An Assessment of Landscape Perception Using a Normalised Naturalness Index in the Greater Seoul Area

Abstract

This study analysed the greater Seoul area (GSA) in terms of naturalness, a representative indicator of natural scenic beauty, and created an assessment map, shifting from a traditional urban development perspective to a landscape perspective. It also developed a “normalised naturalness index” by combining the results of the expert metric score with the Hemeroby index, which was used as a naturalness assessment representative item. Then, it interpreted the naturalness status of the GSA landscape characteristics. As a result, the landscape of the GSA demonstrates the following five characteristics: First, the central business districts in the capital city of Seoul are densely developed areas with a very high degree of human intervention. Second, the satellite cities built to solve Seoul’s housing and logistics problems are rated as “a little less, but still heavily humanised” as a landscape characteristic. These areas are becoming increasingly humanised. Also, it is worth noting that the third characteristic, regarding moderate landscape areas, has a distinctly different meaning for areas outside of the city boundary, as well as those within the city boundary. Although these areas are in the same statistical category, they have two different meanings: one is the area where the average values converged on “moderate” by virtue of urban forests near the city centre, and the other is the area outside of Seoul that has a Hemeroby value of 0.5–0.6, which refers to open spaces such as agricultural lands, wetlands, or coastal areas. Fourth, suburban forests are reserved with legal restrictions to curb excessive urban sprawl, as well as parts of the demilitarised zone along the border areas of North and South Koreas. The last landscape characteristic is illustrated in the scenic area of the eastern woodlands. The normalised landscape naturalness index developed through this study provides an overall understanding of the environmental state of the GSA. Future research may build on the results of this study to refine methods for assessing public perceptions of naturalness.

1. Introduction

1.1. Landscape and Perception

The rapid development of the greater Seoul area (GSA) since the 1960s has transformed it into a busy metropolis, resulting in expansive suburban cities. This has caused the urban population and development to expend to the suburbs, leading to significant environmental problems, with the original landscape of the GSA gradually disappearing. The United Nations was the first to put forward the concept of “sustainable development” in 1987, defining it as “meeting the needs of the present without compromising the ability of future generations to meet their own needs” (https://www.un.org/sustainabledevelopment/ accessed on 18 May 2024). In terms of promoting sustainability, the current rapid environmental change of the earth means declines in biodiversity and the difficulties of carbon neutrality (Lu, Y et al. [1]). Therefore, this study analysed the GSA in terms of naturalness, a representative indicator of natural scenic beauty, and created an assessment map to highlight its current status. The aim of this study is to understand the currently perceived natural state of the GSA landscape in terms of naturalness. In doing so, we created a normalised naturalness index using Hemeroby and expert assessment data.

“Landscape” means an area, as perceived by people, whose character is the result of the action and interaction of natural and/or human factors (Council of Europe Landscape Convention [2]). In other words, human–environment interactions correspond to the concept of landscape, and the ELC defines landscape as “the area that people perceive”, placing the public at the centre of any understanding of the landscape (CoE [3]). As Butler puts it, “Being recognised as an entity perceived by people moves the focus of landscape from being a purely physical area to being dependent on individuals and society” to provide it with meaning (Butler and Berglund [4]). This is related to the cognitive concept of perception also in the natural scientific context. In the majority of studies by Korean experts, landscape is defined as a socio-ecological system where systematic interactions occur among diverse ecosystems and human society (Jun et al. [5]; Kim and Son [6]).

In social sciences, the concept of perception relates more with the intellectual/cognitive sense of perception, which has largely been the domain of environmental psychology (Scott, [7]; Groening, [8]; Butler and Berglund [4] Kim and Son [6]). Scott [7], in a study focusing on the Denbighshire region of the United Kingdom suggests the need to integrate land use with user perceptions of the landscape, argued that “public perception is complex and idiosyncratic, rendering that simple analysis or generalisation hardistic”. However, he also said that, if used well, public perceptions can reveal implications that top-down strategies have failed to address.

