The Challenge of Social Vulnerability Assessment in the Context of Land Use Changes for Sustainable Urban Planning—Case Studies: Developing Cities in Romania

Abstract

Urban growth triggers massive changes in land use cover, exacerbating extreme natural and technological events. In order for land use planning to be efficient, it requires the integration of comprehensive risk and vulnerability assessment. This paper aims to create a bridge between the existing vulnerability theories and their implementation in land use planning policies and proposes an innovative approach to determine whether the changes in the territorial dynamics of cities draw considerable changes in communities’ social vulnerability. The methodology identifies and selects three case studies from the Urban Atlas inventory, representative of the dynamics of large Romanian cities, taking into consideration the following hazards: earthquakes, floods, and technological hazards. Vulnerability was then assessed by assigning each land use class a specific vulnerability level. The methodology involved assessing the level of vulnerability specific to the situation in 2018 compared to 2006. The results showed that major changes in land use are related to the transition of areas with a low level of vulnerability to areas with a higher level of vulnerability as a result of the urban areas expansion to the detriment of natural and agricultural areas. This is generally translated into a higher degree of vulnerability due to an increased density of artificial elements and of population in the residential areas. The findings of the study of territorial dynamics in the proximity of large industrial operators did not reveal a tendency that differed from the general trend. Although many territorial changes have been observed in the period 2006–2018, it is necessary to extend the analysis, with the issue of the new versions of the Urban Atlas, to confirm the identified trends and to express the up-to-date situation.

1. Introduction

Vulnerability, land use, and sustainable development are issues that receive considerable attention in human society nowadays. The land is a non-renewable resource and while demand is constantly increasing, it is imperative to maintain a balance between demand and supply, needs and interests, or between contradictory uses, through land use policies that achieve sustainable development and improve the quality of the environment [1,2]. Very often, a poorly developed urban planning process leads to the changing of more natural land surfaces into artificial ones planned for human activities, therefore increasing social vulnerability. Therefore, the evaluation of the land use change process is important in order to ensure a sustainable development of urban areas and to increase the resilience of territories and communities [3,4]. On the other hand, land use planning may also positively impact the environment by preserving natural resources, enhancing open space opportunities, or providing a significant reduction in traffic pollution [5].

Land use depends on numerous factors, including population, economic status, infrastructure, industrial activities, geographic conditions, land development policies, etc. [6,7] and impacts numerous parameters, including flood risk, landslide probability, biodiversity, urban climate, hydrological processes, and pollution [8,9,10,11,12,13,14,15].

Given the increasing number of disasters over recent years, one of the most efficient and accessible methods for reducing the pressure posed by natural or technological risks is reducing the vulnerability level of communities exposed to a particular hazard [16,17].

Vulnerability is a term used by various stakeholders, which means it has different meanings relevant to those interests: researchers, disaster management agencies institutions, climate change organizations, or development agencies [18,19,20]. Generally, social vulnerability is an inner property of the human system that describes its capacity to prepare for, absorb and adapt to risks posed by environmental shocks and natural disasters [21,22,23]. Social vulnerability focuses on those features of the society–environment that stem from social inequalities and influence ability to respond [17,24]. Social vulnerability characterizes the community regardless of the potential to produce harm or of exposure to hazard [25].

Recent scientific literature has intensely analyzed the two concepts, “social vulnerability” and “land use” taken separately, but barely discusses their potential connection. However, the number of studies aiming to fill this gap is constantly increasing, but most of them pay attention to vulnerability to a specific hazard. For example, the assessment of flood vulnerability within spatial planning is discussed by [11,26,27,28]. Vulnerability assessment for the environmental sector (species) is examined by Berry et al. 2006 [18], while the impact of land changes on land degradation is studied by Makhamreh 2019 [29]. Ştefănescu et al. 2018 [30] analyzed social vulnerability in the case of technological accidents (explosion and toxic dispersion), while Robertson 2020 [31] addressed the technological exposure of a population of individuals.

Vulnerability levels strongly connect to land use classes. Specific indicators in the literature demonstrate that certain demographic characteristics influencing social vulnerability are correlated to land uses [32].

As such, urbanization and industrialization processes have led to an increased proportion of the population living in cities, where the impact of a technological and natural hazard is certainly more significant [33,34]. This high population and household density translates into residential areas, therefore correlating these land use classes to increased social vulnerability. For vulnerability analysis, this study considers the following hazards: earthquakes, floods, and technological hazards.

Earthquakes, as natural disasters, pose a long-term concern due to their highly destructive power, wide range, and significant effects on society, economy, and environment [35,36,37,38,39]. According to the Emergency Events Database (EM-DAT) from the Centre for Research on the Epidemiology of Disasters (CRED) global statistics report, earthquakes were among the most frequent natural disasters from 1998 to 2017. Data show that more than 700,000 people were killed and more than 1,400,000 people were injured, amounting to 540 billion US dollars in economic losses worldwide. The economic development of the affected regions is usually widely impaired due to the costs related to the recovery and reconstruction. Generally, earthquakes have different impacts on people, depending on the proximity to the earthquake fault, soil type and structure, nature of the building stock, population characteristics [40], location of the epicenter, physical vulnerability, and social vulnerability [41]. It was demonstrated that even in the case of earthquakes of the same magnitude, the death tolls varied greatly: death toll and mortality showed a decreasing trend with increasing economic levels [39]. This can be explained by the different implemented risk reduction measures or building standards.

Floods are among the most common and dangerous natural hazards, causing considerable economic and human losses [42]. The impacts are both direct (property and infrastructure damage, severe damages to people and environment, reduced crop yields) and indirect (temporary relocation, disruption of public service utilities) [43,44,45,46,47]. Since most cities are distributed in river floodplains or along the coast, these urban areas are more exposed to flood disasters [48]. Furthermore, the characteristics of urban areas increase flood risk because of impermeable surfaces, which cause higher peak and volume runoff after rainfall [42]. Many studies in the current literature have demonstrated a strong connection between floods and multiple socio-economic characteristics of the affected population [17,43,48,49].

The rapid economic growth and industrialization process have led to increased exposure and risk to technological hazards, resulting in human victims, economic losses, and environmental degradation [50]. Accidents cause consequences to workers and communities living in the vicinity of industrial plants. These are usually those social groups with lower income, and therefore higher vulnerability. Social vulnerability is also enhanced by population structure (proportion of females, young children, and elderly) and access to emergency resources (for example, evacuation routes, nearby hospitals) [51]. Therefore, a distinction should be made between two main phenomena: the fire/explosion and the toxic dispersion, with the extent of the affected area being the main difference between these events.

