The world has been facing an increase in various natural hazards. The coastal regions are recognized as one of the most vulnerable due to high population pressure and climate change intensity. Mediterranean countries are one of the most burnable ecosystems in the world, one of the most exposed to pluvial floods, and have the highest erosion rates within the EU. Therefore, the aim of this study was to develop the first multi-hazard susceptibility model in Croatia for Sali settlement (island of Dugi otok). The creation of a multihazard susceptibility model (MHSM) combined the application of geospatial technology (GST) with a local perception survey. The methodology consisted of two main steps: (1) creating individual hazard susceptibility models (soil erosion, wildfires, pluvial floods), and (2) overall hazard susceptibility modeling. Multicriterial GIS analyses and Analytical Hierarchy Process were used to create individual hazard models. Criteria used (32) to create models are derived from very-high-resolution (VHR) models. Two versions of MHSM are created: 1) all criteria with equal weighting coefficients and 2) weight coefficients determined based on a public perception survey. Both models had similar results and reveal moderate susceptibility of Sali to multiple hazards. The public perceives that the research area is the most susceptible to wildfires. The greatest difference between public perception and the GIS-MCDA model of hazard susceptibility is related to soil erosion. However, the accuracy of the soil erosion model was confirmed by ROC curves based on recent traces of soil erosion in the research area. The proposed methodological framework of multihazard susceptibility modeling can be applied, with minor modifications, to other Mediterranean countries.
Research Article
Multi-hazard Susceptibility Model based on Very-High-Resolution data – a case study of Sali settlement (Dugi otok, Croatia)
https://doi.org/10.21203/rs.3.rs-2096960/v1
This work is licensed under a CC BY 4.0 License
Journal Publication
published 06 Nov, 2023
Read the published version in Environmental Science and Pollution Research →
You are reading this latest preprint version
Multihazard susceptibility
soil erosion
pluvial floods
wildfires
GIS-MCDA
public perception.
In recent decades the world has been facing an increase in various natural hazards which is likely related to climate change (Thomas and Lopez 2015; Wu et al. 2018). Natural hazards may result in life losses and can cause significant damage to the economy, urban infrastructure, environment, and any other aspects of human life (Papathoma-Kohle et al. 2011). Global temperatures rising is resulting in an increase of extreme weather events (high heats, rainfalls, storms), sea level rise, wild-fires ignition, changes in infectious agents, and other related events. In the context of disaster mitigation planning, the important aspect is hazard mapping which provides insight into the distribution of locations threatened by specific hazard (Nugraha et al. 2018). This has encouraged scientists from different fields to study specific hazards such as flash floods/pluvial floods (Azizi et al. 2021; Arabameri et al. 2020; Waqas et al. 2021), landslides (Sestras et al. 2019; Wang et al. 2009), debris flows (Jakob et al. 2012; Abuzied and Pradhan, 2021), forest/wildfires (Zhang et al. 2019, Abedi et al. 2021; Marić et al. 2021), glacier avalanches (Khadka et al. 2021; Gurung et al. 2021), earthquakes (Miao et al. 2022; Batar et al. 2021) droughts (Mokarram et al. 2021) soil erosion (Domazetović et al. 2019). All of the above-mentioned recent researches are focused on single hazards that are typical for the specific research area. Still, some locations are susceptible to multiple hazards which may occur simultaneously or as cascading events (Aksha et al. 2020, Ullah et al. 2022).
The multi-hazard concept was initially introduced through the United Nations Environment Program’s sustainable development policies with the aim to investigate and improve the planning and management of residential areas prone to natural hazards (UNEP 1992). Estimation of the multi-hazard susceptibility is the component part of disaster risk management (Ullah et al. 2022). The development of an effective multi-hazard risk mitigation plan implies assessing single hazard susceptibility and possibility as well as their interactions (Ullah et al. 2022; Pouyan 2021). Susceptibility to hazard is described as the tendency of a region to the occurrence of a specific hazard without considering the moment of the event or potential consequences such as economic losses or fatality (Pouyan 2021).
Previous research on multihazard combines different variations of hazards depending on the research area. Some of the case studies where multi-hazard evaluations have been performed include Iran (Pouyan 2021), Pakistan (Ullah et al. 2022), Nepal (Aksha et al. 2020), Turkey (Yanar et al. 2020), United Kingdom (Guerriero et al. 2022), Greece (Skilodimou 2019) the Adriatic Sea (Gallina et al. 2020) Serbia (Durlević et al. 2021), etc. Coastal regions are particularly prone to natural hazards (Kron 2013; Ballesteros et al. 2018) due to natural, physical, and socio-economic characteristics making them especially vulnerable (EEA 2013; Ballesteros et al. 2018). For example, the Mediterranean due to climate characteristics and geographic location is considered one of the most flammable ecosystems in the world (Pausas et al. 2016; Maric et al. 2021) and one of the most exposed to flash/pluvial floods (Gaume et al. 2016). In addition, the Mediterranean area has the highest total erosion rates within the EU (Panagos 2020), and the lowest levels of soil organic matter (Aguilera et al. 2013; Guittonny-Philippe et al. 2014) accompanied by increasing human pressures (Shukla et al. 2019; Ning et al. 2006) and climate change vulnerability (Montanarella, 2015; Panagos et al. 2015). Loss of organic matter through the erosion process represents a huge cost to the agricultural community and at the same time affects marine pollution (Borelli et al. 2017) by sedimentation of eroded material into the sea (Stipaničev et al. 2007; Pavlek et al. 2017).
As a Mediterranean country, the Republic of Croatia (HR) is recognized as a disaster-prone country. It has a high wildfires ignition risk (Stipaničev et al. 2007; Marić et al. 2021) averaging over 1000 registered wildfires annually. Coastal areas and the islands have an especially pronounced risk of fire ignition (Pavlek et al. 2017; Marić et al. 2021) and soil erosion (Domazetović et al. 2019). According to the values of the average rate of erosion soil loss, Croatia is among the moderately endangered countries of the EU (Ferreira et al. 2022). However, due to different micro-location characteristics, some parts of Croatia are significantly more vulnerable and endangered by the soil erosion process and have very high values of the average soil loss rate (Ferreira et al. 2022). The soil loss by erosion could be especially problematic on small islands with limited agricultural land area. Generating the soil-erosion model is an important part of conservation strategies and soil management (Panagos et al. 2015; Borrelli et al. 2017). Like other European countries, Croatia is also affected by intensive rainfall events when the amount of water exceeds the capacity of the drainage system and causes pluvial floods that result in material damage. Still, this type of hazard is less investigated compared with other types of natural hazards that occur in Croatia (Ciblić et al. 2019).
