Joining Application of Unmanned Aerial Vehicle Imagery with GIS for Monitoring of Soft Cliff Linear Habitats

Abstract

In the coastal zone, two types of habitats—linear and areal—are distinguished. The main differences between both types are their shape and structure and the hydro- and litho-dynamic, salinity, and ecological gradients. Studying linear littoral habitats is essential for interpreting the ’coastal squeeze’ effect. The study’s main objective was to assess short-term behavior of soft cliffs as littoral linear habitats during calm season storm events in the example of the Olandų Kepurė cliff, located on a peri-urban protected seashore (Baltic Sea, Lithuania). The approach combined the surveillance of the cliff using unmanned aerial vehicles (UAVs) with the data analysis using an ArcGIS algorithm specially adjusted for linear habitats. The authors discerned two short-term behavior forms—cliff base cavities and scarp slumps. The scarp slumps are more widely spread. It is particularly noticeable at the beginning of the spring–summer period when the difference between the occurrence of both forms is 3.5 times. In contrast, cliff base cavities proliferate in spring. This phenomenon might be related to a seasonal Baltic Sea level rise. The main conclusion is that 55 m long cliff cells are optimal for analyzing short-term cliff behavior using UAV and GIS.

1. Introduction

Two types of littoral habitats should be distinguished in the coastal zone—linear and areal. The concept of linear littoral habitats, first proposed by Robles-Diaz-de-León and Nava-Tudela [1] and developed by the coastal research team at Klaipeda University, Lithuania [2,3,4,5], has garnered significant international interest in recent years. It has been applied by a worldwide coastal research community on various coastal management aspects—from coastal dune conservation [6] to coastal forest [7] and fishery management [8].

Referring to the broader international context and relevance of this research study, the main postulate of the concept of linear littoral habitats is about the fundamental morphological and ecological differences that distinguish them from areal habitats. This contrast is an essential research tenet. The main features of the linear coastal habitats are the following:

• Shape: Linear littoral habitats are narrow habitats characterized by a vast longitudinal range compared to a relatively thin width.
• Structure: Linear littoral habitats are relatively homogenous. It contrasts with areal littoral habitats, which are typically patchy.
• Distinctive gradients: The boundaries of linear littoral habitats are defined by steep and distinctive hydro- and litho-dynamic, salinity, and ecological gradients, which feature the complexity of these ecosystems.

Understanding linear littoral habitats is vital for responding to problems such as the ‘coastal squeeze’ effect in modern coastal evolution [9]. Pontee [10], p. 206, defines coastal squeeze as one form of coastal habitat loss, where intertidal habitat is lost due to the high-water mark being fixed by a defense or structure (…) and the low-water mark migrating landwards in response to sea level rise. It threatens the integrity of linear littoral habitats by breaking and fragmenting these narrow, long-shore strips. It is acute, where artificial consolidation on the coastline affects habitats and ecosystems that typically retreat landward due to coastal erosion [11,12].

The growing tourism pressure on linear littoral habitats causes coastal development for urban sprawl, tourism, recreation, and industry [13]. Narrow sandy beaches are particularly vulnerable and threatened to disappear due to coastal squeeze [14]. Sandy coasts are highly developed and densely populated due to their recreational amenities and visual appeal. Erosion of these coasts over the last few decades has already resulted in coastal squeeze [10,15]. This study focuses not on soft cliff erosion per se but on a broader, comprehensive consideration of cliff behavior in peri-urban protected shorelines, which simultaneously are attractive tourist destinations [16,17].

In 2022, the authors of the present study investigated current main trends in two interrelated complex research fields pertinent to the objectives of the study: (i) biodiversity conservation and tourism sustainability [18], and (ii) applications of remote sensing for coastal and marine nature conservation [19]. By applying hierarchical cluster analysis with a KH Coder 3.0 tool, major trending research themes on biodiversity conservation and tourism sustainability were identified, including community-based tourism development [20,21,22], national park management for tourism [23,24,25], sustainable tourist motivation [26,27,28], and biodiversity conservation and ecotourism [29,30,31].

Similarly, using the same technique, four trending research themes in applications of remote sensing in coastal and marine conservation were identified, including remote sensing-based classification and modelling [32,33,34], conservation of tropical coastal and marine habitats [35,36,37], mapping of habitats and species distribution [38,39,40], and ecosystem and biodiversity conservation and resource management [41,42,43]. The objectives of our study stem directly from the above findings as summing up the identified key research directions; joining the application of unmanned aerial vehicle (UAV) imagery with GIS for monitoring coastal habitats in protected areas emerges as one of the resulting pivotal research directions.

