2.1.4. The Null Hypothesis: Zero-Effect Reference Sample

One of the basic aspects of this GIS retro-historical evaluation methodology of human impacts in the territory is inter-temporal comparative analysis [35]. In this context, an analogous concept to a control sample of a laboratory test or the null hypothesis of a statistical analysis is necessary. In this sense, the case analyzed is very interesting because it has spatially geo-referenced information of all its territory with aerial photography in the years 1929, 1932, 1945, and 1956. This information is of great interest since this period between 1929 and the mid-1960s predates the arrival of tourism and the urbanization process of the coast, allowing a geo-referenced spatial comparative analysis.

We should also take into account that the level of accuracy in the geo-referenced information of both periods (1929–1956 and 1956–2017) is not the same. However, as can be seen in Table 1, the information available is sufficiently accurate to enable a temporary comparative analysis to be performed using GIS indicators of the impact of the urbanization process, the construction of ports and coastal infrastructures, and the development of large transformation of lands in the coastal perimeter.


**Table 1.** Technical characteristics of geo-referenced data used.

#### *2.2. Spatial Coastal Vulnerability Assessment: Methodological Considerations to the Model*

The assessment of coastal vulnerability must be carried out through contrasted and homogeneous scientific methodologies at a spatial level. An approximate estimation of a coastal global vulnerability Π*<sup>z</sup>* of the different areas of a territory can be obtained by means of an indicator that groups together the most common existing risks. This approximation Π*<sup>z</sup>* can be estimated as the homogenous sum of the different existing vulnerabilities as a result of natural hazards modeled through GIS indices *δ<sup>i</sup>* and

weighted by corrective coefficients *λ<sup>i</sup>* to statistically assess their probability. The format of global index (4), partial indexes (5), and weighting coefficients (6) will be as follows:

$$
\Pi\_z = \sum\_{\Pi} \lambda\_i \delta\_i \quad \text{with } \Pi \text{ \(e, 1\)}\tag{4}
$$

$$\delta = \{\delta\_1; \delta\_2; \delta\_i; \delta\_n\} \quad \text{with } \delta \in [0, 1] \tag{5}$$

$$
\lambda = \{\lambda\_1; \lambda\_2; \lambda\_i; \lambda\_n\} \quad \text{with } \lambda \in [0, 1] \tag{6}
$$

where partial indexes of vulnerability risks *δ* are detected in a territory and the weighting coefficients *λ* are obtained as follows:

$$\delta\_{\vec{l}} = \Phi \left[ \frac{T\_j}{l I\_M} \right] \text{ with } \delta \in [0, 1] \tag{7}$$

$$
\lambda\_i = \Psi \left( \frac{\lambda\_i}{\lambda\_{\text{in}}} \right) \text{ with} \\
\sum\_i \lambda\_i = 1 \tag{8}
$$

The function Φ will model the risk maps for each hazard variable *δ<sup>i</sup>* in a dimensionless way. This function is made up of a density map of average values T for each one of the *j* areas of the territory divided by the maximum values U of the territory. The function Ψ will establish the values of the weighting coefficients *λ<sup>i</sup>* as a function of a probability ratio of each of the variables in relation to an average value *λm*. This probabilistic assessment should be carried out based on a statistical evaluation criterion that can be modeled in a common way for all the risk variables *δ<sup>i</sup>* (in this case, the measurement will be based on the different return periods for flooding for example). It should be noted that the different GIS indicators may be modeling phenomena of a very different nature (floods of different origin, earthquakes, hurricanes, impacts of climate change, fire risk, etc.). Nevertheless, it is important to remember that the effect derived from these elements must be implemented in a homogeneous manner by generating as output units that can be added in a dimensionless way to perform the global coastal vulnerability index.