Therefore, the main objective of this manuscript is to create the first multihazard (soil erosion, wildfires, and pluvial floods) susceptibility model for the micro-location (Sali, Dugi otok) in the coastal part of Croatia. This research was funded and conducted under the Interreg - PEPSEA (Protecting the Enclosed Parts of the Sea in Adriatic from pollution) project. The study area (235 ha) includes the drainage basin of the Sali settlement. To investigate multihazard susceptibility this research combines a public perception survey and geospatial technology (GST).
The study area (235 ha) is the drainage basin of the Sali settlement situated on the Dugi Otok island (Croatia) (Fig. 1). The total population of Sali settlement in 2011 was 740 (Borrelli et al. 2021). The topography of Sali’s area is predominantly up to 200 m above sea level. This is an area of Mediterranean (Csa) climate according to Koppen's classification with maritime precipitation regime. The largest amounts of precipitation fall in the winter months, while the summer is characterized by high heat and droughts (Ciblić et al. 2019). The study area is classified as a moderately endangered area by soil erosion, where the average annual soil loss rate is 5–10 t 〖ha〗 ^ (− 1) y ^ (− 1) (Panagos et al. 2015). There are no permanent watercourses in the drainage basin Sali (HOK 1:5000), with exception of a small accumulation in the natural reservoir Saljsko polje. Therefore, floods in this area are possible exclusively because of intensive rainfalls (pluvial events) or sea levels rising (coastal floods). Pluvial flood has been recorded in 2017 when a large amount of rain fell in a short time and caused material and agricultural damage (DHMZ, 2017). The landscape of Sali is dominated by neglected agricultural areas and forest of Aleppo pine and degraded holm. These vegetation types are characterized by high flammability (Marić et al. 2021). The potential danger of wildfires is increased in the summer months due to higher population pressure (tourists) combined with droughts and high air temperature (Bakšić et al. 2015).
3.1. Public perception of multihazard susceptibility
The public survey examined the level of risk of 33 potential threats (including soil erosion, wildfires, and pluvial floods) to the natural environment. The level of risk is expressed by a linear Likert scale from 1 to 5 (1-very low, 2- low, 3- moderate, 4- high, 5-very high). The questionnaire was conducted among adult citizens (18+), in the period from 25 to 30 June 2020 and encompassed 5% of the Sali population (N = 38). The unit of analysis is household, and each questionnaire was filled at a different address inside the potential vulnerability zones. The selection of respondents was random, and the type of sample was stratified.
3.2. Multihazard susceptibility model
The multihazard susceptibility model (MHSM) is generated from three individual hazard susceptibility models which include soil erosion, wildfires, and pluvial floods. Therefore, the process of model generation included two main steps: (1) Creating the Individual hazard susceptibility models (soil erosion, wildfires, pluvial floods) (2) Overall hazard susceptibility. Both steps consist of several additional sub-steps within GIS-based multi-criteria decision analysis (GIS-MCDA) and AHP (Analytical Hierarchy Process).
The GIS-MCDA process for each model consists of several steps (Domazetović et al. 2019; Marić et al. 2021):
(1) identification of the problem and defining the goal;
(2) determination of criteria and constraints;
(3) standardization of criteria;
(4) determination of weight coefficients;
(5) criteria aggregation;
(6) model validation
The process was simplified using the GAMA method (Domazetović et al. 2019).
Criteria used to generate the individual hazard susceptibility were derived from very-high-resolution models (VHR) produced from imagery acquired by unmanned aerial vehicles (UAVs) DJI Matrice 600 Pro and 210 RTK V2. VHR models include the digital terrain model (DTM), multispectral imagery (5 bands), and orthomosaic. The imagery was acquired using aerial surveys recorded using the double grid missions, with front and side overlaps of 80% and flight height of around 200 meters. Morphometric and hydrological parameters used in the study were derived from created DTM with a spatial resolution of 0.5 meters. Anthropogenically, environmental and other parameters are based on the land cover model which is generated from multispectral imagery using the Geographic Object-Based Image Analysis (GEOBIA) and Support Vector Machine classification algorithm.
Standardization of predisposing criteria was performed using the Decision-maker standardization method, where predisposing criteria were standardized to scale from 1 to 5 (1- very low susceptibility, 5 - very high susceptibility).
Boolean criteria were standardized to binary scale (0-false and 1-true). Using the AHP, selected criteria were ranked according to their level of influence on the individual hazards. Aggregation of the criteria is performed using the formula (Eastman 1999).
P = ∑ 𝑤𝑖 𝑋𝑖 * ∏ 𝐶𝑗
where: P = suitability 𝐶𝑗 = restriction 𝑤𝑖 = weighted coefficient ∑ =sum of weighted criteria; ∏ = sum of restrictions (1 – suitable, 0 – unsuitable) 𝑋𝑖 = criteria value
Subsequently, the individual susceptibility models are combined to create MHSM. Using the AHP process two variants of the model were made: (1) each hazard had the same relative importance (equal weight coefficients) and (2) weighting coefficients were added regarding the local population's perception of the risk of a particular hazard. AHP matrix was evaluated using the λmax maximum eigenvalue of the matrix with the consistency index (CI) and the consistency ratio (CR).
The consistency ratio for each AHP table is calculated according to the following formula (Saaty 1985):
CR = CI / RI
where: CI = consistency index, RI = random index.
Consistency index (CI) is calculated according to the formula:
CI = (λmax – n)/n – 1 (2)
where: λmax = Principal Eigen Value, n = number of criteria, λmax is calculated according to the formula: λmax = Σ of the products between each element of the priority vector and column totals.
The total susceptibility of the research area to natural hazards was calculated using the Weighted Overlay Function in ArcMap. Three generated models (pluvial floods, soil erosion, wildfires) are input parameters to create the final model. For the first scenario of MHSM each model of susceptibility hazard had the same weight coefficients (0.33). The second variant of MHSM had different weight coefficients for susceptibility hazard models determined based on a public perception survey.