The central insight is an increasing coastal habitat vulnerability resulting from growing external impacts like climate change and sea level rise. Thus, if a littoral linear habitat happens to be an attractive peri-urban beach or a soft cliff prone to the combined adverse effects of the urban sprawl, sea level rise and coastal squeeze, coastal managers risk appearing in a ‘perfect storm’ situation regarding coastal cliff behavior. In this situation, the ad hoc flexible application of UAVs owned by coastal managers is convenient for vulnerable littoral habitat monitoring [44].

Furthermore, ArcGIS is a convenient spatial interpretation toolbox available at many coastal conservation and management institutions. Therefore, our study aimed to test the capabilities of the combined use of UAV imagery and ArcGIS to survey and interpret the behavior of linear littoral habitats that are popular tourist attractions in the calm seasons of the year. Olandų Kepurė is a typical soft cliff, i.e., ‘formed through the exposure of rocks that have little resistance’ [45] p. 2, in this case, deposits of glacial till and sandy clay. It is well suited to investigate and understand the behavior of linear littoral habitats under complex stress and its implications for coastal management.

Although the application of UAVs is limited for monitoring in wide and long coastal zones due to stability issues and other technical limitations, for a narrow, ca. 50 m wide, 1.5 km long stretch of a linear littoral habitat like Olandų Kepurė, it is more flexible to study short-term behavior than using satellite data. Hence, we hypothesized that it is possible to study the short-term behavior of the soft cliff as a linear littoral habitat if sufficient precision of remote sensing is achieved and a 1-D interpretation using ArcGIS is applied.

2. Materials and Methods

2.1. Study Area

Considering the diversity of South Baltic coastal habitats and patterns of seaside recreational resources, the South Baltic coastal region can be divided into three sub-regions (Figure 1):

• Southeast Scandinavian coast and islands;
• South Baltic coast and islands;
• Southeast Baltic graded coast.

The morphological and habitat features of the Southeast Baltic graded coast resulted from the post-glacial Baltic Sea level fluctuations combined with the sediment input from large rivers, erosion of moraine promontories, and a longshore marine sediment drift [46]. Here, the graded coast is defined as the one where the headland erosion and deposition of eroded sediments in adjacent embayments grades the shoreline configuration in the long term [47]. Our study area is one of such Southeast Baltic coastal features—the highest sea cliff in Lithuania—Olandų Kepurė, ca. 25 m high at the highest point. Its annual erosion rate ranges from 0.5 m to 2.2 m.

The Olandų Kepurė sea cliff is protected as a landscape reserve within an IUCN category V peri-urban Littoral Regional Park (Figure 2). It is a littoral linear mega-habitat comprising a shoreline-parallel series of the underwater boulder belt, nearshore sand belt, gravel strip, sand beach, cliff base and cliff slope (Figure 3). The areal part of the Olandų Kepurė landscape reserve includes several habitats related to the prevailing mature, up to 120-year-old Scots’ pine forests of a local Eastern Baltic race (Pinus sylvestris var. rigensis Loudon). Also, over 200-year-old English oak (Quercus robur) trees and over 120-year-old Norway spruce (Picea abies) trees grow in the Olandų Kepurė nature reserve, constituting its landscape and ecosystem value [48].

The Olandų Kepurė sea cliff is a 1.5 km long stretch of the Baltic Sea coast north of Klaipeda, the third-largest urban agglomeration of Lithuania (Figure 4). It is the most popular single-day outdoor active visitor attraction on the Lithuanian Baltic Sea coast. A total of 0.4 million visitors annually visit the cliff. One of the essential management tasks of the Olandų Kepurė landscape reserve is to optimize tourism infrastructure to allow visitors to enjoy the park’s natural and cultural heritage values combined with coastal and maritime tourism. Therefore, it is necessary to investigate the cliff behavior in the relatively calm seasons more appealing to visitors.

Here, we define the calm season as a protracted period of the year when the probability of strong winds with an average wind speed over 20 m/s is below 10% [49], and weather conditions at the Olandų Kepurė cliff are formed predominantly by breeze circulation [50]. The calm season in the South Baltic Region is usually between April and August [51], but it varies from year to year. Respectively, calm season storm events are periods with a wind speed over 20 m/s lasting for one week in the calm season related to North Atlantic cyclones originating in the east Atlantic and passing over the southeast Baltic coast [52].

According to the information from the Ranger’s Office of the Littoral Regional Park [53], the climate features of the Olandų Kepurė landscape reserve in the current climatic standard norm period (1991 to 2020) are typical for a temperate Atlantic climate coastal region featured by mild winters, more clear days, more frequent thunderstorms. The annual number of cloudy days is 98 to 111 days per year. About 100 sunny days occur in the warm season (i.e., May–September). The average annual precipitation rate is 696 mm, of which 63% falls in the warm season (April to October).

