Journal of Volcanology and Geothermal Research 288 (2014) 132–143

Contents lists available at ScienceDirect

Journal of Volcanology and Geothermal Research

j ourna l homepage: www.e lsev ie r .com/ locate / jvo lgeores

A new Volcanic managEment Risk Database desIgn (VERDI): Applicationto El Hierro Island (Canary Islands)

S. Bartolini ⁎, L. Becerril, J. MartíGroup of Volcanology, SIMGEO (UB-CSIC), Institute of Earth Sciences Jaume Almera, ICTJA-CSIC, Lluís Solé i Sabarís s/n, 08028 Barcelona, Spain

⁎ Corresponding author. Tel.: +34 934095410.E-mail address: sbartolini@ictja.csic.es (S. Bartolini).

http://dx.doi.org/10.1016/j.jvolgeores.2014.10.0090377-0273/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 16 May 2014Accepted 2 October 2014Available online 28 October 2014

Keywords:Database designVolcanic riskDecision makingEl Hierro

One of the most important issues in modern volcanology is the assessment of volcanic risk, which will depend –

among other factors – on both the quantity and quality of the available data and an optimum storagemechanism.This will require the design of purpose-built databases that take into account data format and availability and af-ford easy data storage and sharing, andwill provide for amore complete risk assessment that combines differentanalyses but avoids any duplication of information. Data contained in any such database should facilitate spatialand temporal analysis thatwill (1) produce probabilistic hazardmodels for future vent opening, (2) simulate vol-canic hazards and (3) assess their socio-economic impact.We describe the design of a newspatial database struc-ture, VERDI (Volcanic managEment Risk Database desIgn), which allows different types of data, includinggeological, volcanological, meteorological, monitoring and socio-economic information, to bemanipulated, orga-nized and managed. The root of the question is to ensure that VERDI will serve as a tool for connecting differentkinds of data sources, GIS platforms and modeling applications. We present an overview of the database design,its components and the attributes that play an important role in the database model. The potential of the VERDIstructure and the possibilities it offers in regard to data organization are here shown through its application onEl Hierro (Canary Islands). The VERDI database will provide scientists and decision makers with a useful toolthat will assist to conduct volcanic risk assessment and management.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Volcanic risk assessment and management are complex issues duelargely to the nature, variety and availability of the data they handle(De la Cruz-Reyna, 1996). The quality of the data will determine theevaluation of the volcanic risk, which is an essential part of risk-baseddecision making in land-use planning and emergency management.The first step in the evaluation of volcanic risk consists of obtainingand organizing all pertinent data derived from disciplines such as geol-ogy, volcanology, geochemistry, petrology and seismology, as well asvulnerability and socio-economic information relating to the elementsthat are potentially at risk. Some of the most relevant issues includehow and where to store the data, in which format should it be madeavailable, and how to facilitate its use and exchange. Thus, it is essentialto design an appropriate database that is specifically adapted to the taskof evaluating and managing volcanic risk.

The design of an appropriate database for risk assessment andmanagement should aim to organize all the available and necessaryinformation on volcanic risk assessment in a standardized way that iseasy to consult and exchange.

As in any other field, the first step in designing a database for volca-nic risk assessment andmanagement is the definition of its architecture.This must allow for effective interaction between the different informa-tion fields and offer users a clear vision of its internal organization andrapid access to its contents. Nevertheless, it will be the quantity andquality of the information contained in the database that will determinethe reliability and validity of the final risk analysis. Subsequent stepswill consist of the creation, maintenance and updating of all data relatedto volcanic risk. It is important to ensure that the databasewill be able toevolve freely from a simple to amore complex structure and be updatedwhen new data are available.

To facilitate its operability and the visualization of the data the data-base must be integrated into a GIS (Geographic Information System). AGIS is an organized integration of software, hardware and geographicdata designed to capture, store, manipulate, analyze and representgeoreferenced information (Longley et al., 2005). In recent years, theuse of GIS and the improvement of the modeling of volcanic processeshave become useful tools in volcanic hazard and risk assessment. Infact, susceptibility, hazard, vulnerability and riskmaps have been gener-ated using GIS tools (Pareschi et al., 2000; Felpeto et al., 2007; Barrecaet al., 2013) and can be represented in a GIS environment as a supportfor spatial decision making (Cova, 1999).