4.1. Soil erosion susceptibility
Three groups of criteria were used to create GIS-MCDA soil erosion susceptibility model:
(1) primary morphometric (slope (SLO), aspect (ASP), profile curvature (PROF), planar curvature (PLAN)
(2) secondary morphometric (topographic wetness index (TWI), stream power index (SPI), length slope factor (LSF), specific catchment area (WAT)
(3) environmental parameters (land cover (LC), Boolean criteria (BLN).
The slope is the most influential criterion in creating the soil erosion susceptibility model due to its effect on the denudation processes (Domazetović et al. 2019; Conforti et al. 2011; Andualem et al. 2020). Steep slopes cause higher denudation intensity and therefore are the most important causative topographical feature.
The second most influential criterion is aspect because it affects the soil moisture content (Ma et al. 2010) and therefore determines the development of the different vegetation types (Domazetović et al. 2019; Conforti et al. 2011). The third affecting criterion is planar curvature (PLAN) which is perpendicular to the maximum slope direction and affects the convergence or divergence of water flow across the surface. Profile curvature (PROF) is parallel to the direction of the maximum slope and is an important criterion due to its effect on downslope flow rate i.e. the acceleration and deceleration of surface runoff (Conforti et al. 2011).
The topographic wetness index (TWI) is the indicator of the soil water content distribution (Ghosh and Maiti 2021). TWI is often used to calculate to what extent the area is saturated with water and to estimate the ability of the area to infiltrate the water (Conforti et al. 2011). Water-saturated areas precondition the more intensive surface runoff and increase erosion susceptibility (Domazetović et al. 2019).
The stream power index (SPI) is an indicator of the energy potential of surface water flow and is based on the correlation between slope, flow path, and accumulation (Andualem et al. 2020). High SPI values indicate a greater potential for soil erosion.
Length-slope factor (LSF) is related to sediment transport capacity and is often used to estimate the intensity of the erosion process (Zhang et al. 2019). High LSF values mean a higher potential for soil erosion.
The erosive power of water flow is also affected by the size of the Catchment area (CA) (Domazetović et al. 2019). Therefore, the specific catchment area for the Sali settlement is calculated using the Hydrology toolset.
Land cover (LC) and constraining Boolean (BLN) criteria are used as an additional group of criteria. High soil erosion susceptibility values have land cover classes like bare soils and low vegetation. Boolean BLN criteria include all areas soil erosion is not possible like build-up areas and water surfaces. The AHP matrix and weighting coefficients for each parameter are presented in Table 1.
| SLO | ASP | PROF | PLAN | TWI | SPI | LSF | WAT | LC | wok | |
|---|---|---|---|---|---|---|---|---|---|---|
| SLO | 1 | 2 | 3 | 6 | 3 | 3 | 3 | 3 | 3 | 3.00 |
| ASP | 0.33 | 1 | 5 | 3 | 0.5 | 0.5 | 0.5 | 6 | 3 | 2.10 |
| PROF | 0.17 | 0.2 | 1 | 1 | 0.33 | 0.33 | 0.33 | 3 | 0.33 | 0.80 |
| PLAN | 0.17 | 0.2 | 1 | 1 | 0.33 | 0.33 | 0.33 | 3 | 0.33 | 0.80 |
| TWI | 0.33 | 2 | 2 | 3 | 1 | 1 | 1 | 5 | 2 | 1.92 |
| SPI | 0.33 | 2 | 3 | 3 | 1 | 1 | 1 | 5 | 2 | 2.04 |
| LSF | 0.33 | 2 | 3 | 3 | 1 | 1 | 1 | 5 | 2 | 2.04 |
| WAT | 0.11 | 0.17 | 0.33 | 0.33 | 0.17 | 0.17 | 0.17 | 1 | 0.17 | 0.31 |
| LC | 0.33 | 0.25 | 2 | 2 | 0.5 | 0.5 | 0.5 | 5 | 1 | 1.39 |
| Total | 3.1 | 9.82 | 20.33 | 22.33 | 7.83 | 7.83 | 7.83 | 36 | 13.83 | 14.38 |
| λmax | 9.003 | |||||||||
| CI | 0.003 | |||||||||
| CR | 0.002 |
4.2. Wildfire ignition susceptibility
According to the literature review (Hong et al. 2017; Pourghasemi 2016; Marić et al. 2021) to create GIS-MCDA wildfire susceptibility model six groups of criteria were used:
(1) primary morphometric (slope, elevation, aspect)
2) secondary morphometric (terrain ruggedness (TRI), topographic wetness index (TWI);
(3) anthropogenic (distance from a road, distance from houses);
(4) other parameters (NDVI, insolation, heat load index (HLI).
Slope influences the rate of fire spread and intensity, as well as provides frictional effects on the response team (Guettouche et al. 2011; Asori et al. 2020). Steeper slopes derive wildfire spread and intensity more than gentler slopes (Guettouche et al. 2011; Asori et al. 2020). In the research area, the southern slopes receive more sunlight. Elevation affects the volume of rainfall, air humidity, vegetation patterns, and exposure to wind (Tiwari et al. 2021, Marić et al. 2021). Aspect is an important topographic element that influences the spreading rate and intensity of wildfires by controlling the wind condition and moisture content of the air (Asori et al. 2020). The type of fuel (land cover) will determine its combustibility or flammability. Higher values of NDVI indicate the tendency of an area to accumulate water (Mattivi et al., 2019, Marić et al. 2021). Greater NDVI values mean a lower wildfire susceptibility. Mostly grassland ecosystems may be more susceptible to wildfires than wetlands ecosystems due to differences in the wetness index of the fuels (Asori et al. 2020). Most fires are caused by human-related causes. Therefore, a closer distance from roads and housing units is linked to higher a probability of fire ignition (Marić et al. 2021; Gigović et al. 2018). The HLI is a parameter that considers the steepness of the slope when calculating the amount of solar radiation received by the slope. Weighting coefficients are calculated and described in the paper Marić et al. 2021.