The average air temperature ranges from +5 °C to +12 °C in spring, + 14 °C to + 17 °C in summer, +12.5 °C to +5 °C in autumn and −4 °C to −1 °C in winter. The Baltic Sea tidal range at Olandų Kepurė is 3.5 cm. Prevailing SW and SE air masses shape the dynamic conditions in the Baltic Sea nearshore. The Olandų Kepurė cliff is exposed to the longest fetches of the severest autumn and winter storms and wave directions (SW, W and NW). The range of the seasonal change in the water level of the Baltic Sea caused by the wind is 22 cm to 28 cm.

2.2. Research Overview

Figure 5 summarizes the research methodology, workflow, and the interrelation of steps. The assessment of cliff behavior along the entire 1.5 km length of the Olandų Kepurė sea cliff during the calm season storm events used a blend of qualitative and quantitative methods, resulting in a mixed technique proven productive in other coastal GIS-based studies [54,55,56,57,58,59]. The methodology involved pilot cliff monitoring using UAV and data analysis using an ArcGIS algorithm adjusted for linear habitats, accompanied by modelling waves and currents in the storm events before the remote surveillance sessions and the Delphi procedure for identifying the sites of cliff scarp slumps and cliff base cavities.

The pilot cliff monitoring using UAV was for collecting the visual database of the georeferenced sites of the erosion sites. The aim of this pilot monitoring stems from the overall study goal to test if UAVs can be applied for coastal monitoring in a peri-urban protected area, which is a popular tourist destination. The challenge of data input was to optimize the flight parameters for collecting the database of the discernable images of cliff scarp slumps and base cavities. The essential aspect of data output was the visual assessment of taken images regarding resolution, the area covered, and the detail of objects in the image.

As the database of erosion sites was too small to train the AI visual recognition models for discerning the cliff scarp slumps and cliff base cavities, the authors employed professional photo interpreters for this task using the Delphi technique. It is a reliable method to help reach expert consensus, especially in fuzzy situations. It has been successfully applied worldwide, including by the authors of this study [60]. The initial data input was over 400 UAV-made images provided by the organizer to the panelists as the visual priors. The data output was a georeferenced database of cliff base cavities and scarp slumps, which experts discerned with an 80% consensus.

Since there are no buoys or other hydrodynamic monitoring gauges in the vicinity of Olandų Kepurė, the next step was to apply hydrodynamic modelling to obtain data on water circulation, waves, and water level during the investigated storm events. The 3-D simulation model was applied to simulate and understand spatiotemporal hydrodynamics of the Baltic Sea near the Olandų Kepurė cliff during three investigated storm events. The input data were obtained from the Klaipeda Maritime Observatory, and the output data were simulated hourly hydrodynamic parameters at Olandų Kepurė during three storm events (25 February–3 March 2023, 1–7 April 2023, and 18–24 August 2023).

The final step of this study was to apply a Geographic Information System (GIS) to interpret the spatial distribution of the cliff scarp slumps and base cavities, thereby determining their distribution’s essential patterns. As academic literature highlights, GIS offers many coastal monitoring and conservation benefits [61,62,63,64]. This study applied the GIS software ArcGIS10.8^® by ESRI^TM (Redlands, CA, USA) to determine the short-term spatial distribution patterns and changes in sea cliff behavior. The input data were quarried from the georeferenced database of the spatial distribution of the discerned erosion sites after each of the three surveyed storm events. The application of GIS by geostatistical analysis enabled discerning the concentration zones of the erosion sites along the cliff’s length.

2.3. Monitoring Using UAV

A UAV Phantom 3 Professional was used, which supports GPS geolocation and is equipped with a Sony EXMOR 1/2.3” camera with a resolution of 12.76 million pixels. The first stage of the image collection included three test flights (on 22 February 2023—twice, and on 27 February 2023) at altitudes of 70 m, 40 m, and 25 m. The objective of the test flights was to find the optimal flight parameters that would be most suitable for collecting the database of the cliff images. The chosen strategy was to start flying from an altitude of 70 m, gradually reducing it to 40 m and 25 m.

The chosen strategy was to start flying from an altitude of 70 m, gradually reducing it to 40 m and 25 m. Since the Olandų Kepurė cliff height reaches 25 m, the minimum flight height of 25 m was reasonable. During each test flight, at least 25 test images were taken. We further assessed the visual quality of the obtained images. Considering that cliff scarp slumps and base cavities might be 1 m in diameter, it was decided to use the flight height of 25 m for further image database collection. The visual wavelength spectrum was applied in the UAV surveillance. We conducted the surveillance flights to collect the image database after the storm events on 3 March 2023, 7 April 2023, and 24 August 2023.