Urban and industrialization growth trigger massive changes in land use, exacerbating extreme natural and technological events. This fact is supported by many studies in the literature [28,52,53]. According to Su et al. (2021) [54], land use change may increase the risk disaster, especially when the land use planning does not include a disaster risk assessment approach. The method proposed in this study, which is based on data envelopment analysis (DEA), represents a tool to assess the relationship between land use change and disaster losses associated with flooding.

Although numerous studies demonstrate the link between specific events and vulnerability, the association of land use classes and vulnerability remains underexplored. To address this gap, the main objective of this paper is to explore the linkages between vulnerability to natural and technological hazards and land use cover and to propose a methodology for assigning a vulnerability level to each land use class. Furthermore, this study includes vulnerability analysis regarding four different hazards, offering a better and more complex understanding of the impact of land use change in the selected areas of study.

Our study aimed to answer the following research questions:

RQ1: What is the correlation between land use classes and vulnerability level?

RQ2: Which were the main land use changes within the time period under consideration?

RQ3: Are technological risks a real concern in the territorial dynamics of developing cities?

RQ4: How does the level of vulnerability vary as a result of the territorial changes that have taken place in Romania’s major cities?

2. Materials and Methods

This study includes a series of methodological steps to determine whether the changes imposed by the territorial dynamics of cities in developing countries, such as Romania, imply a change in the intrinsic degree of vulnerability. The purpose was to determine whether the occurred land use changes contributed positively or negatively to the general level of vulnerability and, implicitly, to the level of risk.

As the present study focuses on the link between land use and the general level of vulnerability, the first methodological stage involves identifying and selecting three case studies from the Urban Atlas inventory. These case studies are representative of the dynamics of large Romanian cities.

The Urban Atlas inventory was used in the analysis because it is a reliable, high-resolution source of land use data, made available to users as an open-source online resource. For the development of the Urban Atlas database, Earth Observation (EO) data, such as SPOT 5, ALOS, RapidEye, Sentinel, and QuickBird, were used, and the inventory was supported by the European Space Agency (ESA) and the European Environment Agency (EEA) [55,56].

Since 2000, the European Commission tasked the European Space Agency with the systematic acquisition of satellite images at regular intervals. The 2006 Urban Atlas products are based on previously mentioned data (high resolution mosaics) with 2.5 m spatial resolution multispectral, including infrared or multispectral merged with panchromatic data, topographic maps (at a scale of 1:50,000 or larger), commercial off-the-shelf (COTS) data and Google Earth (only for interpretation purposes) [57]. All the input data for the UA products are described by metadata in accordance with the INSPIRE metadata profile specifications and guidelines [58]. Similar to the 2006 Urban Atlas inventory, the EO data for the Urban Atlas 2018 is an optical Very High Resolution (VHR) coverage over EU (during the period from 2017 to 2019). The VHR mosaics used were created in pseudo-natural color at 2.5 m resolution by the Joint Research Centre (JRC). The imagery was derived from Pléiades 1A & 1B, SuperView-1, Kompsat-3/3A, PlanetScope, Spot-6/7, TripleSat and Deimos-2 source data with green, red, blue and near the infrared band [59]. A critical aspect that contributed to the choice of the use of the data provided by Urban Atlas was the availability of distinct data sets for periods long enough to capture the changes in land use. The oldest available products refer to the situation in 2006, while the newest ones describe the land use at the level of 2018 [56].

The availability of data for each of the previously mentioned years was the basis for selecting the cases included in the analysis. The number of inhabitants in each of the Functional Urban Areas (FUA) registered in the inventory was another criterion used in the study areas selection. For the analysis results to be representative of the general situation of Romanian cities, the capital city of Bucharest was excluded from the analysis because the large number of inhabitants and their high density determine its own dynamics.

The inventory includes five main land use classes: Class 1—artificial surfaces, Class 2—agricultural areas, Class 3—natural and semi-natural areas, Class 4—wetlands, and Class 5—water. Class 1 is further divided as follows: residential buildings (with six subcategories); industrial areas and transport units (with six subcategories); areas with strong human influence (three subcategories); and artificial areas for recreation (two subcategories). As for the other main classes, the 2006 database includes three classes for agriculture, forests, and water. The 2012 and 2018 databases include ten more subcategories: five for agricultural areas, three for natural and semi-natural areas, one for wetlands, and one for water [60]. The minimum mapping unit is 0.25 ha for artificial surfaces and 1 ha for the other surfaces [61].

The degree of vulnerability is influenced by the exposure of various elements to different types of hazards; therefore, the analysis included two of the most common natural hazards, in terms of frequency, in our country: floods and earthquakes [62]. In addition to natural hazards, the degree of industrialization of large cities in Romania has determined the need to include technological hazards manifested in the form of toxic dispersions, fires, or explosions in the analysis. These hazards may occur due to the technological processes specific to large industrial operators whose activity is regulated by transposing the SEVESO directives in the Romanian legislation.

The land use classes were extracted from the Urban Atlas database for each of the selected case studies. A vulnerability level was then assigned based on how potential natural or technological disasters may affect these cases. The four qualitative rankings of vulnerability were selected by the authors, based on previous studies that provide sufficient details on vulnerability types and rankings and considering the potential consequences that certain disasters may have on the population in the land use areas [32,63,64,65,66,67]. Apart from the direct consequences on society, the indirect impacts were also considered: for example, a flood could damage large surfaces of arable land and permanent crops, resulting in significant monetary losses for the communities.

Four qualitative rankings of vulnerability were established for each land use type in the Urban Atlas inventory and correlated for each of the natural or technological hazards mentioned above. The four vulnerability levels were:

A—very high vulnerability; includes those land uses that are characterized by a very high density of population and anthropic elements (blocks of residential flats, critical infrastructures, etc.); these land uses include also sites of community importance and other special protection areas; a disaster could cause severe damage and human losses and would disrupt social activities.

B—high vulnerability; includes those land uses that are characterized by a high density of population and anthropic elements (residential areas, infrastructure, etc.); a disaster could cause material damage, could affect the population, and disrupt social activities.

C—medium vulnerability; includes those land uses characterized by few anthropic elements or whose destruction would cause minor damages; the environment may be affected by the consequences of a disaster.