4.3. Pluvial floods susceptibility
Pluvial floods occur when the amount of rainfall exceeds the capacity of the soil to infiltrate the water. Therefore, impervious surfaces due to their incapacity for water infiltration are the most susceptible to pluvial floods. There are many criteria affecting susceptibility to PF, but their selection varies from one study area to another. To create the Pluvial Floods Susceptibility Model (PFSM) the most often used criteria are related to the ground morphology/topographical features, environmental and hydrology features (Di Salvo et al. 2018; Arabameri et al. 2020). Some authors to generate a susceptibility model are considering the spatial density of previously observed floods (Di Salvo et al. 2018; Haque et al. 2021; Pham et al. 2020). However, the Settlement of Sali, is data poor area for which the official database of previous floods is not existing.
Therefore, three groups of criteria were used:
(1) topographical (elevation (ELV), slope (SLO), distance to sinks (DS), planar curvature (PLAN), topographic wetness index (TWI), terrain ruggedness index (TRI), stream power index (SPI)
(2) hydrological (drainage density (DD), distance to potential streams (STD) and
(3) environmental (land cover (LC), distance to roads (RD) and NDVI). The weighting coefficients for each parameter are presented in Table 2.
Land cover (LC) is the primary criterium in determining the susceptibility to floods (Vojtek and Vojtekova 2019). Different types of land covers have different infiltration capacities. Impervious (buildings, roads) and bare surfaces increase the surface water flow, whereas green areas reduce surface flow due to higher infiltration capacity (Choudhury et al. 2022). Urbanization, agriculture, deforestation, and other anthropogenic processes directly increase the susceptibility to pluvial floods.
Distance to potential streams (STD) was derived using the Hydrology tools and two additional tools: Stream to feature and Euclidean distance. By increasing the distance from the potential stream, susceptibility to pluvial flood hazard decreases (Choudhury et al. 2022).
The slope (SLO) is an important criterium because it prompts the velocity of the water flow and thus increases the possibility of the occurrence of the hazard’s negative effects (Haque et al. 2021). Steeper slopes are susceptible to torrential flows, while flat areas are more susceptible to water accumulation.
Planar curvature (PLAN) is directly affecting the convergence and divergence of the surface water flow in the drainage basins (Arabameri et al. 2020). Laterally convex surfaces have positive values and are classified as very low susceptibility to pluvial flood while laterally convex surfaces with negative values are classified as very high susceptibility.
Anthropogenic spatial elements like roads are important in PF susceptibility modeling because of the impermeable materials from which they are composed. When a large amount of rain falls, roads have a similar function to riverbeds, i.e. direct and channel the water flow. Therefore, by increasing the distance to roads (RD) the susceptibility to PF is decreasing (Janizadeh et al. 2021). RD is calculated in ArcMap 10.3.1. using the Euclidean distance.
Morphological characteristics of sinks affect the flood depth. The presence of sinks indicates a greater potential danger to people, vehicles, and goods (Di Salvo et al. 2018). By increasing the distance from sinks (DS), susceptibility to danger decreases. The location of sinks is derived using the Flow direction and Sinks tool in Hydrology toolbox.
Drainage density (DD) is an indicator of water flow accumulation, and it provides spatial information about water circulation i.e. helps identify the areas that are more likely to get flooded (Elkhrachy 2015). It is calculated using the Line density tool which calculates a magnitude-per-unit area from line features in the defined radius.
TWI enables the identification of areas saturated with water (Conforti et al. 2011). Water-saturated areas are incapable of further water infiltration and are conditioning the creation of surface runoff. Therefore, high values of TWI indicate higher susceptibility to PF.
The terrain ruggedness index (TRI) indicates the potential stream flow velocities and helps to identify the low-infiltration areas. High values of TWI are related to higher susceptibility to pluvial floods.
SPI is the measure of the protentional surface water flow erosion power (Arabameri et al. 2020). SPI was calculated based on slope and contributing area using the Hydrology toolset in Arc Map and using the Raster calculator. High SPI values indicate a higher susceptibility to pluvial floods.
The elevation is one of the most often used parameters which is considered to affect floods in an inverse way. Generally, lower elevations are more prone to floods i.e., flood frequency increases with elevation decreasing (Gaume et al. 2016; Arabameri et al. 2020; Elkhrachy 2015). In the research, area elevation is the least influential factor due to the small research area.
| LC | STD | SLO | PLAN | RD | SD | DD | TWI | TRI | SPI | NDVI | ELV | wok | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| LC | 1 | 2 | 2 | 3 | 3 | 3 | 4 | 8 | 8 | 8 | 8 | 9 | 0.247 |
| STD | 0.5 | 1 | 1 | 2 | 2 | 2 | 3 | 5 | 5 | 5 | 5 | 7 | 0.153 |
| SLO | 0.5 | 1 | 1 | 2 | 2 | 2 | 3 | 3 | 3 | 4 | 4 | 7 | 0.137 |
| PLAN | 0.33 | 0.5 | 0.5 | 1 | 1 | 1 | 2 | 3 | 2 | 3 | 4 | 5 | 0.088 |
| RD | 0.33 | 0.5 | 0.5 | 1 | 1 | 1 | 2 | 3 | 3 | 3 | 3 | 5 | 0.088 |
| SD | 0.33 | 0.5 | 0.5 | 1 | 1 | 1 | 2 | 3 | 3 | 3 | 3 | 5 | 0.088 |
| DD | 0.25 | 0.33 | 0.5 | 0.5 | 0.5 | 0.5 | 1 | 2 | 2 | 2 | 2 | 3 | 0.056 |
| TWI | 0.13 | 0.2 | 0.33 | 0.33 | 0.33 | 0.33 | 0.5 | 1 | 1 | 1 | 1 | 2 | 0.032 |
| TRI | 0.13 | 0.2 | 0.33 | 0.33 | 0.33 | 0.33 | 0.5 | 1 | 1 | 1 | 1 | 2 | 0.032 |
| SPI | 0.13 | 0.2 | 0.25 | 0.25 | 0.33 | 0.33 | 0.5 | 1 | 1 | 1 | 1 | 2 | 0.030 |
| NDVI | 0.13 | 0.2 | 0.25 | 0.25 | 0.33 | 0.33 | 0.5 | 1 | 1 | 1 | 1 | 2 | 0.030 |
| ELV | 0.11 | 0.14 | 0.14 | 0.2 | 0.2 | 0.2 | 0.33 | 0.5 | 0.5 | 0.5 | 0.5 | 1 | 0.018 |
| Total | 3.86 | 6.78 | 7.31 | 11.87 | 12.03 | 12.03 | 19.33 | 31.5 | 30.5 | 32.5 | 33.5 | 50 | |
| λmax | 12.1 | ||||||||||||
| CI | 0.1 | ||||||||||||
| CR | 0.065 |
5.1. Public perception survey
The sample population (N = 85) comprised different age groups with an average age of 41.35 years. respondents. Most (98%) respondents have the minimum primary level of education and 41% are academically educated. The 81% of respondents are employed. The local population estimated that of the three investigated threats, the occurrence of wildfires poses the greatest risk.