2.4. Delphi Technique

A group of five experts in coastal geomorphology, coastal management, integrated coastal and maritime planning, and coastal conservation, with good knowledge of soft cliffs and who did not interact directly with each other, accomplished the Delphi study by achieving consensus judgment. There is no single opinion regarding the optimal number of panelists in the Delphi method [65], with heterogeneity being an essential factor in eliciting reliable responses [66]. In a systematic review of Delphi studies, Boulkedid et al. [66] found the minimum number of panelists to be three.

First, the experts were introduced to the proposed method of the coastal erosion site, discerning from the UAV-made images and the possible variety of the visual shapes of cliff base cavities and scarp slumps. Then, experts were asked to discern the cliff base cavities and scarp slumps in the images. Next, the organizer collected, summarized, and returned the first judgments to the panelists for further evaluation. These judgments included the responses of other panelists, allowing each expert to scrutinize the opinions of others and adjust their own. In the next two rounds, the study organizer sent the revised judgments, with feedback from other panelists, back to the experts.

For this study, the experts performed three rounds of judgment using an approach drawn from previous Delphi studies in landscape management found in the literature [67,68,69,70,71,72,73,74]. The Delphi study was carried out from February to May 2024 using e-mail for communication within the panel of five experts. During the feedback, there was an apparent tendency among the panelists to converge their opinions and seek a consensus. Thus, the research team successfully narrowed the experts’ judgments. After the third round, the respondents reached a satisfactory consensus of 80% of their opinions regarding the discerning of the cliff scarp slumps and cliff base cavities.

2.5. Nearshore Wave Height, Surge Level and Longshore Current Modelling

We employed a 3-D Shallow water HYdrodynamic Finite Element Model SHYFEM [75], an open-source hydrodynamic modelling platform (http://www.ismar.cnr.it/shyfem, accessed on 1 October 2024), to simulate the hydrodynamic parameters. SHYFEM was developed at CNR-ISMAR, a marine research institute in Venice and successfully applied to many shallow water environments [76,77,78,79,80,81,82,83,84,85]. SHYFEM simulates physical variables, such as water circulation, waves, water level, and temperature, necessary to characterize the shallow water spatiotemporal dynamics, and resolves shallow water hydrodynamic equations for lagoons, coastal seas, estuaries, and lakes. The model uses an unstructured grid (finite elements) to discretize the studied spatial aquatic domain. The finite elements are essential in simulating nearshore hydrodynamics [85].

These finite elements allow for the model’s better adaptation to the system’s morphology and bathymetry, ensuring a more accurate representation of real-world conditions [82]. A detailed description of the model and illustrations of its 2-D and 3-D applications can be found in [82,83,84]. The model calculated sea levels, wave heights, longshore current direction, and speed for these periods in the Baltic Sea nearshore up to a 20 m depth contour varying from 250 m at Olandų Kepurė to 10 km in the Baltic Proper. The atmospheric forcing was interpolated from a regular regional climate model data grid into the finite-element nodes by bi-linear interpolation using the algorithm outlined in [85].

2.6. Application of GIS

ArcGIS 10.8^® by ESRI^TM was chosen since it has proved to be an efficient toolbox for versatile analytical purposes [64,86,87,88,89,90]. However, our purpose was quite specific: to investigate the behavior patterns of linear littoral habitats. For this purpose, two key assumptions were made. (1) For spatial analysis, linear littoral habitats with a considerable ratio of length to width can be treated as linear objects, and (2) the main results of the study can be represented in a 1-D coordinate system without considering the curvatures of the linear object. Using linear planimetry for GIS applications in soil erosion management and coastal planning is a new field of geomorphological and geographical studies [91,92,93].

Nevertheless, creating a 1-D georeferenced linear planimetry is challenging, considering linear littoral habitats’ spatial curvature and gradients. Therefore, the next step was to establish a georeferenced database of the erosion sites along the 1.5 km long Olandų Kepurė cliff using ArcGIS 10.8. The 2-D spatial data were converted into the 1-D georeferenced linear data using our original interpretation algorithm based on the sequence of applying several ArcGIS tools, considering the linear distribution of coastal erosion. We developed the algorithm to extract intended 1-D data using ArcGIS 10.8 Toolkit functions of Points to Line, Generate Points Along Lines, Split Line at Point, Spatial Join, and Snap.

Lastly, using the Attribute Table option in ArcGIS 10.8, the data were exported into a Microsoft Excel format and later used to perform additional calculations. In this way, a GIS database of the research sectors of the Olandų Kepurė cliff was created, which contains information about the center points of coastal erosion processes and their localization. Finally, standard ArcGIS geostatistical tools were applied to interpret the results of our investigation. For this purpose, we calculated the erosion site concentration coefficient K using a standard ArcGIS geospatial analysis tool, Point Density, adapted from 2-D to 1-D structure analysis. It indicates how many erosion sites are present in each cliff cell.