In this study, the flooding of land and marine origin have been selected as major risks to analyze the global assessment of the coastal vulnerability index of this territory. Other natural hazards, such as those derived from the risk of fire, seismic movements, or hurricanes, have not been taken into account since they are not significant cases with a negligible historical occurrence rate. The methodology for the evaluation of both flood risks has been developed as follows.

#### 2.2.1. Flood Risk of Marine Origin

To face the complexity of the different aspects that make up the calculation of the flood throughout this coastal territory, a three-phase methodology based on the criteria of the Methodological Guide of the Spanish National Flood Mapping System [48] has been followed.

In the first phase, the entire coastline is flooded only by the dynamics of the sea level (derived from the effects of astronomical and meteorological tides) without surf. With this approach, there are valid results in the areas where the waves have no relevance (inside estuaries or sheltered from external infrastructures). To do this, the extreme regime of flood elevation from the series of sea level data (available for more than 60 years of data) is adjusted for each position along the coast, calculating the level of flood associated with the return period T = 100 years with the Peaks Over Threshold (POT) technique [49] and adjusting the distribution function by means of the Generalized Extreme Value (GEV) statistical distribution. As a result, the sea level is obtained for each position, from which the corresponding Digital Terrain Model (hereinafter, DTM) benchmark will have to be subtracted from the coast to obtain the openwork on the ground.

In the second phase, the coastal areas where the waves hit directly are corrected, which is the coast line which is not protected from the waves. In this second approach, terrain profiles are drawn to resolve the flood in two dimensions, profile by profile, incorporating the combined effect of waves and sea level. The effect of the swell is evaluated by means of the two-dimensional numerical model IH-2VOF [50], which solves the Navier-Stokes equations, by using the Downscaled Ocean Waves (DOW) wave database [51], obtained from the data series of the C3E project [52]. This allows us to correctly characterize the wave propagated to the coast with a spatial resolution of at least 200 m.

Finally, in the third phase, the flood envelope is obtained by the sum of the flood zone by level and the flood zone by swell.

#### 2.2.2. Flood Risk of Land Origin

The delimitation of the flood zones is carried out by defining the so-called Significant Potential Flood Hazard Areas (SPFHAs). These areas are obtained from the Preliminary Flood Risk Assessment (hereinafter, PFRA) in accordance with Directive 2007/60 of the European Commission [53] in several ways:


In the case of considering structures of rolling or derivation of flows in the hydrological calculation it is considered that the flows are in an altered regime; otherwise they are deemed to be in a natural regime. For the analysis of the flood risks of land origin, the protocols established in the Methodological Guide for the development of the Spanish National Flood Mapping System [48] to model cartographic, hydrologic, geomorphologic and hydraulic boundary conditions have been used. In this case, since the most affected area has a more complex orography (see the results section below), Laser Imaging Detection and Ranging (LIDAR) tools have been used to generate the DTM. This higher level of precision is justified by the need for modeling the surfaces of the watersheds in a reliable way with the actual drainage directions. The cells used are at least 25 × 25 in agricultural or natural land and 5 × 5 in urban areas.

For this method, in relation to the concept of return period, it is important to make certain clarifications. In numerical terms, it is equivalent to the probability of having an equal or higher avenue flow in a given year, that is, the probability of exceeding the flow in a year. For example, for a return period of 100 years, that probability F(x) = 1/T = 1/100 = 0.01 = 1%. Thus, there is a 1% probability that one year this flow value will be exceeded and a 99% probability that it will not be exceeded. However, this does not imply that two or more avenues of such or higher intensity cannot occur within the same year, since the return period is a statistical concept and depends on the duration of the interval considered. Should we wish to calculate the probability of equaling or exceeding this value during a period of N years (statistical concept of Risk) for a return period T, it would be calculated by means of the following expression (9):

$$[1 - [1 - (1/T)]^N] \tag{9}$$

Thus, according to Table 2, an area affected by flooding in a period of 100 years return zone has a probability of 22.2% of being flooded in a period of 25 consecutive years and 39.5% to be flooded in 50 consecutive years.