The wildfire ignition risk is ranked as moderate (3.00) with the response’s standard deviation of 1.16. Pluvial flood risk is ranked as low (2.78), while the standard deviation is 1.15. The risk of soil erosion is perceived to be the lowest (2.24) with a standard deviation of 0.91. The percentage of risk perception of each hazard (1 – very low, 5-very high) is presented in Table 3.
Table 3. Public perception of each hazard – the percentage of each class of susceptibility
5.2. Soil erosion susceptibility
It is established that high and very high SES is present in the coastal area and on the southern slopes of Sali bay, as well as on steeper slopes of the main island ridge. Contrary to the public perception, created GIS-MCDA model has shown that within the wider area of the Sali settlements are prevailing zones of high (4) and very high (5) SES, which cover 55.44% of the study area (Fig. 3.). Zones of low (1) and very low (2) SES encompass 13.53% of the study area. The opposite results of the conducted analysis can be explained by the construction of dry-stone walls therefore residents considered arable land is protected from erosion. Also, the low perception could be the consequence of the decreasing use of agricultural land and depopulation, which are typical for Croatian islands (Faričić et al. 2010). The accuracy of the model was calculated using the ROC curves based on recent traces of soil erosion (Fig. 4). AUC value is 0.873 which confirmed the accuracy of the model. All detected field-determined traces of recent soil erosion are located within the zones of high and very high SES in the created GIS-MCDA model.
5.3. Wild-fires susceptibility
The wider area of Sali is arguably prone to wildfires due to the highly flammable vegetation cover combined with other predisposing factors. WFSM reveals that nearly 60% the of settlement area is high or very highly susceptible to fire ignition (Fig. 5). Still, the local population perceives the threat of wildfire ignition as moderate (x ̅=3.00) in their natural environment. Perception of wildfires as a threat is the highest compared to other threats. This perception is possibly influenced by the recent wildfires in the southern part of the island Dugi otok which was several times threatened by fire (Fig. 6). The Sali wider area has many unmaintained agricultural areas, which are highly susceptible to fire ignition. Although fire experts often warn about the importance of cleaning and maintaining such neglected areas, the population is often ignorant. Perception of risk is crucial in motivating individuals to act for risk mitigation (Šiljeg et al. 2022).
5.4. Pluvial floods susceptibility
The PFSM indicates that the 47.08% of Sali settlement is very low (1) or very low susceptible (2) to pluvial floods which is not surprising considering that it is dalmatian Island with low population density and therefore the low percentage of impervious surfaces i.e. build-up areas. Moderately susceptible is 40.11% of the settlement (Fig. 6). This zones predominantly encompasses the botanical reserve Saljsko polje where dominated vegetation types are old Oliva groves, vineyards, and neglected agricultural areas. On the Saljsko polje, a small puddle is formed due to its topographical and lithology characteristics, and it represents the only bigger accumulation inside Sali drainage basin. High and very high susceptible areas make up 12.81% of the total area and are mainly located in the coastal zone. However, concerning is the fact that in the coastal part, especially Sali’s bay many residential objects, restaurants, bakeries, the local market, and touristic objects are located. Highly susceptible are also some parts of the roads, and fish processing plant “Marušić” located at norther-west coast due to high percentage of impervious surface and specific terrain configuration. Also, the area of the quarry is highly susceptible to pluvial floods due to bare land and low infiltration characteristics (Fig. 7).
5.5. Multihazard susceptibility
MHSM 1 is classified using the equal interval method and results showed that the largest area (65.18%) of Sali drainage basin is moderate (3) susceptible to multiple hazards (Fig. 8). Very low or low susceptibility zones are represented in 14.59% of the basin area. High susceptibility (4) zones are widespread through almost the whole drainage basin, but their total area is approximately only 27%. Very high susceptibility areas are represented to a significantly lesser extent (0.5%). This is not surprising considering the used methodology; similar results are presented in other research papers (Durlević et al. 2021; Aksha et al. 2020). Namely, hazards can have specific inherent causative and different external triggers, which consequently could result in a superposition hazard model. Therefore, multi-hazard models should be analyzed with extreme caution.
In MHSM 2 weighting coefficients were applied according to public perception of hazard susceptibility (Fig. 9). The highest value is given to wildfires (0.374) while the lowest one is to soil erosion (0.279). However, the results are similar to the MHSM 1 (Fig. 11). Most of the basin area is moderately prone to multiple hazards i.e., moderate (3) susceptibility is still the most (58.19%) widespread class within the basin (Fig. 10). Zones of high and very high susceptibility are as in model 1, located near the roads and houses but are not including them. Very high (5) and high (4) classes include 27% of the total area. The most hazard-prone areas are located in the northern part of the basin and near the coast.
However, by changing the classification method and using Quantile for classification, it is easier to observe the potential endangerment of urban areas within the drainage basin (Fig. 10). For example, the most endangered parts of urban areas are in the central part of settlement Sali and are located near the Saljsko polje where dry vegetation, neglected and often noncultivated (bare soil) agricultural areas prevail. Furthermore, geomorphologically this is a natural basin where water accumulates from higher grounds. Therefore, this area is arguably susceptible to each analyzed hazard. Residential and other buildings are not prone to multiple hazards, which is expected because SES and WFSM are excluding the impervious surfaces and because the local population perceived the PF susceptibility as low.
GIS-MCDA and public perception analysis have shown that the wider area of Sali settlement is the most susceptible to wildfires. According to the WFS model, the wider area of Sali settlement is predominately high or very high (60%) susceptible. The residents perceive wildfires as a moderate threat to their environment. The perception is influenced by the recent fires in Dugi otok. The way how population perceives natural threats is extremely important because it affects mitigation behavior. The more aware the population is, the easier it will be to take preventive measures.