D—low vulnerability; includes those land uses that are characterized by a lack of anthropic elements.

The structure of the Urban Atlas inventory involves the division of areas into classes and subclasses of land use. The classification (

Table 1. Vulnerability level assigned to land use.

Urban Atlas Nomenclature			Exposed Elements	Consequences	V. E.	V. Fl.	V. F./Ex.	V.T.
1. Artificial Surfaces
1.1. Urban fabric	1.1.1. Continuous urban fabric (S.S. >80%)		Population Buildings	Trapped people, reduced mobility	A	A	A	A
	1.1.2. Discontinuous urban fabric	1.1.2.1. Discontinuous dense urban fabric (S.S. 50–80%)	Population Buildings	Trapped people, reduced mobility	A	A	A	A
		1.1.2.2. Discontinuous medium density urban fabric (S.S. 30–50%)	Population Buildings	Trapped people, reduced mobility	A	A	A	A
		1.1.2.3. Discontinuous low density urban fabric (S.S. 10–30%)	Population Buildings	Trapped people, reduced mobility	B	B	B	B
		1.1.2.4. Discontinuous very low density urban fabric (S.S. <10%)	Population Buildings	Trapped people, reduced mobility	C	B	B	B
	1.1.3. Isolated structures		Buildings	Damages buildings	C	D	D	D
1.2. Industrial, Commercial, Public, Military, Private and Transport Units	1.2.1. Industrial, Commercial, Public, Military and Private Units		Population Buildings Commercial areas	Trapped people, reduced mobility, multi-risk	A	A	A	B
	1.2.2. Road and rail network and associated land	1.2.2.1 Fast transit roads and associated land	Infrastructure (buildings, transport lines)	Reduced mobility and response capabilities	A	A	A	C
		1.2.2.2. Other roads and associated land	Infrastructure (buildings, transport lines)	Reduced mobility and response capabilities	B	B	B	C
		1.2.2.3 Railways and associated land	Infrastructure (buildings, transport lines)	Reduced mobility and response capabilities	A	A	A	C
	1.2.3. Port areas		Infrastructure (buildings, transport nodes)	Reduced mobility and response capabilities	A	A	A	C
	1.2.4. Airports		Infrastructure (buildings, transport nodes)	Reduced mobility and response capabilities	A	A	A	B
1.3. Mine, dump and construction sites	1.3.1. Mineral extraction and dump sites		Heterogeneous areas	-	D	B	B	C
	1.3.3. Construction sites		Construction sites	Damages buildings	B	C	C	D
	1.3.4. Land without current use		Land	-	D	D	D	D
1.4 Artificial non-agricultural vegetated areas	1.4.1. Green urban areas		Population	Reduced mobility	B	B	C	C
	1.4.2. Sports and leisure facilities		Population	Reduced mobility	B	C	C	C
2. Agricultural
2.1. Arable land			Arable land	-	D	C	D	D
2.2. Permanent crops			Crops	-	D	C	D	D
2.3. Pastures			Pastures	-	D	D	D	D
2.4. Complex and mixed cultivation patterns			Heterogeneous areas	-	D	D	D	D
3. Natural and semi-natural areas
3.1. Forests			Forests	-	D	C	D	D
3.2. Herbaceous vegetation association			Heterogeneous areas	-	D	D	D	D
3.3. Open spaces with little or no vegetation			Heterogeneous areas	-	D	D	D	D
4. Wetlands
			Wetlands	-	D	D	D	D
5. Water
			Water	-	D	D	D	D

V.E.—Vulnerability class earthquake; V.Fl.—Vulnerability class floods; V.F./Ex.—Vulnerability class fires/explosion; V.T.—Vulnerability class toxic dispersion.

) of an area as having a specific level of vulnerability (out of the four described) was carried out at the level of subdivisions from the 2018 Urban Atlas nomenclature [68] as follows:

As there were minor differences in the structure and nomenclature of the products from 2006 and 2018, changes were made in how the vulnerability level was scored in order to produce comparable results. In this regard, the cemeteries (class 1.4.1 in 2018 nomenclature) were assigned the same vulnerability value as class 1.2.1, while classes 3.2, 3.3 and 4 were grouped and assigned the same vulnerability value as class 2.

The continuous urban fabric was considered to have very high vulnerability (A level) due to the high population density and anthropic elements (residential areas, infrastructure, etc.). The same vulnerability level characterizes the port areas and airports because they are considered critical infrastructures, any disturbance having severe consequences in several fields of activity.

The high vulnerability level (level B) includes discontinuous urban areas, transport infrastructure, green areas, and sports and leisure facilities. The discontinuous urban fabric class, similar to the continuous, was considered to be vulnerable due to the high density of population and built elements. Generally, these land use classes may be found in urban or rural areas. Road and railway transport networks were also considered highly vulnerable and were assigned as level B because they were also defined as critical infrastructures, but with a lower density and, evidently, lower exposure. Vulnerable groups, mainly elder and children, spend much time in green areas and sport and leisure facilities, therefore these land uses were considered to be highly vulnerable.

Industrial or commercial units, building sites, water courses, and water bodies are included in level C (medium vulnerability). Building sites were included in this category due to the low number of individuals present onsite, compared to the continuous urban fabric. The exposure of these individuals is limited to a few hours per day.

Level D (low vulnerability) includes the other land uses not discussed previously. Although they may be affected by a disaster (for example, vegetation destruction), the consequences are not severe enough to require special land use planning measures.

GIS (Geographic Information System) techniques were used to determine the extent to which territorial changes contributed to the change of vulnerability level. This degree was expressed as a percentage variation (increase or decrease) on the surface of the areas included in a particular class of vulnerability for the 2018 situation as compared to the initial situation in 2006, using the following equation:

$$\sigma =\frac{\sum S_{2018}-\sum S_{2006}}{\sum S_{2006}}\times 100$$

(1)

where:

σ (%)—vulnerability variation;

S₂₀₁₈ (ha)—surface area of territories included in a particular vulnerability class (hazard dependent) for 2018;

S₂₀₀₆ (ha)—surface area of territories included in a particular vulnerability class (hazard dependent) for 2006.

For example: if the sum of all areas classified as having a very high level of vulnerability for a specific hazard, in 2006, is 100 hectares and the sum of all areas classified as having a very high level of vulnerability for a specific hazard, in 2018, is 120 hectares, we get a positive vulnerability variation of 20 percent. If the sum of all areas classified as having a very high level of vulnerability for a specific hazard in 2018 is smaller compared to the situation in 2006, we get a negative vulnerability variation.

