Derivation of Wildfire Ignition Index using GIS-MCDA from

High-Resolution UAV Imagery Data and Perception Analysis in

Settlement Sali, Dugi Otok Island (Croatia)

Ivan Marić

, Ante Šiljeg

and Fran Domazetović

University of Zadar, Department of Geography, Trg kneza Višeslava 9, 23 000 Zadar, Croatia

Keywords: Wildfire Risk Ignition Index, GIS-MCDA, High-Resolution UAV Imagery, Perception.

Abstract: In recent years, wildfires have become one of the most hazardous natural disasters because of their overall

impact on the natural and urban environment. In this paper, we have generated a wildfire risk ignition index

for the Sali settlement (Dugi Otok, Croatia). This model was generated within the INTERREG PEPSEA

(Protecting the Enclosed Parts of the Sea in Adriatic from pollution) project. Wildfire ignition index is based

on the GIS-MCDA (Multi-Criteria Decision Analysis). The process was performed using 13 criteria grouped

in five clusters. Criteria were derived from high-resolution multispectral (5 bands) orthomosaic and digital

terrain model (DTM) produced from imagery acquired with Matrice 600 Pro and Matrice 210 RTK V2 UAV.

The criteria weights were determined using the AHP (Analytic Hierarchy Process). The model of wildfire

ignition risk was classified into five classes, from very low (1) to very high (5). The model indicates that

14.14 % of the study area falls in a very high (5) ignition risk zone. The fire-risk perception was analyzed and

the wildfire ignition model was evaluated using a questionnaire. The results indicate that all recent wildfire

ignition locations occurred in high (4) and very high (5) risk class. Furthermore, the population recognized

wildfires as a moderate threat to the ecosystem of the wider Sali area. A set of specific management measures

has been proposed to prevent wildfire ignition. This proposed methodological framework and results can

provide valuable information and specific management tools to local government.

1 INTRODUCTION

Wildfire (Pavlek et al., 2017) or wildland fire

(Eskandari, 2017) burns uncontrollably in a natural

environment in which the primary fuel is vegetation.

Wildfire is the one of most hazardous natural

disasters (Bonazountas et al., 2005) and important

cause of land degradation which lead to

desertification, deforestation (Eskandari, 2017), and

destabilization of soil-water conservation (Sharma,

2012). They can have profound effects on global gas

emissions, biodiversity, land cover change, health,

and local economies (Sebastián-López et al., 2008,

Somashekar et al., 2009, Thompson and Calkin,

2011, Ajin et al., 2016). One of the most important

phases in wildfire management are prevention

(Vasilakos et al., 2007, Sebastián-López et al., 2008)

https://orcid.org/0000-0002-9723-6778

https://orcid.org/0000-0001-6332-174X

https://orcid.org/0000-0003-3920-6703

and early detection (Doolin and Sitar, 2005, Hefeeda

and Bagheri, 2007, Vescoukis et al., 2012). Namely,

risk management begins with an assessment of the

areas with the highest possibility of fire ignition

(Gigović et al., 2018). An important measure in fire

prevention is a derivation of the fire ignition risk

(Roland et al., 2015) which can indicate the

vulnerable areas and can provide specific

management tools to authorities (Bonazountas et al.,

2005). Fire ignition risk refers to the chance of a fire

starting as determined by the presence and activity of

any causative agent. It is regarded as an essential

element in analyzing and assessing fire danger

(Vasilakos et al., 2007, Catry et al., 2010).

Identification of factors affecting the ignition of forest

fire is one of the basic tools for forest fire control and

fighting actions. Zonation of fire risk ignition is one

Mari

c, I., Šiljeg, A. and Domazetovi

c, F.

Derivation of Wildﬁre Ignition Index using GIS-MCDA from High-Resolution UAV Imagery Data and Perception Analysis in Settlement Sali, Dugi Otok Island (Croatia).

DOI: 10.5220/0010465000900097

In Proceedings of the 7th International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2021), pages 90-97

ISBN: 978-989-758-503-6

of the basic tools for forest fire control and action

measures (Mohammadi et al., 2010).