According to the GIS-MCDA wider area of Sali is also prone to soil erosion. SES model revealed that nearly 60% of the area is high or very highly susceptible to erosion. Contrarily, analysis of public perception showed that respondents perceived soil erosion as low and very low threat. A possible explanation is that nowadays the local population does not perceive soil erosion as a significant threat because of decreasing use of agricultural land. Also, in the past, residents have constructed dry-stone walls which they have considered to have protected their lands from erosion.
PFSM revealed that the researched area is low susceptible to pluvial floods compared to wildfires and soil erosion. The reason is a low density of impervious surfaces and terrain configuration. Still, if this threat occurs, the most exposed are residential objects, roads, and other objects. This GIS-MCDA model is the most consistent with the population's perception, i.e, in both cases the most represented classes are low and very low susceptibility.
Overall multiple hazard susceptibility is moderated according to both versions of MHSM. The multi-hazard susceptibility generally could be challenging not only to model but also to analyze and interpret due to differences among hazard characteristics. Each hazard could have specific inherent causative and different external triggers, which consequences could result in a superposition hazard model. Therefore, multi-hazard models should be analyzed with extreme caution. With some minor modifications, this framework of multihazard modeling can be applied to other Mediterranean countries. To get better results, in the future we suggest using DSM (digital surface model) acquired with aerialLIDAR, rather than DTM for SES modeling, which would include dry stone walls in the analysis.
Funding
This research was funded and conducted under the Interreg - PEPSEA (Protecting the Enclosed Parts of the Sea in Adriatic from pollution) project.
Competing interests
The authors have no relevant financial or non-financial interests to disclose.
Authors Contributions
All authors contributed to the study’s conception and design. Material preparation, data collection, and analysis were performed by Ante Šiljeg, Silvija Šiljeg, Rina Milošević, Ivan Marić, Fran Domazetović, and Lovre Panđa. The first draft of the manuscript was written by Rina Milošević and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Ethical approval
This article does not contain any studies with human participants or animals performed by any of the authors.
Consent to participate
Informed consent about anonymous participation in survey research was obtained from all participants included in the study.
Consent to publish
The participants have consented to the submission of the manuscript to the journal.
Availability of data and materials
Raw data were generated at the University of Zadar (Croatia). Derived data supporting the findings of this study are available from the corresponding author Rina Milošević on request.- Abedi Gheshlaghi H, Feizizadeh B, Blaschke T, Lakes T, Tajbar S (2021) Forest fire susceptibility modeling using hybrid approaches. Trans GIS 25(1):311–333
- Abuzied SM, Pradhan B (2021) Hydro-geomorphic assessment of erosion intensity and sediment yield initiated debris-flow hazards at Wadi Dahab Watershed, Egypt. Georisk: Assessment and Management of Risk for Engineered Systems and Geohazards, 15(3), 221–246
- Aguilera E, Lassaletta L, Sanz-Cobena A, Garnier J, Vallejo A (2013) The potential of organic fertilizers and water management to reduce N2O emissions in Mediterranean climate cropping systems. A review. Agric Ecosyst Environ 164:32–52
- Aksha SK, Resler LM, Juran L, Carstensen LW Jr (2020) A geospatial analysis of multi-hazard risk in Dharan, Nepal. Geomatics. Nat Hazards Risk 11(1):88–111
- Andualem TG, Hagos YG, Kefale A, Zelalem B (2020) Soil erosion-prone area identification using multi-criteria decision analysis in Ethiopian highlands. Model Earth Syst Environ 6(3):1407–1418
- Arabameri A, Saha S, Chen W, Roy J, Pradhan B, Bui DT (2020) Flash flood susceptibility modelling using functional tree and hybrid ensemble techniques. J Hydrol 125007. doi:10.1016/j.jhydrol.2020.12500
- Asori M, Emmanuel D, Dumedah G(2020) Wildfire hazard and risk modelling in the northern regions of Ghana using GIS-based multi-criteria decision making analysis.Journal of Environment and Earth Science, 10(11)
- Azizi K, Meier CI, Ullah K, Wang Y, Fang Z, Wang L, Rahman M (2021) (2022). Multi-hazard susceptibility mapping based on Convolutional Neural Networks. Geoscience Frontiers, 13(5), 101425
- Bakšić N, Vučetić M, Španjol Ž(2015) Potencijalna opasnost od požara otvorenog prostora u Republici Hrvatskoj. Vatrogastvo i upravljanje požarima, (2.),30–40
- Ballesteros C, Jiménez JA, Viavattene C (2018) A multi-component flood risk assessment in the Maresme coast (NW Mediterranean). Nat Hazards 90(1):265–292
- Batar AK, Watanabe T (2021) Landslide susceptibility mapping and assessment using geospatial platforms and weights of evidence (WoE) method in the Indian Himalayan region: Recent developments, gaps, and future directions. ISPRS Int J Geo-Information 10(3):114
- Borrelli P, Alewell C, Alvarez P, Anache JAA, Baartman J, Ballabio C, Panagos P (2021) Soil erosion modelling: A global review and statistical analysis. Sci Total Environ 780:146494
- Borrelli P, Robinson DA, Fleischer LR, Lugato E, Ballabio C, Alewell C, Panagos P (2017) An assessment of the global impact of 21st century land use change on soil erosion. Nat Commun 8(1):1–13
- Choudhury KN, Yabar H, Mizunoya T (2022) GIS and remote sensing-based spatiotemporal analysis of cumulative flood risk over Bangladesh’s national highways. Asia-Pacific J Reg Sci 6(1):335–364
- Cibilić A, Barbalić D, Rubinić J, Karleuša B, Krvavica N (2019) Upravljanje rizicima od poplava uslijed jakih oborina-projekt rainman
- Conforti M, Aucelli PP, Robustelli G, Scarciglia F (2011) Geomorphology and GIS analysis for mapping gully erosion susceptibility in the Turbolo stream catchment (Northern Calabria, Italy). Nat Hazards 56(3):881–898
- Di Salvo C, Pennica F, Ciotoli G, Cavinato GP (2018) A GIS-based procedure for preliminary mapping of pluvial flood risk at metropolitan scale, vol 107. Environmental modelling & software, pp 64–84
- Domazetović F, Šiljeg A, Lončar N, Marić I (2019) Development of automated multicriteria GIS analysis of gully erosion susceptibility. Appl Geogr 112:102083
- Državni hidrometeorološki zavod (2017) https://meteo.hr/index.php/
- Durlević U, Novković I, Lukić T, Valjarević A, Samardžić I, Krstić F(2021) … urić, V. Multihazard susceptibility assessment: A case study–Municipality of Štrpce (Southern Serbia). Open Geosciences, 13(1), 1414–1431
- Eastman JR (1999) Multi-criteria evaluation and GIS. Geographical Inform Syst 1(1):493–502
- Elkhrachy I (2015) Flash flood hazard mapping using satellite images and GIS tools: a case study of Najran City, Kingdom of Saudi Arabia (KSA). Egypt J Remote Sens Space Sci 18(2):261–278
- European Environment Agency (EEA) (2013) Balancing the future of Europe’s coasts-knowledge base for integrated management. Report No 12. Copenhagen. doi:10.2800/99116
- Faričić J, Graovac V, Čuka A (2010) Mali hrvatski otoci–radno-rezidencijalni prostor i/ili prostor odmora i rekreacije. Geoadria 15(1):145–185
- Ferreira CS, Seifollahi-Aghmiuni S, Destouni G, Ghajarnia N, Kalantari Z (2022) Soil degradation in the European Mediterranean region: Processes, status and consequences. Sci Total Environ 805:150106
- Gallina V, Torresan S, Zabeo A, Critto A, Glade T, Marcomini A (2020) A multi-risk methodology for the assessment of climate change impacts in coastal zones. Sustainability 12(9):3697
- Gaume E, Borga M, Llassat MC, Maouche S, Lang M, Diakakis M(2016) Mediterranean extreme floods and flash floods.