Butler and Berglund [4] argue that landscape characterisation is a process of “understanding the differences between distinct landscape regions and types and that characterisation of landscape is needed to identify what makes each place special not in terms of rating it better or worse”, but rather what makes it unique. This means that cognitive evaluation is needed to uncover the localities that distinguish one area from another in human perception.

Furthermore, in his book, Jones [9] argues that public participation in landscape assessment enables decision making to incorporate public knowledge, values, perspectives, and behaviours, facilitating the seeking of solutions by providing administrators with a better understanding of the problems perceived by the public.

Kim and Son [6] used a Q-methodology to examine the public’s “perceived naturalness” of urban parks in Seoul, South Korea, and found that different groups of the public had different perceptions of naturalness, suggesting that there is a fragmented public perception, as argued by Scott earlier. However, Kim and Son [6] did find a dominant group of perceptions of landscape naturalness, suggesting that there is a representative public perception for assessing naturalness. These cognitive studies suggest that landscape assessments should consider the subjectivity of users or the level of public awareness of the environment.

A notable example of landscape-based regional planning is the United Kingdom’s (UK’s) Landscape Characterisation Assessment (LCA) under the European Landscape Convention (ELC) (Tudor [10]). The landscape characterisation method in the UK is used to describe the features of a local landscape that distinguish it from the surrounding areas (James and Gittins [11]). Experts and local activists combine qualitative and quantitative studies to evaluate landscapes at the national, regional, or local government levels. The LCA step presents an objective expert-level assessment using spatial data, such as GIS, as the basis for public assessment. Experts play a key role in assigning identifiable landscape attributes and selecting specific proxy indicators in order to efficiently distinguish landscape features (Rogge et al. [12]; Vizzari [13]; Martín et al. [14]). The results of landscape characterisation distinguish between “landscape areas” and “landscape types” and are made available to the public for decision making in local communities. These results may be used as the basis for regional planning.

The use of indicators from landscape ecology has also become common (Gobster et al. [15]; Kim and Pauleit, [16]; Fry et al. [17]; Martín et al. [18]). This calls for an eclectic effort to integrate data-driven approaches with those of social sciences to simplify and present them in a single intuitive score. In this regard, the Ministry of Environment in South Korea has produced the Ecological Naturalness Map [19] and the Environmental Conservation Value Assessment Map (Song, W. et al. [20,21]), which are available by region. In addition, the normalised difference vegetation index (NDVI), the enhanced vegetation index (EVI), land cover, etc., are used to identify the degree of physical vegetation integrity (Cho et al. [22]; Kang et al. [23]). However, these efforts are mainly focused on assessing physical or ecological naturalness, with the main purpose of marking areas of special concern for development sensitivity. Therefore, there are limited studies that consider the perceptual aspects of the landscape characterisation process.

1.2. Naturalness of Landscape Character

Naturalness as a concept generally describes how close the current landscape is to a state of nature as perceived by humans (Tveit et al. [24]; Kim and Son [6]). Perceived naturalness may therefore be assessed differently from ecological naturalness.

In Tveit’s study, the dimensions of naturalness are defined as intactness; wilderness or pristine wildness; naturalness; and ecological robustness. The following attributes were selected for assessment: natural features; the structural integrity of vegetation; vegetation/ground cover type; water; management; patch morphology; and edge morphology. Canopy cover, area, and spatial cover were selected as potentially measurable indicators. “Vegetation richness” refers to the amount or density of vegetation—both natural and synthetic—growing in each space. Specific studies using canopy as an indicator of naturalness include one that separated vegetation density into two categories—high and low (Purcell and Lamb, [25]); one that used quantitative values (von Oheimb et al. [26]); and one that surveyed the amount of vegetation around residential areas (Hur et al. [27]).

Ode et al. [28] concluded that vegetation naturalness or transition levels using orthophotos or land cover data may be estimated using the weighted reclassification of various indices. In a landscape beauty assessment study covering Germany, Hermes [29] used the Hemeroby scale as an indicator of naturalness. Kerebel et al. [30] analysed naturalness using the CORINE Land Cover (CLC) classes (European Environment Agency [31]).