The Mediterranean, one of the most flammable

ecosystems in the world (Pausas et al., 2016), is one

of the most endangered areas considering the wildfire

(Catry et al., 2010, Pavlek et al., 2017). As a

Mediterranean country, the Republic of Croatia (HR)

(Fig. 1A-B) has a constant increase in the danger of

wildfires, averaging over 1000 registered wildfires

annually (Pavlek et al., 2017) and it is recognized as

a country with high forest fire risk (Stipaničev et al.,

2007). The HR especially has a pronounced risk of

fire ignition in the coastal zone and on the islands

(Pavlek et al., 2017).

This article presents a wildfire risk ignition result

performed within the INTERREG Italy-Croatia

PEPSEA (Protecting the Enclosed Parts of the Sea in

Adriatic from pollution) project. The wildfire risk

ignition index was derived using GIS-MCDA and

acquired high-resolution (GSD<5 cm) UAV imagery

data (RGB and multispectral). Furthermore, the risk

perception about the dangers of wildfires was

examined by a questionnaire. The main goal of the

research was: (a) generate wildfire risk ignition index,

as a part of a quantitative rating system, using the

quickly definable parameters and GIS-MCDA; (b)

propose fire prevention measures based on available

resources of the study area with a basic aim of

reduction of fire ignitions; (c) examine residents'

awareness of a wildfire hazard. The research was

carried out within the drainage basin of the Sali

located on the island of Dugi Otok in HR.

2 STUDY AREA

The study area (235 ha) includes drainage basins of

the Sali and Sašćica bays located in the settlement

Sali (Dugi Otok island, Croatia) (Fig. 1C). This

landscape is dominated by abandoned agricultural

(dominance of olive groves) areas with Aleppo pine

forests and predominantly degraded holm oak forests.

3 MATERIALS AND METHODS

3.1 GIS-MCDA

Wildfire ignition index was derived using

multicriteria GIS analysis (GIS-MCDA). The GIS-

MCDA process was performed in six steps which

included: (1) identification of problem and definition

of the main goal, (2) determination of criteria and

constraints, standardization of criteria (3),

determination of weight coefficients (ponders) (4),

criteria aggregation (5) and validation of created

model (6) (Fig. 2) (Malczewski and Rinner, 2015,

Domazetović et al., 2019).

Figure 1: Location of Sali settlement.

3.1.1 Selection of Evaluation Criteria

All predisposing criteria of fire ignition were

generated from the digital terrain model (DTM) and

multispectral orthomosaic produced from high-

resolution UAV imagery (Fig. 2). Four groups of

continuous values criteria and two (boolean) criteria

were used in the GIS-MCDA. These groups were (1)

morphometric (slope, elevation, aspect, terrain

ruggedness, topographic wetness index - TWI); (2)

vegetation (land cover, normalized difference

vegetation index - NDVI); (3) climate (insolation,

heat load index - HLI); and (4) anthropogenic

(distance from a road, distance from housing units)

(Fig. 2).

Aspect affects the amount of sunlight the area

receives and temperature. The study area is located in

the northern hemisphere therefore southern slopes

receive more sunlight. Elevation as an important

physiographic variable affects the volume of rainfall,

air humidity, vegetation patterns, and exposure to

wind (Tiwari et al., 2021). Fire ignitions at higher

elevations are generally less frequent due to lower

temperatures and higher rainfall. Most wildfires occur

on slopes between 0 and 20º. It has been found that

the rate of wildfire ignition decreases with a higher

slope (Swanson, 2018). Also, fire ignitions are more

Derivation of Wildﬁre Ignition Index using GIS-MCDA from High-Resolution UAV Imagery Data and Perception Analysis in Settlement

Sali, Dugi Otok Island (Croatia)

Figure 2: Methodological framework of wildfire ignition

index generation using MCDA-GIS.

frequent at less complex (ruggedness) terrain. The

topographic wetness index (TWI) is a measure of

long-term moisture that uses the upslope contributing

areas and slope to determine an index of moisture

(Iverson et al., 2004). Higher values mean more

tendency of an area to accumulate water (Mattivi et

al., 2019) therefore these areas have a lower risk of

wildfire ignition. Land cover is the key factor in the

ignition of wildfire (Carmo et al., 2011). In a

Mediterranean wildfire, risk should be higher for

shrublands, pine stands and, grasslands than

croplands and broadleaf forests. Normalized

difference vegetation index (NDVI) can indicate

higher vegetation dryness due to water stress which is

a predisposing factor for fire occurrence (Maselli et

al., 2003). Decreased values of NDVI in the

Mediterranean can be linked to a higher probability

of fire ignition during summer (Zipoli et al., 2000).