Furthermore, thematic volcanic risk maps can facilitate land-useplanning and appropriate actions required during emergencies. In fact,

133S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

hazard and riskmaps are key tools in emergency management: the for-mer depicts the hazard at any particular location,while the latter showsthe spatial variation of both hazard and vulnerability (Lirer et al., 2001).

To date, the databases used in volcanology have been createdto store and analyze different types of information and have beenemployed to analyze, for example, (1) the impacts of volcanic phenom-ena on people (Witham, 2005); (2) potentially active volcanoes situatedin regions of high geodynamic unrest (Gogu et al., 2006); (3) collapsecalderas (Geyer and Martí, 2008); (4) volcano monitoring data that in-clude instrumentally and visually recorded changes in seismicity,ground deformation, gas emission and other parameters (WOVOdat(Venezky and Newhall, 2007)); (5) global volcanic unrest (Phillipsonet al., 2013); and (6) active faults on Mt. Etna (Barreca et al., 2013).In particular, efforts have been made to construct a Global VolcanicRisk database of largemagnitude explosive volcanic eruptions (LaMEVE(Crosweller et al., 2012)). However, none of the existing databasesis based on a simple architecture that contains all the necessary infor-mation for volcanic risk analysis and management.

Herewe present VolcanicmanagEment Risk Database desIgn (VERDI),the architecture for a geodatabase for volcanic risk assessment andmanagement. The rationale behind constructing this database is theneed to create a comprehensive structure including all known or iden-tified fields that might contribute to the assessment of volcanic risk.The database also aims to make the task of volcanic risk managementeasier for decision makers. Currently, relevant data are stored in a vari-ety of different formats and are not always easily accessible. Thus, thisnew way of compiling extensive data aims to provide an accessibleand useful structure that will facilitate information sharing and riskassessment. This new database has been designed to work in a GISenvironment.

The ultimate aim of VERDI is to create a platform for expanding,updating and sharing information that is open to the incorporation ofnew data. In the future, a website could be set up to make it a trulyuser-friendly application.

In this paper, we also present an example of the applicability ofVERDI, taking as example the island of El Hierro (Canary Islands,Spain). We show how all the available data necessary for conducting apreliminary risk assessment can be integrated and discuss the limita-tions of existing data and the inherent advantages in storing data inthe proposed form.

2. VERDI architecture

A simplified version of the VERDI database design structure is shownin Fig. 1. The full version of the VERDI structure and the usermanual willbe published online on the website of the CSIC Barcelona VolcanologyGroup (http://www.GVB-csic.es/).

The design of the database has taken into account the type of datarequired and possible inter-relationships in order to avoid duplication.

The first steps in the creation of the database model were the collec-tion of metadata, the analysis of the required features and the calculationof the expected output responses. This phase included the creation ofinformation groups and the definition of the table fields and the relation-ships between tables.

In order to optimize the accurate evaluation of volcanic risk, VERDIcontains 13 information groups regarding past and current volcanicactivity and the associated hazards and the potential vulnerability ofthe elements that may be affected by such hazards. The informationincluded in each group is recorded in individual tables. Additionally,VERDI includes spatial features that can be visualizedwith aGIS applica-tion. The rationale behind the VERDI architecture is based on the princi-ple that all the information concerning the evaluation of volcanic riskshould be comparable, consistent and available for future comparisonsand data analyses.

In the following subsections we offer a brief description of eachinformation group and the type of data included therein.

2.1. GroupCore

GroupCore is the central group of VERDI and represents themetadatainformation of all the actions that could be incorporated into thedatabase as new data. This group governs the recorded informationadded to each group of the database, thereby controlling the insertionof new data.