Naturalness was assessed along several dimensions. As shown in Figure 1, “Nature as a wilderness” refers to “nature in its pristine state” and encompasses the natural environment in its sublime and transient ephemerality. It refers to a state that is conserved; self-sustaining; managed for its own sake (Landres et al. [32]); primarily assessed as an ecological indicator; and representative of biodiversity.

When discussing naturalness, “no anthropogenic intervention” refers to a state in which processes of creation, destruction, and change occur by nature itself in the absence of the effects of anthropogenic development. Specifically, there are studies describing changes in edges or changes in ecological cover as a shape index from the perspective of landscape ecology based on the land cover status (Ode et al. [33]), studies that present fractal dimensions as naturalness (Hagerhall et al. [34]), and studies comparing the area of urbanised and agricultural land with the area in a natural state (Wrbka et al. [35]).

Landscape naturalness not only refers to ecosystems, but also to the “degree to which a physical object is visually natural” as a visually observed form or material. When a landscape is said to “contrast with the surrounding environment”, this suggests that naturalness may be more pronounced when contrasted with synthetic elements outside of the boundary. Specifically, a study (Van den Berg et al. [36]) distinguished the restorative effect of green space naturalness, depending on whether the surrounding environment is artificial or natural.

The concept of naturalness is, therefore, grounded in environmental psychology theories such as biophilia, and is often associated with indicators of human preference for landscapes in their natural state, including the greenness of vegetation cover (Real et al. [37]; Wu and Hobbs. [38]; Ode, Å, [33,39]; Walz, [40]). Indeed, psychophysical research shows that nature influences mental resilience and psychological satisfaction, thus emphasising the importance of nature experiences (Kaplan, [41]; Kellert and Wilson [42]).

Among them, this study derived a normalised indicator of “naturalness” that integrates objective indicators from ecological research concepts and public evaluative perceptions. Naturalness is of great importance as an alternative perception indicator for preventing human-induced environmental degradation and development conflicts. Naturalness here refers to the degree of naturalness as perceived by humans.

However, there is a general lack of exploratory research to classify regional characteristics in Korea. Naturalness remains a novel concept in Korean landscape science. The lack of consideration fails to provide not only a basis for countering the logic of development, but also a vision of how peripheral cities with sprawling development should move forward to preserve their local characteristics in the future. This study, therefore, analysed the GSA in terms of naturalness, a representative indicator of natural scenic beauty, and created an assessment map to address its current status.

2. Material and Methods

2.1. Study Area

The GSA consists of the Seoul, Incheon, and Gyeonggi-do Provinces, comprising thirty-three cities in total. These characteristics result in diverse land uses in the GSA, from urban to rural areas, with rapid changes in landscapes occurring in the suburbs.

The GSA covers an area of over 11,868 ${\mathrm{km}}^2$, accounting for 11.8% of the entire area of Korea. Its population is approximately 26,000,000 (Son et al. [43]), representing 50% of the nation’s population. The administrative complex comprises 33 districts, 28 cities, and 5 counties, including Seoul (25 districts), Incheon (8 districts, two2 counties), and Gyeonggi-do (28 cities, 3 counties). As presented in Figure 2, there is a restricted development zone between Seoul and Gyeonggi-do, which is part of an urban planning measure to stop Seoul’s sprawl.

2.2. Method

The study used the Hemeroby index and figures from Kwon, H. et al. [44] to derive normalised scores through min-max nominalisation. The term Hemeroby, introduced by Jalas [45], is derived from the Greek words hemeros (cultivation) and bios (life). This refers to an integrated metric used to determine the ecological components and degrees of human intervention in ecosystems. It is often used to draw landscape-ecological maps using land cover (Kowarik [46]; Steinhardt et al. [47]; Walz and Stein [48]; Kim and Son [49]; Erdős et al. [50]).