**Table 2.** Probability of occurrence for T = 100 years.

On the other hand, it should also be noted that the calculated flood areas have important limitations regarding the flood that would occur in a specific event. Current techniques, although they are very precise, have important restrictions that make the actual flood of an event vary significantly from what was calculated. In this sense, we must highlight two important limitations:


### *2.3. Geostatistical Correlation between Human Actions and Coastal Vulnerability*

Once the distributions at the spatial level of the impact and coastal vulnerability indexes have been obtained, we can evaluate the possible spatial correlation between them by using geo-statistical methods. This analysis will allow us to assess to what extent the transformations made by human activity in the coastal perimeter of a territory have influenced the current coastal vulnerability existing in it. The spatial relationships will be parameterized and assessed through the use of Global Moran's I [54] and Anselin Local Moran's I [55] bivariate statistics, both are geo-processing tools from ArcGIS Pro 10.5.0 (ESRI Corporation, Redlands, CA, USA).

Bivariate global spatial autocorrelation will allow us to assess the statistical correlation of a set of geo-located data obtained spatially and the sign of this autocorrelation (positive or negative). Bivariate Global Moran's I statistic formula is given as *I* (9):

$$I = \frac{n}{S\_0} \frac{\sum\_{i=1}^{n} \sum\_{j=1}^{n} w\_{i,j} z\_i z\_j}{\sum\_{i=1}^{n} z\_i^2} \tag{10}$$

where *zi* is the deviation of an attribute for feature *i* from its mean *xi* − *X* , *wi*,*<sup>j</sup>* is the spatial weight between feature *i* and *j*, *n* is equal to the total number of features, and *S*<sup>0</sup> is the aggregate of all the spatial weights of (11):

$$S\_0 = \sum\_{i=1}^{n} \sum\_{j=1}^{n} w\_{i,j} \tag{11}$$

The *zI*-score for the statistic is computed, as in Reference (12):

$$z\_I = \frac{I - E[I]}{\sqrt{V[I]}} \tag{12}$$

where *E*[*I*] and *V*[*I*] can be calculated as follows:

$$E[I] = -1/(n-1)\tag{13}$$

$$V[I] = EI^2 - E[I]^2\tag{14}$$

Global spatial GIS autocorrelation will return three values: the Moran's I Index, z-score, and p-value. Given a series of spatial features and an associated attribute, bivariate Global Moran's I statistic indicates whether the pattern expressed is clustered, dispersed, or random and its degree of statistical correlation. When the z-score or p-value indicates statistical significance, a positive Moran's I index value indicates a tendency toward clustering, while a negative Moran's I index value indicates tendency toward dispersion. The z-score and p-value are measures of statistical significance which inform us whether or not to reject the null hypothesis. For this analysis, the null hypothesis states that the values associated with features do not have any statistical correlation.

From this information, we will be able to implement, in a geo-located way, the so-called hot and cold points in the mapping through the Local Indicators of Spatial Association (LISA) from Anselin [55]. Each Anselin Local Moran's I statistic of spatial association *I* is given as:

$$I\_i = \frac{\mathbf{x}\_i - \overline{X}}{S\_i^2} \sum\_{j=1, j=i}^n w\_{i,j} (\mathbf{x}\_j - \overline{X}) \tag{15}$$

where *xi* is an attribute for feature *i*, *X* is the mean of the corresponding attribute, *wi*,*<sup>j</sup>* is the spatial weight between feature *i* and *j*, and:

$$S\_i^2 = \frac{\sum\_{j=1, j=i}^n (\mathbf{x}\_j - \overline{X})^2}{n-1} \tag{16}$$

with n equating to the total number of features. The *zI*-score for the statistic is computed as:

$$z\_I = \frac{I - E[I]}{\sqrt{V[I\_i]}}\tag{17}$$

where *E*[*I*] and *V*[*I*] can be calculated as follows:

$$E[I] = -\frac{\sum\_{j=1, j=i}^{n} w\_{i,j}}{n-1} \tag{18}$$

$$V[I] = EI^2 - E[I\_i]^2\tag{19}$$

For this analysis, the null hypothesis states that the values correlation of two elements are randomly distributed. Thus, the higher (or lower) the z-score, the stronger the intensity of the clustering of these values. A z-score near zero indicates no apparent clustering within the study area. A positive z-score indicates clustering of high values. A negative z-score indicates clustering of low values. This numerical evaluation will be implemented through GIS mapping to distinguish configuration patterns of High-High clusters (high levels of impact associated with high levels of vulnerability), Low-Low clusters (low levels of impact associated with low levels of vulnerability), and spatial outliers, either High-Low (high levels of impact associated with low levels of vulnerability) or Low-High (low levels of impact associated with high levels of vulnerability).

Therefore, the bivariate statistical correlation analysis between the distributions of different GIS indicators will help us to understand, spatially, the extent to which the impacts produced by human action affect coastal vulnerability.

#### **3. Results**

The exposed methodology has been applied to the surface area detailed above with the following results. In the first place, the retro-historic GIS analysis of the impacts on the coastal edge has been carried out. Secondly, the coastal vulnerability of each area has been spatially evaluated. Finally, the spatial correlation of both phenomena has been evaluated by geo-statistical methods.

#### *3.1. GIS Retrohistoric Analysis of the Anthropization Impacts in the Coastal Perimeter*

A long-time spatial analysis of the seaside impacts in the area has been carried out through GIS retro-historic methods from the 1960's to the present. The results obtained in the analysis have been differentiated according to the structure detailed in the methodology section.

#### 3.1.1. Urbanization Impacts

The transformation density of the urbanization processes can be observed in a summarized way in Figure 6. To simplify the spatial representation of the GIS indicators for the transformation of the coastal perimeter as a result of urbanization, outputs have been tessellated using the ArcGIS Pro 10.5.0 program (ESRI Corporation, Redlands, CA, USA). The tessellated polygons have a size of 25 × 25 m to allow an understandable visualization of the results at a large scale (in case a tile has a surface area in two or more categories, it is assigned to the category with the most surface area present). At the intensity level, it is observed how the largest and fastest-growing population densities have been generated in the ancient dune cord called La Manga. At a quantitative level of surface transformation, it can be observed that the greatest results in absolute values are found in the urban sprawl from the settlements of the inner perimeter (San Javier and Los Alcazares coastal towns).

**Figure 6.** Mapping of *UTD*1973−<sup>2017</sup> <sup>1929</sup>−<sup>1973</sup> index for seaside impacts associated to urbanization. The accelerated evolution of urban sprawl in the population of Los Alcazares in 1956–1981–2017 is detailed in red on the left. The most significant cases of dune shrinkage phenomena on the beaches of the Mar Menor in La Manga are indicated in yellow on the right (below) and the detailed evolution of the dune profile of one of them from 1981 (marked with a red line) until 2017 (above).

It is interesting to observe in the area of La Manga how sometimes there is no correspondence between the initial natural surface and the actual urbanized surface at the two-dimensional level. The incidents detected correspond mainly to retraction phenomena on the beaches of the ancient dune cord of the Mar Menor. These beaches were formerly fed by the sand from the beaches of the Mediterranean thanks to the prevailing winds from the East. The current "screen effect" generated by the massive construction of buildings in the old dune belt has caused retraction effects in a generalized manner in the dune profile of the Mar Menor beaches, with alarming cases being observed in which the disappearance of the beaches reaches the very foundations of some houses.

