third issue is the map’s resolution, which is related to its scale, and the spatial accuracy, which reflects the mapping technique. In general, older maps are much less detailed than 20th-century maps. Therefore, when analysing the historical development of cultural landscapes, it is important to critically evaluate the value and potential of historical maps alongside remote sensing data. Although these two data sources are fundamentally different, each offers its own advantages.

Despite these limitations, they provide information on the spatial distribution of human activities in terms of land use, and they can be used to analyse the relationship between land use and natural conditions. At the same time, they provide a foundation for analysing cultural landscape development, which can be supplemented and refined using more modern spatial remote sensing data or detailed land surface data (e.g., LiDAR). An appropriately chosen combination of different sources, thoughtfully considered, will enable a comprehensive picture of the past cultural landscape and how it functioned to be constructed.

7.2. Advantages of Using Remote Sensing Data or Derived Datasets for Landscape Ecological Analysis

Classified land cover datasets, such as CLC, high resolution layers and CLC Backbone, are freely available and ready to use. These datasets are also available at the same quality level every six or three years, respectively. Layers showing changes between individual years are also available. Large areas, such as the whole of Europe, can be evaluated simultaneously. An existing transformation table (a matrix of flows) can be used to evaluate changes between years.

However, the minimum mapping unit (e.g., 25 ha in CLC and a minimum change of 5 ha) limits the level of detail in the assessment. It is not possible to identify features such as narrow fields and their composition, boundaries and terraces, or individual trees in grazing forests. These elements create the characteristic image and character of a heritage cultural landscape and could be necessary for identifying its attributes.

Satellite imagery is freely available, and in the case of Landsat, it is possible to go back to the mid-1970s. It surveys a large area at a time and has a short image repeat interval (every 16 days or 5 days, respectively). This enables changes to be observed within a year and automated landscape cover classification methods to be employed. There is also a vast array of online datasets and substantial parallel computing capabilities (e.g., Google Earth Engine [70]). The data is multispectral and different spectral bands and combinations of spectral bands can be used to classify the surface, as well as the calculation of different indices or metrics.

However, Landsat’s resolution (30 m, or 15 m when using the pansharpening function) and Sentinel’s resolution (10 m) only allow basic land use structures and changes of more than 5 ha to be identified and mapped. Some typical features of agricultural landscape heritage, such as terraces, cannot be identified. Overlaying the scene with clouds is possible, but the algorithms for editing satellite images (e.g., atmospheric correction) must be understood, as well as having access to specialised software. There are also issues regarding the spatial and thematic accuracy of interpreted imagery [65].

The resolution of orthophotomaps in Slovakia has improved from 0.5 m in the 1950s to 0.15 m today. They have been freely available since 2017 and are updated every three years. From 2020, it will be possible to calculate the NDVI index using 4-channel images. More detailed structures will also be identifiable, such as narrow fields and their composition, boundaries and terraces, as well as individual trees in grazing forests. This information is important for identifying the elements and structures necessary for mapping the heritage of the cultural landscape.

However, orthophoto images from before 2017 must be purchased or accessed via a web browser. Freely available orthophoto images take longer to obtain than satellite data. It is usually not possible to map changes during the year. There are fewer spectral bands (max. 4), which means it is not possible to calculate multiple indices. An orthophoto image covers a smaller area than a satellite image. At the same time, the image is obtained for a significantly smaller area.

7.3. Using LiDAR Data for Mapping Historical Landforms

Since 2017, LiDAR classified point cloud has been freely available across Slovakia. It offers an elevation accuracy of up to 0.05 m, a positional accuracy of up to 0.16 m, and an average of 20 points per sq. m. It can be used to map vegetation and buildings, create a digital terrain model, and identify roads and terraces. It can also be used to determine the height of objects (e.g., vegetation and buildings) on the surface in a simple way. It can also be used to map historical structures hidden under vegetation or invisible to the naked eye (minimal changes in altitude), which are revealed thanks to a detailed digital terrain model [14,108,109]. Three-dimensional models of vegetation and buildings can be created and parameters such as the number of canopy layers, forest density, crown width, crown height and crown shape can be derived from them. This makes it possible to map the absolute details of the landscape’s composition and structure.