This analysis was reiterated for each type of hazard selected for this study to determine if the trends of territorial change contribute to a more substantial influence for certain types of risk.

Temporal Variation of Vulnerability Level in the Areas Exposed to Technological Hazards—Comparative Analysis for Territorial Compatibility

The legislation regulating the activity of large industrial operators is very well defined, including from the point of view of territorial planning. This well-defined legislation provided us the opportunity to conduct a more detailed analysis (characterized by less uncertainty than that induced by natural hazards) of territorial dynamics in hazardous areas. This analysis was performed for all previously selected case studies and involved identifying large industrial sites operating in the study areas as a first step.

Large industrial operators are referred to as those legal entities that store or use large quantities of dangerous substances and thus fall within the provisions of Romanian Law 59/2016 [69], which transposes the requirements of the Seveso III Directive [70].

Although the list of these operators is available on the websites of county environmental protection agencies, safety studies and major accident prevention policies are not always made public for safety reasons. In this context, the information regarding the location of Seveso operators was taken from the risk assessment study at national level in Romania—the RO-RISK project [71].

The initial selection involved overlapping the vector database containing the locations of the operators and the administrative boundaries of the FUA selected as case studies.

For a more detailed analysis of the areas exposed to technological hazards, a maximum of two Seveso operators were selected per each analyzed FUA. This selection was made based on the availability of information regarding the accident processes and scenarios prepared within the safety documentation from the RO-RISK project.

To determine the changes regarding the vulnerability level in the proximity of Seveso operators, the worst possible accident scenario and the emergence distances of the physical effects on the population were identified for each selected operator. These distances are obtained by modeling the effects of an accident and correspond to a certain level of pollutant concentration, thermal radiation or overpressure that may cause exposed people to lose their lives, and suffer irreversible (permanent damage to the respiratory system, hearing loss, irreversible cellular injury, etc.) or reversible (dizziness, minor skin burns, pain, etc.) effects. Thus, the areas exposed to technological hazards (toxic dispersion, fire, explosion) are considered the areas with reversible effects on the population. After this step, the procedure for quantifying and expressing the vulnerability changes is identical to the one described in the previous section.

The Directive 2012/18/EU (Seveso-III Directive) “on the control of major-accident hazards involving dangerous substances” brings to the fore the issue of territorial planning in the vicinity of large industrial operators. By transposing this into Romanian legislation, the Law 59/2016 and the Ministerial Order no. 3710/1212/99/2017 [72] require the establishment of adequate distances from potential sources of risk within the Seveso sites, imposing the need to establish territorial compatibility. Territorial incompatibility is defined as “the situation in which there is no compliance with the provisions of the methodology regarding the distribution of constructions and functional zonings around the sites that fall within the provisions of Law no. 59/2016” [72]. The provisions and application of the Romanian methodology for establishing territorial compatibility are discussed extensively by Török et al., 2020 [73].

The functional areas described in MO 3710 were classified into four categories (based on the vulnerable territorial elements that define them) to “ensure a minimum level of security for the population, for economic activities, infrastructure, and environment” [72]. Precisely this link between functional areas and vulnerable elements justifies the need to develop an analysis to establish the relationship between territorial compatibility and land use changes. From a practical point of view, this relationship can be established by applying the territorial compatibility matrices for the selected scenarios, both for the situation in 2006 and that in 2018. The territorial compatibility analysis was performed using the Urban Atlas inventory to support the establishment of the functional areas. The classification of Urban Atlas land use classes in categories of functional zones, the establishment of impact areas based on physical effect thresholds for accident scenarios, and application of the compatibility matrix for the existing operators were made following the provisions of MO 3710 and the detailed methodology published by [73]. The accident frequencies for the analyzed scenarios were derived from HSE’s documentation “Failure Rate and Event Data for use within Risk Assessments” [74].

3. Results and Discussions

3.1. Temporal Variation of Vulnerability Level in Developing Cities

By comparing the data available in the Urban Atlas Inventory for Romania for both 2006 and 2018, three representative cities (FUAs) were identified to meet the methodological requirements: Cluj-Napoca, Timișoara, and Oradea (see Figure 1).

These cities were built around a historical center, but the communist regime influenced their development: huge industrial structures were surrounded by residential areas, with numerous blocks of flats designed for the working class, leading to high urban density [75]. In recent years, the urbanization process that characterized these cities resulted in several changes of land use, the natural and agricultural landscapes being transformed into built-up areas with different land uses (residential, commercial, or industrial) [76,77,78,79].

Cluj-Napoca

Cluj-Napoca, the Cluj county seat, located in the northwestern part of Romania, is a big industrial center (ICT sector, chemical, and pharmaceutical industry, financial and health services) and an important university center. Therefore, it attracts many students and professors from the entire region and a high number of people in the high-income sector, such as the growing ICT sector [80]. Cluj-Napoca is one of the few cities in Romania where suburbanization and suburban expansion is not paralleled by an increase in the population or economic contraction. On the contrary, the city is a centralizing pole for investments and funding opportunities, multiplying this potential to the smaller communities in its polarizing region [81].

The topography in the urban core area (Figure 2) has a restrictive character in terms of concentric expansion of artificial surfaces. The city is practically located in a valley bordered by steep slopes. Thus, the pressure for territorial changes manifested itself at the level of the other localities from FUA. In this context, Floresti locality (to the east of Cluj Napoca Municipality) became the rural area with the largest number of inhabitants in Romania. During the analyzed period, the number of inhabitants increased five times from approximately 7000 people in 2006 to over 35,000 in 2018. This trend was also manifested at the level of living spaces: in 2018 there were seven times more homes than in 2006 [82]. These changes were manifested at the level of all localities in the city’s vicinity, the only restrictive factor being represented by the road transport infrastructure.

The data presented in

Table 2. Summary of vulnerability level variation between 2006 and 2018 in Cluj-Napoca FUA.

Vulnerability Class	Natural Hazards		Technological Hazards
	Earthquake	Floods	Toxic dispersion	Fire/Explosion
A (Very high)	23.4%	23.4%	23.1%	23.4%
B (High)	24.2%	50.0%	40.8%	54.7%
C (Medium)	135.2%	−3.9%	26.5%	50.7%
D (Low)	−3.6%	6.7%	−3.9%	−4.1%

indicate that this dynamics directly impacts the vulnerability level by generating a tendency of transition from areas with low vulnerability to areas characterized by a higher vulnerability degree for all types of hazard brought into question. This growing trend of vulnerability comes amid urban expansion as uptake of agricultural area (about 70% of the total territorial changes).