It is interesting to observe how the phenomenon of dune shrinkage of the beaches occurs mainly in the lower half of the old dune cord. This question is explained by La Manga being chronologically urbanized progressively from the South to the North, from the mid-1960's to the late 1990's. As a consequence of this, the phenomena of dune shrinkage (which are deferred in time and often require an average of 10–15 years to significantly emerge) appear mainly in the buildings of the Southern half that were built in the decades of the 1970's and 1980's. In addition, this duality is accentuated by the fact that the so-called "screen effect" is more intense in the Southern half, given that the urbanization in that area is denser. However, this does not mean that the same situation will not occur in the northern half in the future, when the construction in that area (which is not yet fully built) has been consolidated and the dune shrinkage phenomena has taken enough time to bring out its first consequences.

#### 3.1.2. Ports and Coastal Infrastructure Impacts

The detected impacts derived from marinas and coastal infrastructures are widely distributed spatially (Figure 7). The distribution of the impacts has also been tessellated in meshes of 25 × 25 m to make the characterization of the phenomenon on a large scale more understandable.

Its impact on the coastal perimeter extends over the last decades and it is observed that it varies depending on two main parameters. On the one hand, it is necessary to take the geographical situation of each element into account. Variables such as the proximity of golas, tall buildings, or the mouths of the wadis can influence the global impact. Obviously, the behavior of a coastal infrastructure in the Mar Menor will be very different from that of one in the Mediterranean Sea. However, even the situation of the infrastructure within the Mar Menor shows influence in its effects on the coast.

On the other hand, the type of infrastructure built exerts the main influences. In the case of breakwater dikes, there is no great variability, with only their length being the differentiating variable. Nevertheless, in the case of ports, regardless of their size (those of greater size may have a greater impact), we find three distinct impact typologies. First, we have the traditional ports whose dock is on land reclaimed from the sea through sheltered dikes. Secondly, we find ports built with interior marinas inland, or naturally sheltered in bays. These ports are usually located in the urban plot of the coastal towns. As the third and last case, we have so-called "island ports". This mixed case includes the ports that are separated from the coastline and are linked to it through an element that allows the passage of water and sedimentary dynamics.

In the case of the dikes or breakwaters we find elements that do have an impact on the coastal dynamics, as well as elements that do not, the length of each element being a determining factor of the intensity from the impact. Among the elements that have an impact, it should be noted that there are positive, negative, and mixed consequences. Positive impacts are understood as those that contribute to stabilize the dune profile for the maintenance of the beach surface or those that correct the negative impacts of an area derived from the action of ports or dikes from neighboring areas. Negative impacts are understood to be those that negatively affect the dune stability of the area in which they are located, causing generalized phenomena of dune retraction. Finally, mixed elements are those that simultaneously generate results that can be cataloged as negative and as positive, such as increasing the beach surface on one side at the cost of reducing it on another, or maintaining the dune stability of one area, generating instability in another nearby area.

In the case of marinas, their size does seem to be directly correlated with their coastal impact. For example, the Tomas Maestre marina (numbered as 7 in Figure 4), the largest marina in Spain (and one of the largest in Europe with 1700 moorings) located in the North of La Manga and crossed by one of the golas, does not offer a remarkable behavior at the level of alteration of the dune profile. Nevertheless, we find common differentiated behaviors based mainly on the location and typology.

**Figure 7.** Tessellation of significant structural CIR impacts in the seaside edge associated to coastal alterations caused by marinas and coastal infrastructures between 1970 and 2017.

In the case of ports located on land reclaimed from the sea and harboring the port dock through breakwaters, we can find usual coastal behavior in these infrastructures in the Mediterranean. The beaches located to the North of the port increase their surface area while those located to the South experience important retraction phenomena over the years. In this section, the case of the San Pedro del Pinatar marina (numbered as 1 in Figure 7), located on the Mediterranean side, is especially interesting, and it has made almost 80% of the well-known La Llana Beach disappear within a period of 20 years (see the evolution in the upper part of Figure 7).