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

(Gigović et al., 2018). The heat load index (HLI) is a

parameter that takes into account the steepness of the

slope when calculating the amount of solar radiation

received by the slope. Area solar radiation tool was

used to calculate the insolation across a study area. In

both parameters, higher values indicate a higher risk

of fire ignition. Using the geographic object-based

image analysis (GEOBIA), built-up areas (I) and

water surfaces (II) were extracted and used as (5)

boolean criteria. These criteria represent the areas

where wildfire can't occur.

3.1.2 Production of Digital High-Resolution

Terrain Model (DTM) and

Multispectral Orthomosaic Model

First, the high-resolution digital surface model was

derived from aerial images collected with the

unmanned aerial vehicle (UAV) (Matrice 210 RTK +

Zenmuse X7-16mm) (Fig. 3

1.1

). The flight was

performed in the DJI GSPro application. Aerial

imagery was acquired using the double grid mission

with front and side overlap of 80% (Fig. 3

1.2

). The

flying height was around 200 m. Camera self-

calibration was done in Agisoft Metashape 1.5.1. The

image workflow process was done in five steps: (1)

orientation of aerial imagery; (2) addition of ground

control points (GCP); (3) creation of a dense cloud;

(4) creation of a digital surface model (DSM); and (5)

creation of a digital orthophoto (DOP). A total of 10

GCP were collected with the GNSS device Stonex

S10 (Fig. 3

1.3

). Total RMSE of GCP was 4.3 cm.

Since the terrain characteristics are important for

deriving the criteria the final step involved generating

a digital terrain model (DTM) through correction and

filtering of the DSM. A correction was performed

using the DSM2DTM algorithm (Chirico et al., 2020)

which gradually removes anthropogenic and natural

elements elevated above the bare ground and

smoothes the final model by removing surface

irregularities (Fig. 4).

Multispectral orthomosaic was derived from

aerial images collected with UAV DJI Matrice

600Pro on which a Red Edge-Mica SenseMX camera

was mounted (Fig. 4). Radiometric calibration of the

multispectral camera using a calibrated reflectance

panel (CRP) was done before and after each mission.

GISTAM 2021 - 7th International Conference on Geographical Information Systems Theory, Applications and Management

Figure 3: Derivation of DSM and multispectral orthomosaic

model.

Figure 4: Conversion of DSM in DTM.

3.1.3 Standardization of Criteria

Standardization of criteria was conducted so their

mutual comparison on the same numerical scale

would be possible (Malczewski and Rinner, 2015).

Standardization was performed with a numerical

interval from 1 to 5 where class (1) referred to very

low ignition risk, (2) low, (3) moderate, (4) high, and

(5) very high. The criteria were standardized by

combining two methods. The decision-maker

standardization method (Domazetović et al., 2019)

was used for criteria that were used in similar case

studies. Jenks (natural breaks) classification method

was used to optimize the arrangement of a set of

values into "natural" classes for criteria where it is

difficult to accurately determine class boundaries to

the risk of a wildfire ignition risk.

3.1.4 Analytical Hierarchy Process (AHP)

Determination of weighting coefficients (Table 1)

was performed using the AHP, which ranks the

selected criteria according to their importance, i.e. the

level of influence on the decision or model. The

criteria were compared based on a scale of absolute

values that represent the extent to which one criterion

dominates over the other (Saaty, 2001). Furthermore,

a wildfire risk model has been created where all

criteria have the same weighting factor.

Table 1: Weighting coefficients (Woc) for wildfire ignition

predisposing criteria.