It particularly indicates the degree of human interference with the natural environment, and the grade of human impact is typically assessed using a scale of seven levels, with the lowest values (Ahemerobic) corresponding to “natural” or “undisturbed” landscapes such as wetlands, and the highest values (Metahemerobic) to “fully disturbed” or “man-made” landscapes, such as urban areas (Tian et al. [51]; Wu et al. [52]).

This study did not use the existing seven scales but converted scores into five scales to suit the land cover of the GSA. The land cover classification of the Ministry of Environ-ment, South Korea, using 2021 data [19] was used as the base map. Kwon, H. et al. [44] was referred to in order to determine the degree of human perception of naturalness for each Hemeroby class according to the ecosystem type as the expert survey value.

The expert survey was conducted by the National Institute of Ecology and was calculated using the Delphi method. According to the report, the expert survey was an online survey. The Delphi method is a flexible approach to gathering expert views on an area of interest. It relies on the core assumption that predictions made by a group are more accurate than those made by an individual; the Delphi technique constructs predictions from expert consensus through structured iterations (Barrett and Heale. [53]). In this study, land cover scores based on the Hemeroby index were compared to ecological function matrix values from experts. We then calculated the average value per unit of land cover.

As shown as Figure 3, we firstly assigned scores to 25 land cover-based ecosystem types in Korea by comparing them with corine land cover types in Europe and previous studies (Tian et al. [51]; Kim and Son. [49]). Secondly, we normalised the scores assigned to each Hemeroby index and the expert land cover assignment scores to take on a value between 0–1.

Specifically, the classification of urbanised land cover was given one point. As shown in

Table 1. Land cover score weighted by Hemeroby and the ecological function matrix score.

Hemeroby	Degree of Naturalness	Human Impact	Korea Landcover Type (Referred from CLC)	Hemeroby Score	Ecological Function Matrix Score
Ahemerobic	natural/Almost no human impacts	none	Bare rock	5 points	8 points Bare rock
Oligohemerobic	close to natural/weak human impacts	limited removal of wood, pastoralism, depiction through air and water	broadleaf forest, intertidal flats, mixed forest (potential natural vegetation), coastal lagoons, beaches, dunes, sands, estuaries, inland marshes, sea and ocean, peat bogs, saltern		10 points broadleaf forest, coniferous forest, mixed forest, sea, and ocean
					9 points beach, water body, river, lakes, and marshes
Mesohemerobic	semi-natural/moderate human impacts	clearing and occasional ploughing, clear cut, occasional slight fertilisation	coniferous forest, transitional woodland shrub, mixed forest (not potential natural vegetation), sparsely vegetated areas, natural grasslands, other grasslands, burnt areas, river, lakes and marshes, beach	4 points	8 points natural grasslands, inland marshes, foreshore, bare rock
					5, 6 points irrigated arable land, ranch farm, extra bare land, pastures, golf field, other grasslands
$\alpha$-euhemerobic	relatively for from natural/moderate strong human impacts	application of fertilisers lime and pesticides, ditch drainage	irrigated arable land, farmland, ranch farm, complex cultivation patterns, cemetery, extra bare land, pastures, land principally occupied by agriculture with significant areas of natural vegetation	3 points	3, 4 points farmland, unploughed farmland, complex cultivation patterns, cemetery
$\beta$-euhemerobic	far from natural/strong human impacts	deep ploughing, drainage, application of pesticides and intensive fertilisation	non-irrigated, arable land, vineyards, complex cultivation patterns, orchard	2 points	all most 3 points, orchard only 6
			facility plantation, greenhouse		2 points
Polyhemerobic	strange to natural/very strong human impacts	single destruction of the biocenosis and covering of the biotope with external material at the same time	sport and leisure facilities, discontinuous urban fabric, construction sites, mineral extraction sites, dump sites	1 point	1 point
Metahemerobic	artificial/excessively strong human impacts	biocenosis destroyed	continuous urban fabric, port areas, airports, industrial or commercial units, road and rail networks, housing		0 points

, non-irrigated arable land and complex cultivation patterns were assigned a score of 4, excluding orchards, which had a higher perceived score of 6. Ranches, aquaculture farms, and pastures were assigned six points each. Non-irrigated rice paddies/fields were given 3 points. However, water bodies such as ponds, rivers, and lakes were considered to have a similar degree of naturalness within the city; they were, therefore, given 8 points. In addition, the bare-rock space received eight points. Green forest spaces, such as coniferous forests, broadleaf forests, and mixed forests, were rated 10 points.