In the case of exempt ports called "island ports", the detected behavior is different. The area in front of the port experiences a growth of the dune surface according to the port-coast union axis. This growth has constituted, in the ports of this typology, what we would call a half-tombolo after 20–30 years of impact and it is foreseeable that it will end up eventually forming a complete tombolo (a bar of sand or shingle joining the port to the mainland). This dune growth also tends to be controversial, since when it is generated in the bottom of the sea currents, it tends to accumulate sludge on the beach that turns out to be very annoying for bathers. In addition, this growth takes place at the expense of the neighboring beaches North and South of the port, which tend to shrink and even in some cases disappear.

Finally, in the case of the marinas located in the interior of the land zone, the casuistic is more varied and depends more on the geographical position and the physical conditions that surround the port. In the case of the port of Cape of Palos (numbered as 6 in Figure 7) in the Southern Mediterranean end of La Manga, a pure inner land port, it can be seen in Figure 7 that there is no effect on the neighboring beaches (when, on the contrary, there are alterations on the beaches on the other side of the cape as a result of a dike construction, as can be seen in the Northern part of the photos). Nevertheless, we also find other cases such as the port of Lo Pagan (numbered as 4) and the Dos Mares yacht club (numbered as 5), rather half-inner land infrastructures, in which there are hard accumulations around the port. These growths are related to the boundary conditions of each port infrastructure and also lead to the accretion of sludge that damages neighboring beaches. The most significant set of impacts are detailed in Table 3.


**Table 3.** CIR and main impacts detected of the marinas on their neighboring beaches.

#### 3.1.3. Coastal Land Alterations Impacts

The analysis of the impacts derived from the direct alterations of the coastal space denotes a heterogeneous distribution associated to specific actions with diverse results. The area most subjected to this type of impact is mainly the old dune cord currently urbanized called La Manga (Figure 8). This area has various actions such as the direct alteration of its surface geometry, the artificial modification of two of the so-called "golas" (natural communication channels between the Mar Menor and the Mediterranean Sea, numbered as 1), the internal dredging of some areas such as the artificial clover called Veneziola (numbered as 3), or the landfilling of large marine areas such as the one called El Vivero (numbered as 2), or the road to connect the inland area with the Ciervo Island (numbered as 5).

**Figure 8.** Tessellation of the *DLT*2017−<sup>1974</sup> <sup>1974</sup>−<sup>1929</sup> mapping for significant impacts in the seaside edge associated to land alteration transformations between 1974 and 2017.

The behavior of the GIS evolution for all these elements is not homogeneous either. On the one hand, we find those whose new physical configuration has consolidated over time reaching a static equilibrium situation. This is the case of alterations in the layout of the dune cord consolidated by road infrastructures and buildings or the landfilling of El Vivero. On the other hand, we find cases in which there is a certain dynamic equilibrium (in which a static equilibrium position is not reached) or in which nature always directly opposes with forcefulness in the same sense as human action in a reiterative way. The first case can be found in the two so-called artificial "golas" or the dredging of Veneziola. In these cases, the artificial dredging made to reach a larger draft thus enabling large vessels to navigate it is not maintained at stable depths due to coastal dynamics, forcing periodical dredging. The other three natural golas also warrant special mention, whose surface and depth have moved in different directions over the last decades. However, in this case, a clear and direct cause cannot be established, since there are several nearby elements with a certain impact capacity such as sports ports, dredging, or alterations to the coastal surface. Finally, we found situations whose completely unbalanced configuration generated important alterations and whose impacts were not acceptable either environmentally or socially by the population, forcing the restoration of them to their original situation. One such case is of the road that connected the Ciervo island, whose obstruction of the sea currents generated such a "dam effect" to the North of the same that forced its dismantling.