1234567 8 9 10 11Woc

1 1 2 3 3 4 4 5 5 6 7 1 0.26

20.50 1 2 2 3 3 4 4 5 6 0.500.18

3 0.33 0.50 1 1 2 2 3 3 4 5 0.330.12

4 0.33 0.50 1 1 2 2 3 3 4 5 0.330.12

5 0.25 0.33 0.50 0.50 1 1 2 2 3 4 0.250.08

6 0.25 0.33 0.50 0.50 1 1 2 2 3 4 0.25 0.08

7 0.20 0.25 0.33 0.33 0.50 0.5 1 1 2 3 0.200.05

8 0.20 0.25 0.33 0.33 0.50 0.5 1 1 2 3 0.200.05

9 0.17 0.20 0.25 0.25 0.33 0.33 0.50 0.50 1 2 0.17 0.03

10 0.14 0.17 0.20 0.20 0.25 0.25 0.33 0.33 0.50 1.00 0.14 0.02

11 0.11 0.13 0.14 0.14 0.17 0.17 0.25 0.25 0.33 0.50 0.11 0.02

1 - land cover, 2 - distance from roads, 3 - aspect, 4 - area solar

insolation, 5 - distance from housing unit. 6 - NDVI, 7 - HLI, 8 -

slope, 9 - elevation, 10 - TWI, 11 – ruggedness

3.2 Collection of Vegetation Data

After multispectral orthomosaic production, the field

(in-situ) vegetation samples were collected. Samples

were acquired to make easier identification of

vegetation species and to facilitate the process of

deriving the land cover (LC) model. Samples were

Derivation of Wildﬁre Ignition Index using GIS-MCDA from High-Resolution UAV Imagery Data and Perception Analysis in Settlement

Sali, Dugi Otok Island (Croatia)

collected with a process divided into four steps (Fig.

5). A total of 390 samples were collected.

Figure 5: Methodological framework for collection of

vegetation type data.

3.3 Wildfire Perception Analysis

The questionnaire was conducted in the period from

25 to 30 June 2020. It involved 38 respondents, which

is 5% of the Sali population. In the project, the degree

of potential 33 threats (including wildfire) to the

natural environment was examined. Each

questionnaire was conducted at a different address

(Fig. 6). The type of sample was stratified, and the

selection was random. Only adult citizens were

selected.

Figure 6: Surveying respondents in Sali.

4 RESULTS AND DISCUSSION

4.1 Wildfire Ignition Index Models

Two models of ignition index have been derived (Fig.

7). In the first model weight coefficients were

determined by the AHP method (Table 1). In the

second model, all defined predisposing criteria had

equal weighting coefficients (0.091).

Figure 7: Derived models of wildfire ignition index.

In both cases, the models were classified using the

equal interval method. The share of risk classes in the

total studied area was calculated. In model 1, very

high (5) and high (4) classes occupy almost 60% of

the total area. The most risk area includes the

neglected agricultural area and evergreen vegetation

near the roads. In model 2, very high (5) and high (4)

classes include 47% of the total area. However, the

areas of the highest risk of fire ignition generally

coincide with the first model.

Vegetation cover, dominated by Aleppo pine,

maquis, other shrublands, and neglected agricultural

areas, combined with other predisposing factors make

the wider area of Sali very risky from fire ignition.

This is not surprising given that the Dalmatian coast

and islands are classified as the most endangered

areas of wildfire ignition in HR. Therefore, we

believe that the proposed framework, with minor

modifications, can be applied in other Mediterranean

countries that have a high risk of wildfires.

GISTAM 2021 - 7th International Conference on Geographical Information Systems Theory, Applications and Management

4.2 Proposed Measures for Wildfire

Prevention

A system of surveillance cameras has been proposed

with the aim of preventing and timely detection of

wildfires. Within the study area, a binary visibility

analysis was performed at two proposed locations

where cameras could monitor very high risk (class 5)

areas of wildfire ignition. In total, (model 1 + model

2) 39.21 ha of the study area has a very high risk of

wildfire ignition (Fig. 8A).

The camera at a height of 10 m and the range of

surveillance with a radius of 2 km was assumed. The

installed cameras could monitor an area of 114.58 ha,

of which 38.21 ha falls into the category of the very

high risk (Fig. 8B), which means that 97.45% of this

surface could be monitored from proposed locations.

Other measures are also proposed: raising the

level of awareness about the dangers of wildfire;

revitalization of abandoned and neglected

agricultural plots; thinning and cleaning of forests and

construction of narrow, cleared paths to achieve

easier movement in the terrain in the event of a

wildfire spreading.

Figure 8: Coverage of very high (5) wildfire ignition risk

area with surveillance camera.

4.3 Results of Risk Perception Analysis

The average age of 38 respondents was 41.36 years.