The most relevant discussion for the Cluj-Napoca FUA case study can be carried out around the risk of floods. Although the areas with high and very high vulnerability in case of floods have expanded considerably during the analyzed period (Figure 3), the general perception is that the level of risk for floods is relatively low due to lack of exposure. This lack of exposure is linked to the fact that the flow of the largest river in the analyzed area (Somesul Mic) is regulated through hydrotechnical facilities located upstream of the city. The real situation is completely different, though: the analysis showed that the transition to artificial surfaces took place exactly along the tributaries to the Somesul Mic river, where flows are not regulated and floods can occur (especially flash-floods). On the other hand, urbanization implies an increase of the built-up and impermeable areas, and it has several hydrological impacts, such as increased runoff or decreased water infiltration. These aspects can lead to urban flooding in case of heavy rains when the sewage system is overloaded. Such local events occurred with an increased frequency in Cluj-Napoca over the past years. Even though these events covered small areas and had a minimum impact on the community, both the frequency and the associated consequences of heavy rains are expected to increase in the context of climate change.

For this case study, the peak ground acceleration (PGA) specific to a 1000 year return period earthquake ranges between 101–250 cm/s² throughout the entire area [83]. From this point of view, any territorial change that influences the level of vulnerability also determines a variation in the risk associated with earthquakes in Cluj-Napoca.

The next section presents a detailed analysis of the variation of vulnerability level in the areas currently exposed to technological hazards. Nevertheless, the spatial planning process must consider the expansion of areas vulnerable to these hazards near future industrial sites.

Timișoara

Timișoara, the Timiș county seat, is situated in the western part of Romania, and its industrial tradition has continued even after the socialist period. Today, it is an important industrial and academic center, also associated with emigrants’ flows and traded goods [75]. After the post-socialist period, the population concentrated in the surrounding rural areas. However, the next phase of urban development will transform more distant rural localities into urban districts [84]. Furthermore, the urban restructuring that took place in recent years changed the functionalities of different areas: derelict production plants may be seen in the inner-core of Timișoara [85], while the new marginal urban areas fulfill more functionalities, such as residential, industrial or other uses (Figure 4).

From the perspective of the limiting character constituted by the topography, Timișoara is located on the other side of the spectrum compared to Cluj-Napoca, as it is situated in a plain area, thus facilitating the uniform expansion of the built elements. This expansion of urban areas to the detriment of agricultural areas translated into 90% of the changes in land use. The way these changes materialize in the variation of the vulnerability level is expressed in

Table 3. Summary of vulnerability level variation between 2006 and 2018 in Timișoara FUA.

Vulnerability Class	Natural Hazards		Technological Hazards
	Earthquake	Floods	Toxic Dispersion	Fire/Explosion
A (Very high)	5.5%	5.5%	3.7%	5.5%
B (High)	10.4%	44.6%	27.1%	59.4%
C (Medium)	329.7%	−5.4%	6.4%	−6.0%
D (Low)	−4.2%	56.5%	−4.7%	−4.4%

and the graphical representation of these variations is made in Figure 5. The most significant transition was the expansion towards high vulnerability (Class B) areas. However, the situation is better than Cluj-Napoca’s in terms of proportions and actual surfaces where the vulnerability level has increased.

In the specific case of earthquakes, we can see that the vulnerability class C (medium vulnerability) displays the highest increase (three times the initial situation). This increase is due to the conversion of 276 hectares of predominantly agricultural land into discontinuous, very low-density urban fabric areas. While PGA values of 1000 year return period earthquakes vary between 251–500 cm/s² [83], we can say with certainty that territorial changes occurring in this area contribute to a higher level of risk due to the expansion of vulnerable areas.

Through data analysis, we also see that the vulnerability variation trend is favorable to the expansion of areas with low vulnerability in the case of floods. The lower vulnerability level resulted from converting agricultural land into unused land and expanding water-covered areas. Vulnerability to technological hazards increased during the analyzed period. Increasingly more areas fall into vulnerability classes A and B with the density increment of artificial surfaces.

Oradea

Oradea (Figure 6), the Bihor county seat, is located in the northwest of the country, and it is a successful example of urban environment regeneration. Abandoned industrial sites were transformed into residential areas or commercial units and integrated into urban areas [77]. In recent decades, numerous privately or publicly funded projects have invested in public spaces and leisure facilities, brownfields clean-up for future development plans, and extended green areas. Following the emergence of the metropolitan area and the city’s economic development, large areas of agricultural land shifted to other land use categories: residential, commercial, or industrial [6].

The case of Oradea Municipality is very similar to that of Timișoara in terms of the area topography (plain area), location near the western border of the country and development trends. The general trend was urban expansion to the detriment of agricultural lands in two-thirds of the cases, but also loss of artificial areas (15% of the cases) due to the decommissioning of old industrial areas. This also translates into the dynamics of the vulnerability level in the 2006–2018 period (

Table 4. Summary of vulnerability level variation between 2006 and 2018 in Oradea FUA.

Vulnerability Class	Natural Hazards		Technological Hazards
	Earthquake	Floods	Toxic Dispersion	Fire/Explosion
A (Very high)	8.3%	8.3%	4.5%	8.3%
B (High)	6.9%	46.8%	46.2%	55.1%
C (Medium)	713.8%	−7.1%	−2.4%	−11.5%
D (Low)	−5.3%	72.6%	−5.4%	−5.0%

and Figure 7); the proportions are similar for the two case studies.

The substantial increase observed in earthquake vulnerability class C is due to the conversion to the 395 hectares of discontinuous, very low-density urban fabric used in agriculture. The PGA values of the 1000-year return period earthquake events in this area range between 101–250 cm/s² [83].

The expansion of low flood vulnerability areas results from converting 200 hectares of agricultural land into lands without current use. Regarding exposure to floods in the studied area, it is worth mentioning that the main water collector (Crișul Repede) goes through a series of hydrotechnical works for damming and flow control, thus reducing flood probability and floodable areas.

The more significant increase of high vulnerability areas in case of technological accidents is due to the density increase trend of the artificial elements built and, implicitly, of the population in the residential areas.