The tables contained in this group are ACTION, ACTION_TYPE,PROJECT, REPORT and SUPPORT (see Fig. 1 in Supplementary material1): ACTION and ACTION_TYPE correspond to actions and the type ofactions, respectively, that generate new data (volcanic event, fieldwork,bibliography, etc.); PROJECT is a reference to a project undertaken by aninstitution such as aministry or an institute; the REPORT table describesthe action and includes information about the project related to the ac-tion; and SUPPORT adds information about the entity that is managingor funding the project.

2.2. GroupVolcano

This group contains information about the volcano or the studiedvolcanic area and includes data on the volcanic event itself, the charac-teristics of the type of volcanism and the magnitude of the event(Fig. 2).

The table VOLCANO provides general information about the loca-tion of the volcano and volcanic area, which will normally be associ-ated with spatial information included in a shapefile of polygonsor in raster images. Spatial features contain a folder with additionalinformation such as Digital Elevation Models (DEMs), hillshades andorthophotos.

VOLCANO_TYPE completes the information about the volcano andidentifies different types and features of volcanoes (stratovolcano,shield volcano, etc.). ERUPTIVE_EVENT provides information abouteruptive events including date and location and enables the volcano-stratigraphy of the volcano and the study area to be obtained.ACTIVITY_TYPE characterizes the eruptive behavior of the volcanothus: Hawaiian, Strombolian, Vulcanian, Peléan/Plinian, Plinian, Ultra-plinian and/or Caldera. The size and magnitude of the eruption arecontained in the VEI_MAGNITUDE table, which includes parameterssuch as volume, column height, fragmentation index, dispersionindex, Dense-Rock Equivalent (DRE), magnitude (Pyle, 2000) and theVolcanic Explosivity Index (VEI) according to Newhall and Self's(1982) classification.

2.3. GroupSusceptibility

Volcanic susceptibility (i.e. the probability of vent opening) repre-sents an important step in simulating eruptive scenarios anddevelopinghazard maps (Martí and Felpeto, 2010). Thus, GroupSusceptibilitycontains information on all structural elements such as vents, dykes,faults, fractures and eruptive fissure-alignments obtained from bothgeological and geophysical studies. The location of gas emissionsor water springs, as well as thermal anomalies related to the volcanicactivity, are also included in this group. All of these elements enable sus-ceptibility maps in long-term analyses to be generated. During volcanicunrest episodes, real-time monitoring information – in particularregarding the location of the volcano-tectonic seismicity and surfacedeformation – can be added to permit the susceptibility to be re-evaluated. This group also contains a GEOPHYSICS subgroupwith infor-mation on structural geophysics that includes data derived fromstructural studies using different geophysical techniques such as self-potential, tomography, magnetometry, magnetotelluric and gravime-try. This type of geophysical data is useful in susceptibility analysesand in both short- and long-term hazard evaluations. In addition, itis useful for studying dispersed volcanic fields and their relation tolocal tectonics (Barde-Cabusson et al., 2014) and can thus facilitatea complete analysis of the probability of future activity in monogenetic

Fig. 1. VERDI database design structure.

134 S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

fields and improve understanding of the internal structure of compositevolcanoes (Rout et al., 1993; Blakely et al., 1997; Connor et al., 2000;Kiyosugi et al., 2010).

Moreover, in both short- and long-term hazard assessments themonitoring and interpretation of geophysical parameters such as tem-poral gravity changes, seismicity and ground deformation can benefitfrom integrationwith structural geophysical data. Fig. 3 shows the orga-nization of this group.

2.4. GroupHazard

GroupHazard contains basic data for computing volcanic hazardsto be employed in simulation models that take susceptibility infor-mation into account. This group constitutes the informationon which territorial and emergency plans should be based and hasbeen divided into LONG-TERM and SHORT-TERM subgroups (seeFig. 4).

Fig. 2. GroupVolcano structure.