- Ghosh A, Maiti R (2021) Soil erosion susceptibility assessment using logistic regression, decision tree and random forest: study on the Mayurakshi river basin of Eastern India. Environ Earth Sci 80(8):1–16
- Gigović L, Jakovljević G, Sekulović D, Regodić M (2018) GIS multi-criteria analysis for identifying and mapping forest fire hazard: Nevesinje, Bosnia and Herzegovina. Tehnički Vjesn 25(3):891–897
- Guerriero L, Di Napoli M, Novellino A, Di Martire D, Rispoli C, Lee K, Calcaterra D (2022) Multi-hazard susceptibility assessment using analytic hierarchy process: the Derwent Valley Mills UNESCO World Heritage Site case study (United Kingdom). J Cult Herit 55:339–345
- Guettouche MS, Derias A, Boutiba M, Guendouz M, Boudella A (2011) A fire risk modelling and spatialization by GIS. J Geographic Inform Syst 3(03):254
- Guittonny-Philippe A, Masotti V, Höhener P, Boudenne JL, Viglione J, Laffont-Schwob I (2014) Constructed wetlands to reduce metal pollution from industrial catchments in aquatic Mediterranean ecosystems: A review to overcome obstacles and suggest potential solutions. Environ Int 64:1–16
- Gurung N, Thakuri S, Chauhan R, Ghimire NP, Ghimire M (2021) Dynamics of Lower-Barun Glacier and Glacial Lake and its GLOF Susceptibility Using Geospatial Analysis and Modelling. Jalawaayu 1(2):57–77
- Haque M, Siddika S, Sresto MA, Saroar M, Shabab KR (2021) Geo-spatial analysis for flash flood susceptibility mapping in the North-East Haor. Wetland) Region
- Hong H, Naghibi SA, Moradi Dashtpagerdi M, Pourghasemi HR, Chen W (2017) A comparative assessment between linear and quadratic discriminant analyses (LDA-QDA) with frequency ratio and weights-of-evidence models for forest fire susceptibility mapping in China. Arab J Geosci 10(7):1–14
- Jakob M, Stein D, Ulmi M (2012) Vulnerability of buildings to debris flow impact. Nat Hazards 60(2):241–261
- Janizadeh S, Pal C, Saha S, Chowdhuri A, Ahmadi I, Mirzaei K, Tiefenbacher S, J. P (2021) Mapping the spatial and temporal variability of flood hazard affected by climate and land-use changes in the future. J Environ Manage 298:113551. doi:10.1016/j.jenvman.2021.113551
- Kaur H, Gupta S, Parkash S, Thapa R (2018) Application of geospatial technologies for multi-hazard mapping and characterization of associated risk at local scale. Ann GIS 24(1):33–46
- Khadka N, Chen X, Nie Y, Thakuri S, Zheng G, Zhang G (2021) Evaluation of Glacial Lake Outburst Flood susceptibility using multi-criteria assessment framework in Mahalangur Himalaya. Front Earth Sci 8:601288
- Kron W (2013) Coasts: the high-risk areas of the world. Nat Hazards 66:1363–1382. doi:10.1007/s11069-012-0215-4
- Ma J, Lin G, Chen J, Yang L(2010), June An improved topographic wetness index considering topographic position. In 2010 18th International Conference on Geoinformatics (pp. 1–4). IEEE
- Maric I, Siljeg A, Domazetovic F(2021) Derivation of Wildfire Ignition Index using GIS-MCDA from High-Resolution UAV Imagery Data and Perception Analysis in Settlement Sali, Dugi Otok Island (Croatia). In GISTAM (pp.90–97)
- Miao Z, Peng R, Wang W, Li Q, Chen S, Zhang A, Hu C (2022) Integrating Data Modality and Statistical Learning Methods for Earthquake-Induced Landslide Susceptibility Mapping. Appl Sci 12(3):1760
- Mokarram M, Pourghasemi HR, Hu M, Zhang H (2021) Determining and forecasting drought susceptibility in southwestern Iran using multi-criteria decision-making (MCDM) coupled with CA-Markov model. Sci Total Environ 781:146703
- Montanarella L (2015) Agricultural policy: Govern our soils. Nature 528(7580):32–33 (9)
- Ning SK, Chang NB, Jeng KY, Tseng YH (2006) Soil erosion and non-point source pollution impacts assessment with the aid of multi-temporal remote sensing images. J Environ Manage 79(1):88–101
- Nseka D, Kakembo V, Bamutaze Y, Mugagga F (2019) Analysis of topographic parameters underpinning landslide occurrence in Kigezi highlands of southwestern Uganda. Nat Hazards 99(2):973–989
- Nugraha AL, Awaluddin M, Sasmito B(2018), February Modelling multi hazard mapping in Semarang city using GIS-fuzzy method. In IOP Conference Series: Earth and Environmental Science (Vol. 123, No. 1, p. 012002). IOP Publishing
- Panagos P, Ballabio C, Poesen J, Lugato E, Scarpa S, Montanarella L, Borrelli P (2020) A soil erosion indicator for supporting agricultural, environmental and climate policies in the European Union. Remote Sens 12(9):1365
- Panagos P, Borrelli P, Poesen J, Ballabio C, Lugato E, Meusburger K, Alewell C (2015b) The new assessment of soil loss by water erosion in Europe, vol 54. Environmental science & policy, pp 438–447
- Papathoma-Köhle M, Kappes M, Keiler M, Glade T (2011) Physical vulnerability assessment for alpine hazards: state of the art and future needs. Nat Hazards 58(2):645–680