3.2. Temporal Variation of Vulnerability Level in the Areas Exposed to Technological Hazards—Comparative Analysis for Territorial Compatibility

The industrial sector has played an essential role in the development of Romanian cities since the communist period. Many industrial parks were created precisely to attract people from rural areas. Therefore, a more detailed analysis was conducted to see whether the trends in current development and spatial planning influence in some way the vulnerability level in areas exposed to technological hazards.

Five cases were selected from the more than 300 Seveso sites operating in Romania, based on the criteria described in the methodology, as shown in Figure 8. Regarding the number of Seveso operators in the three case studies, we can see that Timișoara has the largest number of such operators (eight sites), followed by Oradea (four sites) and Cluj-Napoca (two sites). The larger number of sites in Timișoara and Oradea can be attributed to the local territorial development policies and the strategic position of these two cities at the western Romanian border. This positioning facilitates both the supply of raw materials and the sale of end products on Western markets using the more developed infrastructure in neighboring countries.

The availability of data criteria played the most important role in selecting the analyzed sites. Thus, for the case studies from Timișoara and Oradea, two sites were selected in each city. In the case of Cluj only one operator met the analysis conditions. The names of the operators and the exact position are not disclosed for security reasons, but their characteristics and the results of the temporal vulnerability variation analysis are described below:

Cluj-Napoca 1—lower-tier Seveso establishment storing petroleum products and their derivatives. The relevant hazardous substances are gasoline and diesel, and the maximum storage capacity at the site is more than 15,000 cubic meters at the borders of the Urban Core, in a predominantly industrial area.

The worst-case scenario identified by the operator refers to a BLEVE (boiling liquid expanding vapor explosion) event that occurred at a 5000 cubic meter gasoline tank. Neither the accuracy nor the software used to carry out the accident modeling by each operator is the subject of this analysis. The results were considered relevant as the authorities accepted them based on the legal regulations in force. The distances resulting from the accident modeling refer to the effects of thermal radiation on the population and are divided into three areas:

High lethality area—over a distance of 109 m and corresponds to the fireball radius;

Irreversible effects area—the threshold located at a distance of 459 m for the static heat radiation value of 5 kW/m²;

Reversible effects area—over a distance of 890 m (the exact threshold value was not stated in the documentation).

Timișoara 1—higher-tier Seveso establishment operates in the storage and bottling of liquefied gases. The maximum storage capacity at the site is over 2200 m³; the largest quantities of hazardous substances are recorded by butane (stored in spheres of 1000 m³) and propane (stored in tanks of 100 m³). The establishment is located at the extremity of the study area, in a predominantly industrial zone mixed with areas used for agriculture.

The highest emerged distances of physical effects and consequently the worst accident scenario are represented by an accident that occurred at one of the spheres of butane storage, causing a BLEVE type event. The distances related to the thresholds for the emergence/occurrence of the physical effects taken from the documentation are 223 m for the high lethality area, 733 m for the irreversible effects zone, and 1308 m for the reversible effects zone. The safety report does not include the exact threshold values used in consequence modelling.

Timișoara 2—lower-tier Seveso establishment which also operates in the storage and bottling of liquefied gases. The maximum storage capacity at the site is approximately 135 metric tons of liquefied petroleum gas (LPG) of various compositions. The LPG is stored in several pressurized tanks with volumes of 50 and 100 cubic meters. This site is located closer to the center of Timișoara urban core, in an area with mixed land use patterns.

The worst-case scenario described in the establishment’s Major-Accident Prevention Policy refers to a BLEVE event produced at a propane pressurized storage tank. The distances related to the thresholds for the physical effects emergence taken from the documentation are 287 m for the high lethality zone (radius of the fireball), 639 m for the “beginning of lethality” threshold (10 kW/m² for 60 s exposure), 901 m for the irreversible effects area (5 kW/m² for 60 s exposure) and 1400 m for the reversible effects area (2 kW/m² for 60 s exposure).

Oradea 1—lower-tier Seveso establishment in the field of energy production. Hydrazine hydrate (24–35%), diesel, heavy fuel oil (HFA), ammonia, acetylene, and liquefied oxygen are amongst the substances stored on site. The site is located in a predominantly industrial area of the city.

The worst-case scenario identified by the operator refers to an accidental release of ammonia from the storage tank. For a 60-min exposure, based on threshold concentrations specified in Acute Exposure Guideline Levels (AEGLs), the following distances resulted from the modeling: 91 m for the beginning of lethality threshold (AEGL 3–1100 ppm), 246 m for the irreversible effects area (AEGL 2–160 ppm) and 575 m for the reversible effects area (AEGL 1–30 ppm).

Oradea 2—higher-tier Seveso site with activity in the domain of transport, warehousing, and logistics. Various amounts of more than 150 hazardous substances are stored on-site. The site is located on the outskirts of the study area in a predominantly industrial area. The worst-case scenario described in the Safety Report refers to a toxic dispersion of sulfur dioxide as a result of a terrorist or air attack destroying warehouses. The sizes of the areas (as stated by the operator) with effects manifested on the population after exposure to the toxic cloud are: 1400 m for the high lethality threshold (LC₅₀, lethal concentration from which 50% of the affected population dies), 2200 m for the irreversible effects area (IDLH, Immediately Dangerous To Life or Health Values), and 3500 m for the reversible effects area (AEGL 1, 0.2 ppm).

The territorial changes that occurred in the 2006–2018 period, and implicitly the changes in the vulnerability level, were quantified for the areas delimited by the threshold of reversible effects on the population. The summary of the vulnerability variation presented in

Table 5. Summary of vulnerability level variation between 2006 and 2018 within the area with reversible effects for worst-case scenarios.

Vulnerability Class	Seveso Establishment
	Cluj 1	Timisoara 1	Timisoara 2	Oradea 1	Oradea 2
A (Very high)	19.3%	0.9%	2.4%	-	3.2%
B (High)	69.3%	17.0%	57.9%	−21.5%	57.9%
C (Medium)	100.0%	0%	0%	0%	−18.6%
D (Low)	−13.7%	−0.8%	−18.9%	37.5%	−7.2%

indicates that for all cases, with one exception, the change in land use determines a dynamic of increasing the vulnerability level in areas with reversible effects on the population in case of a technological accident.

The analysis of territorial changes in the hazard area specific to the Cluj 1 site (Figure 9) revealed an increase of the surfaces registered in the A, B, and C vulnerability classes.