Min-max normalisation can make comparisons easier through evaluating each data point on the same scale (0, 1) by performing a linear transformation on the original data. This land cover-based assessment is advantageous as the results can be derived quantitatively via reflecting qualitative values when assessing perceived naturalness on a wide scale.

$$x_{scaled}=\frac{x-x_{min}}{x_{max}-x_{min}}$$

Thirdly, the average score of the two normalised scores was assigned to a thirty-meter grid scale of registers and evaluated using the Hemeroby equation.

$$NHI={\displaystyle \sum }_{i=1}^h\frac{S_i}{S_A}\times h$$

NHI = normalised Hemeroby index;

h = number of degrees of normalised Hemeroby (here: n = 10);

S_A = total area of grid unit;

S_i = area of cover types with interference level i.

Based on Tian et al. [51], Wu et al. [52], and Kim and Son [49]’s studies, the grid-specific assigned scores for perceived naturalness are combined with the actual situation to provide a score to the landscape types in the study area (Table 1).

Finally, we used a Zonal statistics algorithm to organise the characteristics of each region. The zonal statistics operation calculates statistics on the cell values of a raster within an area defined by a specific vector dataset, and, in this study, we utilised QGIS [54]. The Zonal statistic algorithm is a statistical method that calculates the mean by summing all cell values and then dividing by the number of cells in the zone.

$$\overline{x}=\frac{1}{N} {\displaystyle \sum }_{i=1}^N{x}_i$$

$\overline{x}$ = mean

x_i = observed values

N = number of observations

3. Results

3.1. Integration of the Naturalness Score

The two different scores were normalised, averaged, and assigned to the grid. As shown in Figure 4, several land cover scores differed between the perceived naturalness Hemeroby scores and the expert scores (see Appendix A).

There was a difference observed in the perceptions of irrigated and non-irrigated arable land. In Figure 4, the expert scores are higher than the Hemeroby scores, indicating that this land type was more strongly perceived as being natural, relative to the degree of heavy human intervention.

In contrast, flat landscapes, cemeteries, other grasslands, inland marshes, foreshore, and saltern were found to have higher Hemeroby scores than expert scores. Characteristically, in the perception of landscape naturalness in Korea, orchards were rated highly by experts. However, they were not rated highly on the Hemeroby scale based on actual land cover—this reflects the reality that most orchards are covered by greenhouses.

There was also a slight discrepancy between the expert and Hemeroby scores for forest landscapes, where experts consistently rated forests with many trees as having the highest naturalness scores. When assessed based on the degree of human disturbance and land cover, however, mixed forests and most coniferous forests are rated as having less naturalness than broadleaf forests.

The scores for mining areas also show significant differences. While the mining area received an expert score of zero, it received a high Hemeroby score of 0.75. On the other hand, there were some land types with common scores. These include the majority of areas (111–163) that correspond to urban planning facilities. Land cover representing these urban facilities all scored zero. In addition, broadleaf forests scored one across all dimensions, and seas and oceans were also rated as highly natural.

Look at the results assigned to the raster in Figure 5. The scores averaged over the two dimensions represent the landscape character of the GSA. First, Seoul is rated as having low naturalness; most of the land cover is urbanised. These areas received a score of zero. They represent urban sprawl—mainly transport networks, residential communities, and commercial/business areas. On the other hand, Icheon City, which is a rural area in GSA, shows a lot of cultivated land.

Except for a few urban centres, most areas in this region scored between 0.38 and 0.5. The land covers that fall between 0.38 and 0.5 are “Irrigated arable land”, “Ranch farm”, “non-irrigated arable land”, “Orchard”, etc.