135S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

The tables of the LONG-TERM hazard subgroup contain mainlydata regarding the products generated during the past activity ofthe volcano. The information required comes mainly fromgeological and historical records and laboratory analyses. Thissubgroup includes the following information split into different tables:magma and volcanic products (lava flows, pyroclastic deposits, etc.);petrological and geochemical data from volcanic rock samples; grain-size classification of pyroclasts based on sieved samples; andmorphometry.

The SHORT-TERM hazard subgroup tables contain monitoring datacollected during an unrest episode. These data are useful for short-

Fig. 3. GroupSuscepti

term hazard assessment in which also the monitoring data are takeninto account. The information is usually organized in terms of volcanicmonitoring networks (seismicity, deformation, gas, thermal, groundwa-ter, remote sensing images, etc.).

2.5. GroupMeteorology

GroupMeteorology includes the information required for the analysisof wind profiles, atmospheric parameters and precipitation data (seeFig. 2 in Supplementarymaterial 1). These parameters are very important

bility structure.

Fig. 4. GroupHazard structure.

136 S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

as inputs for ashfall simulations. Other important parameters included inthis group are related to the atmospheric diffusion coefficient, the erup-tion style and the grain-size classification. Ashfall simulations are very

useful in volcanic risk assessment and consider the impact of volcanicash not only on the population and infrastructures but also on aircraftsafety (Johnson et al., 2012).

137S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

2.6. GroupLaboratory

GroupLaboratory contains information supplementing theGroupSusceptibility and GroupHazard groups that relates to the labora-tories in which sample analyses are conducted. This group specifies thekind of samples used, the analytical tests applied and the results obtained(see Fig. 3 in Supplementary material 1). This group is importantfor controlling the quality of data used to characterize the expected typeof eruption (e.g. lava composition) by means of the analysis of pastproducts.

2.7. GroupDevice

GroupDevice provides information about the measurement devicesin tables such as PETROLOGY, SELF_POTENTIAL, MONITORING, andWIND. A large amount of information in the database is obtainedthrough the use of instruments such as seismographs and microscopesand the DEVICE table (see Fig. 4 in Supplementary material 1) containsthe names, models, types and functions of these devices.

2.8. GroupVulnerability

This group includes all the elements that could be affected by adestructive volcanic event.

Vulnerability is the potential of exposed elements to be directly orindirectly damaged by a given hazard (Scaini et al., 2014). There aremany types of vulnerability – physical, infrastructural, social and eco-nomic – and in combination they constitute the vulnerability of the sys-tem (Menoni et al., 2011). Physical vulnerability due to volcanic activityhas been widely observed and studied, in particular in recent decades(Blong and McKee, 1995; Annen and Wagner, 2003; Spence, 2004;

Fig. 5. GroupVulnera

Baxter et al., 2005; Spence et al., 2005; Gomes et al., 2006; Martí et al.,2008; Zuccaro et al., 2008; Scaini et al., 2014).

Thus, the VERDI database includes administrative divisions, infra-structure networks (TRANSPORT, ELECTRICITY, and WATER_SYSTEMtables), as well as a socio-economic table that includes POPULATION in-formation, FACILITY, BUILDING and LANDUSE (Fig. 5). A LAND_USE clas-sification is included because correct land-use planning is fundamentalin minimizing both loss of life and damage to property (Pareschi et al.,2000). The information contained in this part of the database is veryimportant in the organization of evacuation plans, the reduction of po-tential losses caused by the impact of volcanic and associated hazards,the design of land-planning measures, and the evaluation of potentialeconomic losses.

2.9. GroupCosts

GroupCosts (see Fig. 5 in Supplementary material 1) represents thehuge economic losses (human life, infrastructure, property, productivi-ty, etc.) that volcanic activity can cause. Estimating the economic costsassociated with volcanic eruptions is very difficult due to their durationand the variety of the types of impacts (Annen and Wagner, 2003).However, the quantitative estimation of economic losses is of primaryimportance when providing mitigation recommendations aimed at re-ducing damage (Spence et al., 2005).