The increase of almost 20%, observed for the very high vulnerability areas occurred due to the construction and completion of the bypass in the eastern part of the city in the period between the two reference years. As this bypass is part of a national road regime and the nearby lands do not physically delimit it, the construction of this type of road infrastructure is for the owners of the neighborhood a better way to access their lands and provides the opportunity to change the land use to increase its value. This dynamic is also suggested by the increase in the number of areas included in the category of high vulnerability. Many of the areas used in agriculture in 2006 are now covered by low-density urban fabric and isolated structures.

In the 2018 data inventory, there is also an area framed as green urban area and, therefore of medium vulnerability. Following a more detailed analysis based on recent satellite images, the authors found that the construction of the bypass generated a possible error in the Urban Atlas product. The UA algorithm assumes the existence of green urban spaces where “at least two sides (of forests or green areas) are bordered by urban areas and structures” [73,86]. Since there are no “visible traces of recreational use” on site [86], we can conclude that this area was incorrectly assigned to the green urban area class, as it was in fact an area covered by vegetation characterized by a low vulnerability degree.

Low vulnerability level areas experienced a downward trend, as arable land and pastures were converted into land with other uses. The values of forest areas remained the same.

In the area exposed to hazard specific to the Timișoara 1 site (Figure 10), there were no significant changes in the type of land use; the increase observed in high vulnerability areas is mainly due to the expansion of the local road network. Although an expansion by 17% is observed compared to the situation in 2006, the net expansion is only 1.4 hectares.

The situation specific to the Timișoara 2 site (Figure 10), on the other hand, is slightly different due to its position closer to the city center, which implies a greater pressure from the developers of real estate and commercial projects. Thus, in the exposed areas, the level of vulnerability increased because of the discontinuous low and very low-density urban fabric in areas initially occupied by vegetation or used in agriculture.

A singular case is represented by the changes in the use of land which occurred in the areas exposed to the hazards specific to the operator Oradea 1 (Figure 11). In this case, there is a decreasing trend of vulnerability degree due to restructuring of the old industrial areas. Currently, these areas are represented by land without use.

In the relatively large area exposed to the hazards specific to the Oradea 2 site, between 2006 and 2018, 50 hectares without use or previously used in agriculture were transformed into areas with low and very low-density urban fabric.

Following the analysis of territorial compatibility based on the legal methodological norms published within the Ministerial Order 3710/2017 and described in detail by [73] territorial incompatibilities have been identified only in two out of the five cases. The main reason that determined these results is related to the very low frequencies that characterize the worst scenarios.

The failure rates and event data published by the Health and Safety Executive indicate a specific frequency for the Cluj 1 scenario between the values of 1 × 10⁻⁵ and 1 × 10⁻⁶ per year. By contrast, all other scenario frequencies fall below the threshold of 1 × 10⁻⁶ per year. This led to the application of the most permissive matrices in terms of compatibility.

For the scenario characteristic for the Cluj 1 site, the territorial incompatibility occurs in the area with irreversible effects through the existence of a residential group (discontinuous dense urban fabric) spread over an area of approximately two hectares. The situation of territorial compatibility has not changed over time, as the results are similar for both 2006 and 2018.

For the scenario described in the documentation drafted by the operator of the Timișoara 2 site, territorial incompatibilities were identified within the area where the effects of the accident determine the beginning of lethality in the exposed population. This incompatibility is manifested on an area of sixteen hectares covered by continuous urban fabric and discontinuous dense urban fabric. Since at the level of these two classes of land use there were no changes in the period 2006–2018, the situation of territorial compatibility remained unchanged.

3.3. Uncertainties

Three sources of uncertainty influence the veracity of the results presented in this study. The first source is credited to the way in which the Urban Atlas algorithm assigns land use classes in a manner close to reality. The other sources of uncertainty are derived from the way in which the level of vulnerability is assigned for the different classes of land use and the way in which the potential effects produced on the population (as a result of a technological accident) were quantified by the industrial operators.

A thematic accuracy assessment (

Table 6. Thematic accuracy assessment for 2006 UA data (Source: UA 2006 LU/LC product delivery report).

	Overall Accuracy
	Cluj-Napoca	Timisoara	Oradea
Urban Classes	88.5%	93.6%	88.5%
Rural Classes	83.2%	90.8%	83.2%
Overall Area	76.4%	85.1%	76.4%

) and (

Table 7. Thematic accuracy assessment for 2018 UA data (Source: UA 2018 LU/LC product delivery report).

	Overall Accuracy
	Cluj-Napoca	Timișoara	Oradea
Urban Classes	89.6%	89.2%	82.2 %
Rural Classes	83.7%	81.2%	78.9%
Overall Area	76.3%	78%	71.8%

) for the Urban Atlas products used in this study was derived from the information contained within each product delivery report. The results of this evaluation indicate that the data is mostly reliable: >80% thematic accuracy for both urban and rural classes.

Depending on the tendency of erroneously assigning the actual type of land use in different classes, the general vulnerability level can be underestimated or overestimated by up to 20%. This tendency has been studied in the literature by comparing the Urban Atlas products with data from the Cadastre of Real Estate. The results indicate that the discrepancies occur especially in the case of artificial areas [55].

The manner of assigning the vulnerability degree is characterized by uncertainty because different areas, although included in the same land use class, may have different specific characteristics. In this sense, it is important to mention that the study results are rather general and can help develop sustainable policies in land use. Therefore, a quantitative assessment of vulnerability indicators for homogeneous areas is recommended to implement targeted measures for vulnerability level reduction.

The last-mentioned source of uncertainty is present only in the section related to technological hazards and territorial planning in the area where they affect the health of the population. The analysis showed that different operators used different thresholds, different software, and different parameters in modeling the physical effects. Still, the results of the security studies were considered relevant since they were within the legal norms at the time of their publication.

3.4. Land Use Planning in the Context of Hazard Vulnerability

The rapid development of urban areas represents a complex challenge for the decision-makers involved in the land use planning process, this process being an important tool for a sustainable development and impact mitigation. The planning system is based on policy instruments that can influence the spatial changes regarding the land cover and should include environmental protection principles in order to achieve urban sustainability [87,88]. The inclusion of the sustainability concept in the urban planning and strategies making was supported by the adoption of Directive 2001/42/EC (Strategic Environmental Assessment—SEA Directive), which offers a legal framework and a ground for sustainable development [87,89].