Nevertheless, it requires the necessary knowledge to work with LiDAR data for more complex analyses. It is a very large dataset and takes up a lot of disk space. When working with it, it is necessary to divide the area into tiles.

Over the past decade, research published by Chudý et al. [88] and Slámová et al. [84] has marked a technological advance in the use of detailed LiDAR products for producing sub-metre accurate digital terrain and surface models. The studies’ findings indicated that LiDAR and photogrammetric methods demonstrated comparable levels of accuracy in microscale landform research. Specific components of historical trackways were reconstructed using a validation dataset that employed GNSS tracks and LCP with sub-metre precision. The findings showed that using a high-resolution DTM derived from LiDAR was an effective way to identify historical trackways, even in a forest environment.

On the other hand, the declared positioning accuracy of 3 m for point positioning, as stated by the manufacturer, was exceeded several times during mapping, with tracks showing even greater inaccuracy. GNSS mapping is efficient in the initial phase of recognising or verifying trackways in the field. Further analyses can be performed using precise, high-resolution digital models [84]. In the context of aerial photogrammetry, the presence of vegetation, particularly in broadleaved forests, must be considered to account for data gaps under tree canopies. It is important to note that each model represents generalised results. Misinterpretations of accuracy may also stem from random errors occurring during DEM generalisation. It is impossible to eliminate these errors entirely.

Earlier studies in environmental history were often lacking in scientific credibility. History was mainly interpreted from written materials, but detailed data was limited in terms of reconstructing past environmental relations. Therefore, we introduced geospatial analysis combined with quantitative statistics to examine the context of terraces and relevant factors. Our research into the spread of agricultural terraces in Horný Tisovník [89] has provided new insights into their emergence and spread in specific cultural, historical, and natural conditions. We employed innovative methodological procedures to test hypotheses concerning the origin and distribution of terraces within the cadastral territory and historical estates, utilising GIS and multivariate correlation analysis of selected factors [93].

However, ordination methods are typical of biogeographic studies. Recently, multivariate statistics have also gradually disseminated within the field of landscape ecology. In the field of environmental history, this involves firstly collecting geographic and historical data from various sources, processing it in GIS, and preparing a dataset for statistical analysis. This is a lengthy and time-consuming process requiring precise selection of the variables to be used in evaluating correlations among the factors influencing terrace distribution [89].

Identifying micro-scale landforms requires precise and detailed DTMs to be generated. Integrating data collected through a range of different technologies is essential. DTMs and land cover interpretation were integrated with the NGRDI. However, the vegetation index under consideration is not the most suitable for differentiating between living green vegetation and dried herbaceous vegetation. This is because the numerical difference between the green and red bands has lower thresholds than the numerical difference in the near-infrared red-green-blue imagery (NIR-RGB). In this study, the RGB orthomosaic was obtained free of charge, unlike the NIR-RGB aerial imagery. It is important to note that vegetation indices are most effective in identifying catchworks when the soil humidity within catchworks is higher than in surrounding areas [92].

Conversely, the investigation of catchworks was limited by the availability of precise, detailed terrain data. Catchworks were detected using terrain data interpreted within a TIN model. EUROSENSE Ltd. conducted the delineation of break lines. This process helped to identify potential zones where catchworks could occur in the terrain. A DTM was created from the TIN using interpolation algorithms, with the inverse distance weighting and nearest neighbour methods applied to generate a regular grid. However, it was determined that the point density (0.014 points/m²) and grid pixel size (4.61 m/px) were inadequate for detecting the microtopography of catchworks [92].