For the mapping of the distribution of the DLT transformation indicator during the period 1967-2107, four different levels have been established to categorize the four most representative casuistic of the phenomenon. In the first place we have those cases in which the transformations are limited to the change of land use, with which the functions *G*<sup>i</sup> and *J*<sup>i</sup> barely have differences of less than 20%. This category fundamentally includes transformations from natural to agricultural land or significant changes in the type of agricultural use (urbanization transformations are assessed in the 3.1.1 indicator). Within this first case, the substitution of the old terraced crops by horticultural plantations of intensive agriculture stands out in the area. In the second category, we find those cases in which there has been an alteration of the land-sea configuration, but which has been consolidated in a static equilibrium over time without new alterations. This is the case of the mentioned coastal landfills and geometry changes consolidated by the road infrastructures in the old dune cord. In the third level we find those transformations of the land-sea geometry that have not reached a static equilibrium and are currently in what we have called a dynamic equilibrium. This is the case of artificially-dredged golas whose draft tends to be reduced by the effect of coastal dynamics, or of the golas that still remain natural, whose surface and draft has been changing in the last decades in different directions as a side effect of several nearby impacts. Finally, in the fourth category we find land-sea configurations that have undergone important transformations in different ways as a result of the generation of a dynamic equilibrium that is "not stable". This is the case of the execution of coastal infrastructures such as the road that connected the old dune cord with Ciervo Island, whose serious alteration of coastal dynamics forced the infrastructure to be dismantled.

In Figure 8, the distribution of the impact has been tessellated in meshes of 25 × 25 m to make the characterization of the phenomenon on a large scale more understandable. For this analysis, we must highlight that in the case of the golas, the z parameter is not obtained by geo-referencing. It has been obtained based on the abundant bathymetric documentation of the nautical charts of the area that exist for many dates. It should be noted that the level of detail of these charts for navigation is lower than in the case of the rest of the information; with the information basically being lines of homogeneous depth. However, this can be understood as an acceptable simplification of the model, since in artificial golas the depth is the result of dredging to facilitate navigation, so the bottom tends to be of a homogeneous depth. Likewise, in the so-called natural golas, since these are fairly shallow waters located between the two seas, the slope of the terrain is usually flat.

#### *3.2. GIS Analysis of Coastal Vulnerability*

A spatial analysis of the coastal vulnerability for land and marine flooding has been carried out based on the geo-referenced data available in the Ministry of Environment, Agriculture, and Water of Spain [56]. A return period of T = 100 years has been chosen for the forecast simulation, since it provides spatially significant results, whilst at the same time it results in an order of magnitude close to the total period evaluated, making the results more interesting in order to establish conclusions later.

#### 3.2.1. Coastal Vulnerability Associated to Risk of Marine Flooding

A forecast simulation of coastal vulnerability in the territory analyzed was carried out for a return period of T = 100 years according to the criteria set out in the methodology section. The envelope of worst case scenarios for maximum events indicates an impact mainly concentrated in the Mediterranean side of the old dune cord of La Manga (Figure 9). The inner coastal perimeter of the Mar Menor scarcely suffers from the flooding of the beaches, apart from occasionally some single-family homes located on the front line of the coast. In the case of La Manga, the most affected part surpassed by a possible flood is the area corresponding to the natural golas, located in the far North (which is logical). Regarding the urbanized areas, it is interesting to observe how the effects in the Northern area are much greater than in the Southern zone. The urban configuration of the Southern area is more resistant to the phenomena of flooding of marine origin due to its greater density and more compact structure.

The most vulnerable urban areas are found in the Northern section (especially in the points with the smallest width of the dune cord and with low density urban configurations or with partial building) and in areas with artificial "golas", which are completely flooded by water. Given that a homogeneous distribution of coastal vulnerability is not found, this will be discussed in the discussion section, based on the results of the geo-statistical analysis as to whether the action of man has contributed to worsen or improve the existing situation in this section.

**Figure 9.** Forecast modeling of the danger by marine flooding on a simplified MDT of the coastal strip for a return period T = 100 years. Several detailed examples of the impact in different areas are attached (data source for GIS analysis: see Reference [56]).