- Pausas JG, Alessio GA, Moreira B, SegarraMoragues JG (2016) Secondary compounds enhance flammability in a Mediterranean plant. Oecologia 180(1):103–110
- Pavlek K, Bišćević F, Furčić P, Grđan A, Gugić V, Malešić N, Moharić P, Vragović V, Fuerst-Bjeliš B, Cvitanović M (2017) Spatial patterns and drivers of fire occurrence in a Mediterranean environment: a case study of southern Croatia. Geografisk Tidsskrift-Danish Journal of Geography 117(1):22–35
- Pham BT, Avand M, Janizadeh S, Phong TV, Al-Ansari N, Ho LS, Prakash I (2020) GIS based hybrid computational approaches for flash flood susceptibility assessment. Water 12(3):683
- Pourghasemi HR (2015) GIS-based forest fire susceptibility mapping in Iran: a comparison between evidential belief function and binary logistic regression models. Scand J For Res 31(1):80–98. doi:10.1080/02827581.2015.1052750
- Pouyan S, Pourghasemi HR, Bordbar M, Rahmanian S, Clague JJ (2021) A multi-hazard map-based flooding, gully erosion, forest fires, and earthquakes in Iran. Sci Rep 11(1):1–19
- Saaty TL(1985) Decision making for leaders.IEEE transactions on systems, man, and cybernetics, (3),450–452
- Sestraș P, Bilașco Ș, Roșca S, Naș S, Bondrea MV, Gâlgău R, Cîmpeanu SM (2019) Landslides Susceptibility Assessment Based on GIS Statistical Bivariate Analysis in the Hills Surrounding a Metropolitan Area. Sustainability 11(5):1362
- Shukla PR, Skeg J, Buendia EC, Masson-Delmotte V, Pörtner HO, Roberts DC, … Malley J(2019) Climate Change and Land: an IPCC special report on climate change, desertification, land degradation, sustainable land management, food security, and greenhouse gas fluxes in terrestrial ecosystems
- Šiljeg S, Milošević R, Panđa L, PUBLIC PERCEPTION OF THE URBAN PLUVIAL FLOODS RISK—CASE STUDY OF POREČ (CROATIA) (2022) J Geographical Inst “Jovan Cvijić” SASA 72(2):147–158
- Skilodimou HD, Bathrellos GD, Chousianitis K, Youssef AM, Pradhan B (2019) Multi-hazard assessment modeling via multi-criteria analysis and GIS: a case study. Environ Earth Sci 78(2):1–21
- Stipaničev D, Hrastnik B, Vujčić R(2007), May Holistic Approach to Forest Fire Protection in Split and Dalmatia County of Croatia, Wildfire 2007 Int. In Conference, Sevilla, Spain
- Thomas V, López R(2015) Global increase in climate-related disasters. Asian Development Bank Economics Working Paper Series, (466)
- Tiwari A, Shoab M, Dixit A (2021) GIS-based forest fire susceptibility modeling in Pauri Garhwal, India: a comparative assessment of frequency ratio, analytic hierarchy process and fuzzy modeling techniques. Nat Hazards 105(2):1189–1230
- UNEP (1992) Agenda 21. Technical Report, United Nations Conference on Environment and Development. https://www.un.org/en/conferences/environment/rio
- Van Westen CJ, Montoya L, Boerboom L, Badilla Coto E(2002) Multi-hazard risk assessment using GIS in urban areas: A case study for the city of Turrialba, Costa Rica. In Proceedings of the regional workshop on best practice in disaster mitigation, Bali 120–136
- Vojtek M, Vojteková J (2019) Flood susceptibility mapping on a national scale in Slovakia using the analytical hierarchy process. Water 11(2):364
- Wachinger G, Renn O, Begg C, Kuhlicke C (2013) The risk perception paradox—implications for governance and communication of natural hazards. Risk Anal 33(6):1049–1065
- Wang WD, Xie CM, Du XG (2009) Landslides susceptibility mapping based on geographical information system, GuiZhou, south-west China. Environ Geol 58(1):33–43
- Waqas H, Lu L, Tariq A, Li Q, Baqa MF, Xing J, Sajjad A (2021) Flash flood susceptibility assessment and zonation using an integrating analytic hierarchy process and frequency ratio model for the Chitral District, Khyber Pakhtunkhwa. Pakistan Water 13(12):1650
- Wollschlaeger S, Sadhu A, Ebrahimi G, Woo A (2022) Investigation of climate change impacts on long-term care facility occupants. City and Environment Interactions 13:100077
- Wu J, Han G, Zhou H, Li N (2018) Economic development and declining vulnerability to climate-related disasters in China. Environ Res Lett 13(3):034013
- Yanar T, Kocaman S, Gokceoglu C (2020) Use of Mamdani fuzzy algorithm for multi-hazard susceptibility assessment in a developing urban settlement (Mamak, Ankara, Turkey). ISPRS Int J Geo Inf 9:114
- Zhang G, Wang M, Liu K (2019) Forest fire susceptibility modeling using a convolutional neural network for Yunnan province of China. Int J Disaster Risk Sci 10(3):386–403
- Zhang S, Zhang X, Pei X, Wang S, Huang R, Xu Q, Wang Z (2019) Model test study on the hydrological mechanisms and early warning thresholds for loess fill slope failure induced by rainfall. Eng Geol 258:105135