7.4. Pros and Cons of LiDAR Data in Vegetation Mapping

Despite the fact that LiDAR provides highly accurate and detailed data, it is imperative to consider the inherent limitations of the technology. A significant constraint pertains to the paucity of data concerning the species composition of vegetation, a crucial element in the ecological evaluation of landscapes. LiDAR technology provides accurate data on vegetation structure; however, it does not capture information on plant species, which is essential for a comprehensive ecological assessment. Consequently, it is recommended to combine LiDAR outputs with other data sources, such as orthophotomaps, and to supplement them with classical botanical and dendrological monitoring, which provides a more comprehensive picture of ecosystem structure [110].

Satellite imagery and orthophotomaps provide valuable insights into the spatial layout and historical distribution of wood pastures, particularly through the visual interpretation of landscape texture and tree patterns. Their large-scale coverage and the availability of historical data (e.g., from the 1940s and 1950s) make them essential for long-term landscape analysis. However, they are limited by their two-dimensional nature, which prevents the detection of vertical vegetation structure and the characteristics of individual trees. Image quality is also affected by resolution, seasonality and weather conditions during data capture. LiDAR and orthophotomaps complement each other in identifying wood pastures. LiDAR provides high-resolution 3D data, enabling the detection of tree height, canopy structure and micro-relief features, even under dense vegetation. This is crucial for distinguishing veteran trees, open structures, and succession stages. Orthophotomaps, on the other hand, offer clear visual context and historical comparisons. However, LiDAR lacks species-level information and orthophotomaps remain limited to 2D interpretation. Together, they enhance accuracy, but field validation is required for ecological assessments. Due to its high spatial and vertical resolution, LiDAR technology generates very large datasets. These datasets require substantial storage capacity, computational power and specialised software for processing and analysis. While this enables highly detailed mapping, it can present challenges in terms of data handling and accessibility, particularly in large-scale or resource-limited projects.

In conclusion, LiDAR is a highly effective tool for acquiring precise and objective data on the structure of wood pastures. Its application is pertinent to the conservation, restoration and effective management of these ecologically valuable landscape features, which play a significant role in the sustainability of the landscape and its biodiversity.

8. Conclusions

Although the study of cultural landscapes has been a traditional geographical research topic for almost a century, it continues to evolve conceptually and methodologically. Technological and digital advances are providing new data on the landscape and new ways to process and interpret it. However, the challenge of combining traditional methods based on historical maps and field mapping with modern technologies remains. New data from remote sensing can significantly improve the resolution and detail of research. At the same time, they enable new, updated data to be supplied at short intervals, significantly improving the early detection of changes to the cultural landscape and the identification of trends in its development or potential threats. In our contribution, we presented traditional landscape ecology methods for researching cultural landscapes, which are based on historical map analysis, and critically evaluated their interpretive value. The increased use of historical maps is facilitated by their growing online accessibility as web map services and by the current focus of research activities on cultural landscape development. Detailed historical map analysis allows us to study all phenomena occurring in a landscape, thus revealing processes that would otherwise remain hidden or appear isolated.

We also presented up-to-date remote sensing data sources, such as aerial and satellite images, as well as LiDAR data, and demonstrated their high informational value and accuracy compared to older sources. Through case studies, we demonstrated the effectiveness of LiDAR data in identifying small-scale landscape features, such as historical roads, agricultural terraces, and catchment systems. LiDAR data can also be used to detect and monitor changes in tree vegetation due to the overgrowth of valuable relics of the agricultural landscape, such as wood pastures. This integrated research approach can be perceived as a multitier landscape model, in which historical maps provide spatial context and temporal depth, while aerial and satellite images provide a spatial base layer for cultural landscape assessment. This layer can later be enhanced with highly precise surface/terrain data (such as LiDAR) to capture smaller landscape features.

Analysing the development of the cultural landscape is essential for understanding its internal laws, which determine its future development potential. Incorporating information on past landscape situations, land use and ecosystem distribution into practical nature conservation management takes into account the important factor of time. This information can be used to evaluate the authenticity of ecosystems, determine or adjust the boundaries of nature reserves or biosphere reserves, and restore ecosystems following past negative changes or natural disasters.