Figure 6 presents the naturalness assessment map of the GSA. The map demonstrates the average values of the two dimensions for each region. The colour differences of the 10 levels are based on Jenks’ natural break optimisation. The first level, which is red (0.159–0.270), indicates Seoul, the location of the central business district (see Appendix B). Incheon-Jung-gu is also coloured red because it is a central business district. The southern part of the GSA, coloured orange, is the logistics and historical housing centre for cities such as Incheon and Suwon.

The upper right side of the GSA is dominated by dark green (0.700–0.850) mountainous terrain that is mostly forested. South Korea features high mountain ranges to the right, and 70% of the country is forested. Due to these features, areas such as Yangpyeong and Gapyeong counties appear to be highly natural on the map.

In contrast, the southern part of the GSA is a plain area with several factories and production facilities. It was, therefore, rated as moderately natural (0.410–0.560) on the naturalness map. Typical areas with these characteristics include Incheon, Pyeongtaek, and Hwaseong.

3.2. Landscape Characteristic Difference of Naturalness in the GSA

The landscape characteristics of the GSA, according to its natural features, are shown in Figure 7, Figure 8, Figure 9, Figure 10 and Figure 11. Figure 7 is a predominantly urbanised area with the central business district located in Seoul. The mean value of all scores within the vector polygon is 0.159–0.270. This area has served as the original city centre of Incheon and the capital of Seoul since before the Joseon Dynasty. Songpa-gu, known for its high floor area ratio, is also significantly urbanised.

Areas less heavily urbanised which still have high levels of human intervention can be seen in Figure 8. This area had something in common with a satellite city built to solve Seoul’s housing and logistics problems.

Some areas represent densely populated residential zones that were developed as a national effort to resolve Korea’s housing problems in the 1980s. The others represent the industrial and logistics areas due to factory relocations and facility expansions since 2000. These areas have dense apartment urban forests. Because they were created via systematic urban planning, they also have city parks. Typical examples include Jungnang-gu and Hanam-si in Seoul, Yeonsu-gu in Incheon, and Goyang-si, a relatively recent residential development.

Also, it is worth noting that the third characteristic, which is in regard to moderate landscape areas, has a distinctly different meaning for areas outside of the city boundary as well as those within the city boundary (Figure 9). These regions fall into the same statistical category, but have two different meanings.

First, areas within Seoul are green urban areas with forests or rivers nearby. These have a very high standard deviation of land cover values near urban forests, shown in 3-1 in Figure 9. Examples include Nowon-gu and Dobong-gu in Seoul.

Second, the GSA areas outside of Seoul converge around a mean value between 0.5 and 0.6, representing open spaces such as agricultural areas or farmland, or moderate landscapes with a perceived naturalness of around 0.5 due to wetlands or coastal areas. These areas have a very consistent landscape and are characterised by a predominantly flat landscape. However, some of these rural areas are also undergoing urbanisation due to residential development.

Third, suburban forests are reserved with legal restrictions to prevent urban sprawl and include parts of the demilitarised zone in the border areas of North and South Korea (Figure 10). However, some of these areas are greenbelts under threat of development and are in natural decline. As the development restriction zones surrounding the outer periphery of Seoul have become increasingly encroached upon, the degree of naturalness has not been as high.

Paju and Yeoncheon are also noteworthy as they are located in the Korean Demilitarised zone, an area of high ecological value that is richly preserved due to the military demarcation line, making it difficult for civilians to enter. Paju has recently been built as a new city, which has reduced its naturalness in terms of the total area; it is, however, still considered a highly natural area (0.560–0.700).

Finally, the easternmost outskirts of the GSA are characterised by the following features (Figure 11): high physical and perceived naturalness, rich in forests, lakes and wetlands, and high-tourism value. Most of the terrain is forested, and it is a vast scenic resource due to its valleys and lakes.