3. Results

3.1. Spatial Behavior Patterns of the Olandų Kepurė Cliff

A ‘cliff scarp slump’ is a cliff erosion case occurring when a coherent bulk of unconsolidated sediments moves down a slope for a short distance [45,94]. The term ‘cliff base cavity’ defines the coastal erosion case in which the hydraulic action of waves during a storm surge forms concave basal hollows of various sizes and duration [95,96,97]. The cliff base is a fuzzy, dynamic interface between the cliff surface and the foreshore [45].

In total, 662 cliff scarp slumps and 348 cliff base cavities were identified in all the investigated cases after three calm season storm events (Figure 6). The geostatistical analysis made it possible to distinguish the concentration zones of the erosion sites along the length of the Olandų Kepurė cliff. For this aim, the erosion site concentration coefficient K was calculated using the standard ArcGIS geospatial analysis tool Point density adapted from 2-D to 1-D structure analysis. It shows how many erosion sites are present in different cells when assessing the condition of various cliff cells.

Figure 6 shows that in all three investigated cases, the cliff scarp slumps occurred relatively evenly along the entire length of the investigated coastline. At the same time, the cliff base cavities were more concentrated in the northern part of the coastline. It can also be seen that in all three cases, the locations of cliff scarp slumps and base cavities and their distribution over the surveyed calm season are unrelated. If erosion occurred in different cases at some sites of the cliff base, then the geographical distribution or intensity of the cliff scarp slumps did not depend on this at all.

The increase in erosion in the northern part of the cliff, rather than with the distribution of slumps, correlates with the intensification of the long-term retreat of the cliff from south to north. In 1954–2003, the previously stable cliff in the northern part of the Olandų Kepurė section retreated by ca. 30 m to 50 m, i.e., 0.6 m to 1 m per year [48]. Meanwhile, at the highest central section of the cliff, the shoreline retreat remained minimal. The intensification of cliff base erosion and long-term shoreline retreat from south to north is associated with the predominance of more erosion-resistant moraine deposits and their large volume in the highest (up to 25 m high) central part of the cliff.

3.2. Spatial Patterns of the Cliff Scarp Slump Distribution

A cliff scarp slump is a common form of mass wasting when loose sediments lie above more cohesive layers such as glacial till, like in the case of the Olandų Kepurė cliff. However, because of their small scale (typically up to 10 m in length and up to 150 m³ in mass) and their ephemeralness, it would be better to label the slumps of the Olandų Kepurė cliff as micro-slumps (Figure 7).

Despite the visually ostensible, quite regular spatial distribution pattern apparent in Figure 6, the distribution of slumps in the study area is far from uniform. The erosion site concentration coefficient K, describing the occurrence of slumps along the Olandų Kepurė cliff scarp, varies significantly, ranging from 0.27 to 1.73 (Figure 8). Regardless of the season, the spatial distribution of slumps along the Olandų Kepurė cliff scarp is very uneven—several investigated cells in the north (on the right in Figure 8), where the cliff is the lowest, feature only a few slumps, i.e., K was equal to 0.27.

Our research also indicates a clear seasonal pattern in the occurrence of slumps on various strips of the Olandų Kepurė cliff scarp (Figure 8). Their probability decreases in the spring to early summer period, and a particularly noticeable decrease is in the second part of this period. If the average K value decreased slightly (−0.01) from March to May, from June to August, it decreased significantly (−0.12). This volatility in the seasonal development of cliff scarp slumps is particularly evident in the central, i.e., the highest part of the cliff, where the geomorphology of the cliff is the most complicated [98].

3.3. Spatial Distribution Patterns of the Cliff Base Cavities

Like in the cases of the cliff scarp slumps, in the case of the Olandų Kepurė cliff, basal cavities would be better labelled as micro-cavities due to their minuscule size (up to 1 m high) and ephemeralness (Figure 9). Yet, one should not underestimate their geomorphological role. Even small cavities forming during the storm surge undercutting the cliff base can induce cliff failure [95,96,97].

As mentioned, our survey’s most evident quantitative result is that cliff base cavities occur much more seldom than slumps in all three investigated post-storm cases (Figure 10). This feature was especially noticeable in the spring when the occurrence of slumps was 3.5 times more frequent than the occurrence of cliff base cavities. In contrast to slumps, cliff base cavities occurred most frequently in August. The average erosion site concentration coefficient K value, specifically for the concentration of cliff base cavities in all the investigated cells, was 0.26 in March. From April to August, it increased to 0.44.

Meanwhile, cavities formed by storm waves in the Holocene sand and sandy clay deposits to the south of the cliff’s glacial till strip need to persist longer to be spotted by a UAV. Remarkably, though, even further south, in the southernmost part of the Olandų Kepurė cliff, where sand and sandy clay deposits form the cliff’s base, cavities occurred and were recorded after the storm in August. In this regard, the geomorphological consequences of the late summer storm in terms of the formation of cliff base cavities significantly differ from the consequences of early spring storms.