In comparison to all analyzed threats to the natural

environment of the settlement Sali, respondents

(n=38) have evaluated the risk of wildfire ignition in

17th place out of 33 analyzed threats (Fig. 9).

Respondents have rated the risk of wildfire ignition

as moderate (3.00), while the standard deviation in

responses was 1.16.

Figure 9: Perceptions of the threats to the natural

environment of the Sali wider area.

Since official data about the historical location of

wildfires in Sali settlement do not exist, respondents

have detected recent locations of wildfires on the

generated high-resolution DOP. It is necessary to

point out that this is the main drawback of the

research. Namely, there is a lack of wildfire

occurrence data to accurately validate the model.

Respondents were able to detect only a few recent

wildfires. The difference between the models is

Derivation of Wildﬁre Ignition Index using GIS-MCDA from High-Resolution UAV Imagery Data and Perception Analysis in Settlement

Sali, Dugi Otok Island (Croatia)

difficult to determine since in both models all wildfire

ignition locations are located within (4) high or very

(5) high-risk classes. (Fig. 10).

Figure 10: Detected locations of wildfire ignition.

5 CONCLUSION

High-resolution UAV imagery (RGB and

multispectral) and GIS-MCDA were used to derive a

wildfire ignition index. The wider area of Sali

settlement can be considered as a high-risk area for

wildfire ignition. Risk perception analysis showed

that the respondents perceived wildfires as a moderate

(x ̅=3.00) threat to their natural environment. A set of

specific measures (surveillance cameras, forest

thinning, etc.) has been proposed to prevent wildfire

ignition. In future research, the presented

methodology framework will be applied to a larger

study area. The GIS-MCDA will be expanded with

additional criteria (e.g. power lines, landfill sites)

depending on the characteristics of the study area.

Also, more wildfire occurrence data will be collected

for model validation.

ACKNOWLEDGEMENTS

This work has been supported by INTERREG

PEPSEA project and Croatian Science Foundation

under the project UIP-2017-05-2694.

REFERENCES

Ajin, R. S., Loghin, A. M., Jacob, M. K., Vinod, P. G., &

Krishnamurthy, R. R. (2016). The risk assessment study

of potential forest fire in Idukki Wildlife Sanctuary

using RS and GIS techniques. International Journal of

Advanced Earth Science and Engineering, 5(1), 308-

318.

Bonazountas, M., Kallidromitou, D., Kassomenos, P. A., &

Passas, N. (2005). Forest fire risk analysis. Human and

Ecological Risk Assessment, 11(3), 617-626.

Carmo, M., Moreira, F., Casimiro, P., & Vaz, P. (2011).

Land use and topography influences on wildfire

occurrence in northern Portugal. Landscape and Urban

Planning, 100(1-2), 169-176.

Catry, F. X., Rego, F. C., Bação, F. L., & Moreira, F. (2010).

Modeling and mapping wildfire ignition risk in

Portugal. International Journal of Wildland Fire, 18(8),

921-931.

Chirico, P. G., Bergstresser, S. E., DeWitt, J. D., & Alessi,

M. A. (2020). Geomorphological mapping and

anthropogenic landform change in an urbanizing

watershed using structure-from-motion

photogrammetry and geospatial modeling techniques.

Journal of Maps, 1-12.

Domazetović, F., Šiljeg, A., Lončar, N., & Marić, I. (2019).

Development of automated multicriteria GIS analysis

of gully erosion susceptibility. Applied geography, 112,

102083.

Doolin, D. M., & Sitar, N. (2005, May). Wireless sensors

for wildfire monitoring. In Smart Structures and

Materials 2005: Sensors and Smart Structures

Technologies for Civil, Mechanical, and Aerospace

Systems (Vol. 5765, pp. 477-484). International

Society for Optics and Photonics.

Eskandari, S. (2017). A new approach for forest fire risk

modeling using fuzzy AHP and GIS in Hyrcanian

forests of Iran. Arabian Journal of Geosciences, 10(8),

190.

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 vjesnik, 25(3), 891-897.

Hefeeda, M., & Bagheri, M. (2007, October). Wireless

sensor networks for early detection of forest fires. In

2007 IEEE International Conference on Mobile Adhoc

and Sensor Systems (pp. 1-6). IEEE.