4. Discussion

The results of the environmental characterisation of the GSA’s unique landscapes are intriguing; the physical and perceptual assessments of naturalness are in agreement in several places. First, the normalised naturalness index developed for the landscape assessment of the GSA is an effective representation of the characteristics of each area. Cities such as Suwon and Seongnam, which have long been developed as satellite residential cities, scored low in both physical and perceived naturalness. In the past, the original environment before development had disappeared, with the heavy advancement of urbanisation, and subjective assessments also acknowledged these areas as urbanised.

In contrast, land cover that had not been developed because of political and social circumstances, such as demilitarised zones and restricted development zones, had higher naturalness scores and similar perceived scores. Previous studies have demonstrated that more vertical landscape elements, such as forests and tree canopies, indicate more natural attributes. These results support the findings of Walz and Stein [48] and McMahan et al. [55], indicating that higher ecological quality is associated with higher naturalness values.

When the study evaluated the mean value between the perceived and actual naturalness scores, an unexpected finding occurred: neighbourhoods with more irrigated cropland, plains, grassland, and open space are more likely to have higher naturalness scores. While these regions are statistically in the same category, they have two different implications.

While they may have the same mean score, different characteristics within a neighbourhood indicate different aspects of their normalised Hemeroby. This demonstrates either that values at the extremes have converged to the mean, or that values at the extremes are moderate. The latter happened in two cases. First, areas within Seoul are green urban regions with forests or rivers nearby, which means they have a very high standard deviation of land cover values (like urban spaces). Second, areas outside of Seoul converge around a mean value of 0.5–0.6, indicating a moderate landscape with a naturalness of around 0.5, due to farmland or agricultural fields; wetlands or coastal areas; and other open spaces.

Thus, the results support existing research findings that perceived naturalness at the macro level—which equals the sum of internal and external environmental conditions—is perceived through contrast and association with the surroundings (Liding et al. [56]; Ferrari et al. [57]; Van den Berg et al. [36]; Kim and Son. [49]).

5. Conclusions

As urbanisation progresses, the environment that is harmful to human mental and physical health increases. As a result, due to various climate changes and pollution hazards in the social environment, modern society’s demand for good natural landscapes and environmental conservation grows.

Naturalness aims to grasp the human perception of nature based on the physical ecology in nature, and is recognised as important in terms of identifying a consensus on the natural environment, ultimately forming the motivation for environmental stewardship. The importance of the fundamental link between nature and human health is increasingly recognised in global and regional policy development.

However, previous assessments of perceived naturalness have been conducted, mainly through psychometric surveys using photographic data (Carrus et al. [58]; Marselle et al. [59]), and there have been a few attempts to transfer perceptions of naturalness through qualitative methods to mapping data. As a result of these limitations, perceived naturalness has mostly been assessed within a limited space with the opinions of a small number of subjects, requiring the application of a broader spatial scale.

In this regard, this study, therefore, analysed the GSA in terms of naturalness, a representative indicator of natural scenic beauty, and created an assessment map to address its current status. To assess naturalness, we developed an index that integrates physical and cognitive landscape naturalness. As mentioned in the UK’s LCA case and the literature, decision making in environmental science requires the consideration of the human perception of landscapes. Intellectual/cognitive perception, a domain of environmental psychology in the social sciences, was therefore incorporated into the assessment of local landscapes.

Fortunately, the normalised naturalness index that was developed with two dimensions of indicators accurately reflects the environmental characteristics of the GSA. The results provide an overall understanding of the environmental status. The result of this study can, therefore, be used to assess the naturalness of large areas efficiently.

Of course, the limitation of this study is that the normalised naturalness index does not reflect the subjective naturalness evaluations of public members. However, this study devised an alternative method to evaluate perceived naturalness using expert evaluation values for publicly available land covers. The actual public perception of naturalness in the future should be devised based on the results of this study.

Globally, landscape research is moving towards an integrated perspective to uncover local character and to promote landscape as a regional attraction. The perspective of looking at megacities in terms of sustainable environmental planning should also be explored in the direction of typifying landscape characteristics.