3.4. Nearshore Hydrodynamics at the Olandų Kepurė Cliff

The volatile calm season development of the Olandų Kepurė cliff scarp slumps and base cavities and the spatiotemporal disentanglement of these two erosion processes feature the complex nature of coastal erosion drivers. The increase in the cliff base cavity development from April to August is most probably related to the slowing down of other erosion processes, which caused relatively more active erosion of the cliff base. However, the Olandų Kepurė cliff is in the temperate Atlantic climate coastal region [99]. Therefore, besides the wave action, other drivers influence the cliff’s geomorphological processes in the calm season, especially in the early spring.

Although wave hydraulic action is often the most pivotal driver of soft cliff erosion, the results of the hydrodynamic modelling of three calm season storm events at the Olandų Kepurė cliff show that during the storms of the calm season, it may not be the case even if the cliff faces westerly or southwesterly wave fetch.

Table 1. Modelled parameters of the nearshore hydrodynamics at the Olandų Kepurė cliff.

Storm Event		02.25–03.03.2023	03.31–04.06.2023	08.17–08.23.2023
Parameter
Wave height (m)		0.22	0.15	0.17
Wave direction		249°	138°	261°
Water level (m +NN)		0.27	0.19	0.05
Surface current speed (m/s)		0.02	0.04	0.02
Surface current direction		172	197	188
1 m deep current speed (m/s)		0.02	0.02	0.02
1 m deep current direction		226	223	188

shows that all three investigated calm season storm events created only a minimal storm surge caused by low wave energy, which dissipated in the foreshore covered in moraine boulders, pebbles, and gravel. Low-energy wave action was due to a seasonal low water level in the Baltic Sea, which is usually in summer.

In the calm season, characterized by the low seasonal water level, a shallow nearshore plays an essential role as a wave energy damping factor. Only the highest, low-probability waves reached the cliff base, creating basal cavities. The nearshore currents had low energy and could not carry away eroded sediments from the foreshore. Such a low-energy wave action explains the disentanglement between the spatial distribution of the observed cliff base cavities and scarp slumps. Therefore, slumps must result from other erosion drivers, most probably seepage caused by rainfall during the storm events.

The results of the hydrodynamic modelling also show that the nearshore currents play an important role during the storms in calm seasons in transporting the eroded sediments from the coast to the nearshore, supplying suspended sand and gravel particles to the resulting northbound longshore littoral sediment drift. It is the pivotal process for ensuring the integrity of the cliff and the adjacent accretion areas as a linear littoral mega-habitat and a littoral cell with erosion and accretion strips [100]. Notably, these litho-dynamic features are coherent with the processes observed in the same South Baltic soft cliffs during autumn and winter stormy periods [101].

The 1455 m long Olandų Kepurė cliff was divided into 5 m long cliff cells, which is sufficient and optimal for determining locations of cliff scarp slumps and base cavities throughout the profile of the cliff. In this way, 291 cliff cells were distinguished. Thus, the cliff erosion clusters, their epicenters and buffers in the horizontal projection of the cliff were determined, distinguishing individual zones of shore activity types.

To identify and localize erosion processes and geostatistical analysis of spatial distribution and change, the zones of prevailing shore erosion forms were identified, determining the continuity coefficients K of cells. After transferring the obtained data to a spectrogram, erosion zones were distinguished (Figure 11). It is possible to identify three different behavior zones of the Olandų Kepurė cliff: southern, central, and northern.

3.5. Cliff Cells and Behavior Units of the Olandų Kepurė Cliff

Cliff dynamics reflect site conditions; each cliff problem is specific [45]. The interrelated geomorphological processes affect the entire backshore, foreshore, and nearshore [102]. Therefore, it is possible to analyze each cliff-lined coastal area as a strip of a linear littoral habitat representing a narrow block from the cliff top to the nearshore of various lengths featuring similar drivers and geomorphological responses or behaviors. Lee and Clark [45] aptly labelled these strips as cliff behavior units (CBUs). Thus, it is possible to group any cliff of various lengths into CBUs as homogeneous elements [103].

Lee ad Clark argue [45] (p. 10) that “the reason to stress ‘behavior’ is that the interrelationships between both process and form are central to explaining the episodic and uncertain nature of the recession process. In this context it is useful to consider cliffs as open sediment transport systems characterized by inputs, throughputs and outputs of material, i.e., they are cascading systems”.