Iverson, L. R., Prasad, A. M., & Rebbeck, J. (2004). A

comparison of the integrated moisture index and the

topographic wetness index as related to two years of

soil moisture monitoring in Zaleski State Forest, Ohio.

In: Yaussy, Daniel A.; Hix, David M.; Long, Robert P.;

Goebel, P. Charles, eds. Proceedings, 14th Central

Hardwood Forest Conference; 2004 March 16-19;

Wooster, OH. Gen. Tech. Rep. NE-316. Newtown

Square, PA: US Department of Agriculture, Forest

Service, Northeastern Research Station: 515-517.

Malczewski, J., & Rinner, C. (2015). Multicriteria decision

analysis in geographic information science (pp. 220-

228). New York: Springer.

Maselli, F., Romanelli, S., Bottai, L., & Zipoli, G. (2003).

Use of NOAA-AVHRR NDVI images for the

estimation of dynamic fire risk in Mediterranean areas.

Remote Sensing of Environment, 86(2), 187-197.

Mattivi, P., Franci, F., Lambertini, A., & Bitelli, G. (2019).

TWI computation: a comparison of different open

source GISs. Open Geospatial Data, Software and

Standards, 4(1), 1-12.

GISTAM 2021 - 7th International Conference on Geographical Information Systems Theory, Applications and Management

Mohammadi, F., Shabanian, N., Pourhashemi, M., & Fatehi,

P. (2010). Risk zone mapping of forest fire using GIS

and AHP in a part of Paveh forests. Iranian Journal of

Forest and Poplar Research, 18(4), 569-586.

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.

Pausas, J. G., Alessio, G. A., Moreira, B., & Segarra-

Moragues, J. G. (2016). Secondary compounds enhance

flammability in a Mediterranean plant. Oecologia,

180(1), 103-110.

Roland, V., Marić, I., & Milošević, R. (2015). Application

of GIS technology in firefighting. Vatrogastvo i

upravljanje požarima, (1.), 57-71.

Saaty, T. L. (2001). Fundamentals of the analytic hierarchy

process. In The analytic hierarchy process in natural

resource and environmental decision making (pp. 15-

35). Springer, Dordrecht.

Sebastián-López, A., Salvador-Civil, R., Gonzalo-Jiménez,

J., & SanMiguel-Ayanz, J. (2008). Integration of socio-

economic and environmental variables for modelling

long-term fire danger in Southern Europe. European

Journal of Forest Research, 127(2), 149-163.

Sharma, L. K., Kanga, S., Nathawat, M. S., Sinha, S., &

Pandey, P. C. (2012). Fuzzy AHP for forest fire risk

modeling. Disaster Prevention and Management: An

International Journal.

Somashekar, R. K., Ravikumar, P., Kumar, C. M., Prakash,

K. L., & Nagaraja, B. C. (2009). Burnt area mapping of

Bandipur National Park, India using IRS 1C/1D LISS

III data. Journal of the Indian Society of Remote

Sensing, 37(1), 37-50.

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.

Swanson, V. (2018). Fire Ignition and Burn Risk: A Study

in Trinity and Los Angeles Counties in California.

Thompson, M. P., & Calkin, D. E. (2011). Uncertainty and

risk in wildland fire management: a review. Journal of

environmental management, 92(8), 1895-1909.

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.

Natural Hazards, 105(2), 1189-1230.

Vasilakos, C., Kalabokidis, K., Hatzopoulos, J., Kallos, G.,

& Matsinos, Y. (2007). Integrating new methods and

tools in fire danger rating. International Journal of

Wildland Fire, 16(3), 306-316.

Vescoukis, V., Doulamis, N., & Karagiorgou, S. (2012). A

service oriented architecture for decision support

systems in environmental crisis management. Future

generation computer systems, 28(3), 593-604.

Zipoli, G., Costantini, R., Romanelli, S., Bottai, L., &

Maselli, F. (2000). Use of satellite and ancillary data for

the evaluation of structural and meteorological forest

fire risks in Tuscany (Central Italy). In Proceedings of

ECAC.

Derivation of Wildﬁre Ignition Index using GIS-MCDA from High-Resolution UAV Imagery Data and Perception Analysis in Settlement

Sali, Dugi Otok Island (Croatia)