In Romania, the spatial planning process was characterized by a strict control of the urban development during the communist regime, followed by a lack of legal framework after the fall of this regime. Therefore, the urban expansion took place quickly and chaotically, becoming an issue [88]. Moreover, the integration of European legislation was slowly and poorly done, the increased urban development taking place during a period in which policy planning was inefficient and scarce [87]. Țîncu et al. [90] stated that the policies in Romania should focus more on vulnerable communities and planning strategies should target the areas where land use planning is inefficient. Another issue bring into discussion was the fact that these strategies are paying more attention to economic development and less to reducing the impact on society. In this study, it was demonstrated that the land use changes that took place in the past years increased the urban vulnerability to different hazards, highlighting the need of decision support tools that integrate disaster risk assessment as well vulnerability analysis. According to [91], risk-based land use planning allows the policy-makers to determine what is considered an acceptable level of risk and to develop future land use plans taking into account the potential consequences of hazards.

The growing of artificial surfaces to the detriment of natural ones may induce several negative consequences, such as the increase of frequency and intensity of natural disasters, pollution of soil and water resources, decreasing quality of life, etc. [54]. For example, in the case of flood hazard, the artificial surfaces affect the water infiltration and soil permeability, increase the surface runoff, and decrease evapotranspiration and groundwater recharge, all these effects being discussed in detail by Oudin et al. [92]. Furthermore, many urban areas are developing along water courses, in flooded areas, thus more and more socio-economic assets and people are exposed to flood hazard. Therefore, these aspects must be taken into consideration in the land use planning process, besides socio-economic profit. This can be done through a cost benefit analysis, which should also include disaster risk mitigation measures.

Furthermore, Rouillard et al. [93] state that for successful policy implementation in the field of disaster risk management, public participation and social learning are two important factors. The engagement of stakeholders and population in decision-making activities facilitates/promotes a participative process, in which decisions are made taking into account public needs and are based also on local knowledge. Moreover, social learning enables collaborative actions for disaster management, increasing public awareness and implication in the decision-making process [93].

In this study, social vulnerability to different hazards was assessed in the context of land use change, applying a simplified method that uses publicly available information regarding the evolution of land use in time, and data about the characteristics of different hazards. This approach allows the rapid identification of the most vulnerable areas to the analyzed hazards and offers an overview of the connection between land use change and disaster risk. The results showed that most of the differences in land use evolution are assigned to urban areas, these areas presenting a growing trend and extension over natural land use classes. Furthermore, it was demonstrated that these changes have a direct impact on the level of vulnerability. Of course, the level of vulnerability depends on many factors, such as the exposure (are the analyzed elements present in the affected area?), the characteristics of the exposed elements (for example, the density of building within the urban class), and the type and magnitude of the hazard. If the first two factors are based on information extracted from land use data, the latter factor refers to information regarding the frequency and return periods of the hazard. All these factors were taken into consideration when the vulnerability levels were assigned.

Considering the fact that the natural land use classes have a lower level of vulnerability compared to built-up classes, the transition of natural areas into built-up areas resulted in an increase in vulnerability. These results are consistent with other research in the field, as in the study by [54] who quantitatively analyzed the relationship between land use change and disaster risk, demonstrating that land use changes have caused high disaster risk. Rahman et al. (2021) [94] analyzed the changes in land use over a period of 17 years, the results showing that the increase in built-up areas led to an increase of flood vulnerability. This impact was associated on the one hand with the increased exposure of highly vulnerable elements associated with urban areas, and on the other hand to the fact that the growth of urban areas induces changes in soil characteristics, such as increased runoff on artificial surfaces.

The understanding of land use change and the impact that this change may have on vulnerability level can be used for future prediction regarding the relationship between land use and disaster risk. Furthermore, this information can be used for territorial planning, decision-making regarding risk reduction and sustainable urban development. The method proposed in this study can be included as an integrated part of the land use planning process, offering a preliminary and rapid analysis of vulnerability to different types of hazard and providing a ground base for future planning of land use. The proposed method presents some other advantages, compared to other studies in the field: (i) the use of GIS tools for data processing and analysis allows the visualization of spatial distribution of data and results, facilitating the understanding of the evolution processes; (ii) the use of high resolution data (10 m) provides more reliable results. According to Han et al. (2020) [95] the resolution of land use data represents a critical factor in the analysis of the impact of increased flood occurrence in built-up areas; (iii) moreover, the methods are not specific for the areas analyzed in this study, therefore can be applied in other areas and for different type of hazard.

4. Conclusions

This study is not intended as a manifesto against the expansion of cities in developing countries. However, its results indicate an increase in vulnerability level during the process.

Qualitatively speaking, since the general vulnerability level is an intrinsic characteristic of each type of land use, development and spatial planning policies should harmonize populated areas’ expansion processes with natural and technological risk reduction processes. With a higher vulnerability level, risk mitigation measures should aim in particular to reduce exposure, implement technical measures to minimize the consequences and improve the resilience of communities and infrastructures.

The particular case when the construction of the road infrastructure immediately generated a dynamic in the type of land use in its vicinity suggested that a monitoring period of 12 years might not be enough to understand the complexity of these phenomena. That is why the type of products registered in the Urban Atlas inventory are very important both in terms of their accuracy and resolution and in terms of their update frequency. The methodological processes indicated in this work should be repeated with the publishing of new Urban Atlas inventory products for a better understanding of urban dynamics, thus establishing the need for future studies on this topic. The impact on territorial changes produced by the 2017 legislative regulations on territorial compatibility in the vicinity of large industrial operators is also expected to produce quantifiable changes in the coming years and in the next versions of Urban Atlas.

Continuous, long-term monitoring of land use in developing cities is important so as not to draw wrong conclusions about the vulnerability trend. The trend observed in all three case studies for twelve years was the conversion of agricultural land to areas without current use. This trend may initially indicate a decrease in the level of vulnerability to flooding. In the long run, however, this stage can prove to be only transitional if these lands without current use are the basis for the development of future residential areas, generating, in total, a higher value of vulnerability.

The vulnerability maps representing the situation of territorial development for 2018 can build a foundation for territorial planning and development policies for both authorities and developers from all FUAs listed in the Urban Atlas inventory.

The scientific opportunity of studies aimed at spatial planning in relation to vulnerability level is revealed by the possibility to obtain comparable results for all areas listed in high resolution land use inventories. From a practical point of view, this type of approach facilitates the development of unitary and consistent policies, through which decision makers ensure an acceptable level of risk in relation to territorial changes.