The CBU concept allows the differentiation of the shoreline management approaches depending on the priorities, e.g., conserving the natural state of an undeveloped protected coast like Olandų Kepurė [102]. Therefore, eliciting CBUs is essential for soft cliff surveillance and management, featuring the long-term links between the cliff’s shape and geomorphological processes. CBUs comprise three coherent sub-systems: cliff top, surface, and foreshore [45]. For a CBU, we need to know the nature of geomorphological events, their magnitude and frequency, and the trends, rates and occurrences of large-scale events shaping the cliff, the foreshore and the nearshore [104].

A CBU as a system comprises two essential components: the shoreline sub-system and the cliff sub-system [105]. The impacts on the CBU include a generalized hierarchy of the environmental activation mechanism and primary responses. A typical feature of many CBUs is the sporadic material supply to the foreshore. For beaches, this will equal an irregular delivery of pebbles, gravel, and sand, resulting in changes in beach volume [45]. Considering more processes (beaches, human action, longshore sediment drift), it is possible to describe the CBU behavior better, and the predictions are closer to reality [103].

The Olando Kepurė cliff is classified as a composite cliff system [45], i.e., in our terminology, a complex linear littoral habitat comprising a partly coupled series of homogeneous CBUs. Based on the results of our investigation, three CBUs were distinguished along the 1.5 km length of the Olando Kepurė cliff, each with specific physical features, sediments, cliff dynamics, and the intensiveness of scarp slump and base cavity formation (

Table 2. Cliff behavior units (CBUs) of the Olandų Kepurė cliff.

	CBU	Southern CBU	Central CBU	Northern CBU
Parameter
Average height (m)		16.4	17.7	12.3
Length (m)		527	628	379
Cliff sediments		Sand, sandy clay	Till	Sandy clay
Average recession rate (m/year)		0.93	0.63	0.72
Cliff type		Active cliff	Active cliff	Bluff
Total number of scarp slumps		202	279	181
Total number of base cavities		58	196	94

). The output from one CBU may not always necessarily form an input for the next, although the boundaries between them are not strictly demarcated [48].

The southern and central CBUs of the cliff enjoy limited use of the beach, as the narrow foreshore between the cliff and the sea (<50 m) is a bottleneck for visitor flows, especially when the seasonal sea level is high, usually in spring. The cliff is recessing actively in the southern part. The northern CBU is stabilizing and acquiring vegetation and soil cover. In Lee and Clark’s terms, it is a bluff [45], p. 5: “Steep vegetated coastal slopes, termed bluffs, are generally former cliffs degraded by weathering; most are stable”.

4. Discussion

A range of interrelated drivers may cause the detachment of loose material, resulting in soft cliff erosion [95,96]. The most active drivers are:

• Wave action, including hydraulic action and abrasion, and fluid shearing by up-rushing waves during large storms;
• Seepage erosion;
• Surface erosion, i.e., run-off and wind erosion;
• Gravitational mass movement (creep).

In the case of the Olandų Kepurė cliff erosion in the calm season, the most pivotal drivers are wave action, but especially seepage erosion. Lee and Clark [45], p. 33, define seepage erosion as a cliff denudation process involving the detachment and removal of cliff surface particles by the seepage drag of groundwater flowing out of an exposed soil face with the detachment of sand or till particles from degrading landslide blocks and scarp slumps being the primary process. It reduces the resistance of till, resulting in a scarp slump [94]. Water brings eroded particles down the cliff and deposits on the foreshore [45].

Seepage erosion can shape small-scale slumps in loose sediments [104], like in the case of the Olandų Kepurė cliff. However, seepage is an insufficient erosion driver in lithified materials. Hence, seepage erosion is a prevailing erosive process of soft cliffs, which in a moderate climate zone is particularly prominent in spring after the snow melts, the sediment thaws and the saturation of the groundwater layer induces active infiltration into sediments [106,107,108]. Yet, monitoring and measuring seepage erosion processes on soft cliffs using remote sensing methods is more complex than monitoring wave action.

The essential findings of the results of the presented study led to an apparent notion that it was an efficient approach to consider the coasts as 1-D linear objects in the quantitative analysis and monitoring of coastal erosion processes. For instance, it is valuable in modelling the boundaries of littoral (longshore sediment drift) cells as coastal structures [109]. Due to longshore sediment transport, the 1-D coastline model can accommodate essential processes on the coast, such as perpendicular transport to the beach, changes due to sea level rise, and long-term trends in shoreline changes [110].

It is possible to study coastal behavior of a soft cliff as a linear littoral habitat if sufficient precision of remote sensing using UAVs is achieved and a linear (1-D) interpretation using ArcGIS is applied. Erosion processes of coastal cliffs, especially those with slow retreat rates, are notoriously difficult to monitor by remote sensing methods [111,112,113]. Hence, the advantage of interpreting the eroded coast as a 1-D linear littoral habitat surveyed and monitored by applying the combined approach of the UAV-based field surveillance with data storage and analysis in the ArcGIS-based 1-D database.

The novelty of the results lies in a comprehensive conceptual interpretation and analysis of the eroded coast as a linear littoral habitat. Despite its narrowness, or maybe because of it, this complex linear littoral habitat is a dynamic boundary between the aquatic areal habitats and the terrestrial ones. It is, by convention, the structural element which, in remote sensing interpretation, is referred to as the “instantaneous shoreline” [114].

From the perspective of previous studies, the central practical aspect of the presented study as an illustration of the littoral habitat concept is the deliberation of coastal squeeze as a coastal conservation and management challenge. The development of coastal green infrastructure as a cliff management strategy may deal with a coastal squeeze in eight steps [115]: adequate diagnosis; conservation of mass and energy fluxes; mimicry of natural functions; use of local resources; local actor participation; design and adaptation of solution; resistance and resilience of the intervention; adaptability of the solution. Monitoring is essential in all eight steps.

Also, active communication among scientists, decision-makers and local stakeholders is necessary. Despite Olandų Kepurė being a managed nature reserve, there is one site on the northern part of the cliff top threatened by a coastal squeeze where relocation is impossible as a management strategy (Figure 12). The project developers should involve local stakeholders. They should contribute to the design, implementation, and monitoring processes. In this way, the scientific community will propagate the coastal squeeze mitigation strategies to balance natural processes with human interests and needs [12,116].

Coastal cliff management needs an awareness of geomorphology and changes resulting from coastal processes [117]. It typically aims to break the cliff recession or reduce it. However, what if the cliff is in a managed nature reserve protected as a natural monument? In the UK, cliffs are the most significant landscape assets protected within National Parks and Areas of Outstanding Natural Beauty or as heritage coasts [45]. Cliffs are biodiversity reservoirs with many plant species and communities of coastal cliffs protected by Council Directive 92/43/EEC, and their monitoring is obligatory in EU countries [118].

The essential limitation of the work is the accuracy of the UAV measurements of the cliff surface at a non-vertical angle. Another limitation of the study is in possible human error in the identification process of the erosion sites—cliff scarp slumps and cliff base cavities. We made efforts to eliminate any human error, which the Delphi method enables in such studies. In our opinion, the varying spatial distribution of erosion sites during the investigated storms, which occurred in calm periods, has natural causes and it is not caused by a human error.

It should be an important, albeit not pivotal, method in a broader, comprehensive littoral monitoring system combining methods to survey coastal slope, orientation, geology, substrate, wave exposure, water temperature, human impact, and their impact on the distribution of littoral habitats [119]. Combining the usage of UAV with GIS analysis tools enables researchers to recognize and map the cliff plant species, derive and measure the habitat and species distribution area, count surveyed individuals, and gather quantitative data on their projected area [118].

Future research directions in combining the application of UAVs with GIS tools for coastal cliff surveys will focus on technological improvements in UAVs to help tackle the image quality problem. AI algorithms will become more efficient in recognizing cliff scarp slumps, foot cavities and other ephemeral forms resulting from coastal erosion processes. Machine learning techniques are popular for classifying regularly acquired remotely sensed imagery [120]. The latest promising techniques include regression-based applications of deep learning models for multi-scale image detection on UAVs, which are more effective than the models with a small number of trainable parameters [121].

5. Conclusions

We tested using a UAV as a survey means, gathering remote sensing images of the Olandų Kepurė cliff (eastern Baltic Sea coast, Lithuania). Five photo interpreters independently analyzed the acquired images using the Delphi technique. Using ArcGIS, the distribution of cliff scarp slumps and cliff base cavities as the main ephemeral geomorphological features resulting from coastal erosion processes was mapped. The main conclusion is that the survey results were not affected by photo interpreters and are of sufficient discerning quality for decision making in cliff conservation and management.

By increasing the density of surveyed cliff cells in the analysis of the distribution of slumps and cavities, the results become more uniform, and by reducing the density of surveyed cliff cells, the results become more fragmented. After performing initial calculations using cliff cells of different lengths, the conclusion is that a 55 m long cliff cell is the most suitable to analyze cliff behavior. This insight led to the further conclusion that there is no direct relationship between wave action and the cavity and slump formation. The behavior of the Olandų Kepurė cliff is complex and, during the calm season, depends primarily on seepage erosion.

Even though the Olandų Kepurė cliff is a unique coastal strip for the Lithuanian Baltic Sea coast, it is not included in the National Coastal Belt Management Program for 2023–2032 approved by the Lithuanian Government. Therefore, no specific coastal management measures are designed for the cliff. Since it is Lithuania’s most visited natural object, such a situation is paradoxical. Nevertheless, various soft cliff management measures related to visitor flow management are implemented yearly, such as installing wooden paths and stairs down to the beach. The key challenge is understanding the cliff behavior in different seasons and weather conditions.