The ECONOMIC_LOSSES and VOLCANO_IMPACT tables refer tothe economic and human losses evaluated after a volcanic crisis andthe economic impact for a specific volcanic event. The third table,SCENARIO_IMPACT, represents a support table that allows a cost evalu-ation to be added when a volcanic hazard scenario is computed andenables the human losses expected during a volcanic crisis to becalculated.

bility structure.

138 S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

2.10. GroupManagement

GroupManagement (see Fig. 6 in Supplementary material 1) is a use-ful group for decision makers and risk managers that should includeideally all types of emergency services (police, fire department, RedCross, NGOs, etc.), although in most cases Civil Protection bodies willtake responsibility during a volcanic crisis. Volcanic crises requirecontinuous close collaboration between Civil Protection bodies andscientists in order to best analyze observational and monitoring data,to evaluate short-termhazards, to drawupplans for optimizing existingmonitoring networks, to install new instruments and to provide advicein decision making (Bertolaso et al., 2009).

2.11. GroupReferences

GroupReferences contains contact information for key people andinstitutions, as well as bibliographic references (see Fig. 7 in Supple-mentary material 1) related to the data contained in the database.This group is important for obtaining the reference for any inputinto the VERDI database and thus enables the origin of the data to beknown; in this way, if necessary, the person or team in question canbe contacted if there is any explanation needed for the data introduced.

2.12. GroupModels

GroupModels contains examples of hazard-modeling tools. Itincludes the most relevant available software and a summary of boththe required main input parameters and the output formats.

In recent years, new tools have been developed for generatinghazard and risk maps, evaluating long- and short-term hazards,simulating different eruptive scenarios and designing evacuationplans. Examples of these tools include QVAST (Bartolini et al.,2013), VORIS (Felpeto et al., 2007), a model for lava flow simulation(Connor et al., 2012), HASSET (Sobradelo et al., 2014), BET_EF(Marzocchi et al., 2008), BET_VH (Marzocchi et al., 2010), HAZMAP(Bonadonna et al., 2002; Macedonio et al., 2005), FALL3D (Costaet al., 2006; Folch et al., 2009), TEPHRA2 (Connor et al., 2001),PUFFIN (Patra et al., 2013), VOLCFLOW (Kelfoun and Druitt, 2005),TITAN2D (Sheridan et al., 2005) and EJECT (Mastin, 2001).

Simplified schematic tables are given in the Supplementarymaterial2 with the main input parameters required for the above-listed tools.

2.13. GroupResults

This group contains all the results and outputs derived fromdifferentsimulations and analyses obtained using the database information. Thisgroup should facilitate the exchange of information that will be usefulfor future comparisons, analysis of data, and evaluation of volcanic risk.

3. VERDI usefulness: case study of El Hierro

One of the main obstacles when attempting to develop a robustdatabase is the lack of quality, well-gathered data. This issue can bemade simpler and easier in part by selecting small areas in which totest the operability database. With this aim in mind, a pilot project tocheck the feasibility of VERDI was set up with information availablefrom the island of El Hierro (Canary Island, Spain).

The last eruptiononElHierro that occurred in2011–2012 (López et al.,2012;Martí et al., 2013)demonstrated the importance of reliable data andtools that can enable scientific advisors and decision-makers to considerpossible future eruptive scenarios. Furthermore, this was the first evereruption in the Canary Islands to tracked in real-time (López et al., 2012).

Most of eruptions occurring on El Hierro are similar in type andin size, and consist of the emission of mafic lava flows, the ballisticprojection of pyroclasts and proximal fallout from low fire-fountains(Becerril et al., 2014). Its simple volcanic history, relative homogeneous

petrology and the new data collected during the last eruption, amongother factors, prompted us to select El Hierro as a case study for testingthe methodology proposed here.

In order to show the functionality of the VERDI database, we describehere two hypothetical phases of the volcanic risk assessment and the di-sastermanagement cycle (UNISDR, 2009).We analyze themost represen-tative and necessary information in each of the two periods in El HierroIsland: (1) the mitigation phase: actions aim to reduce the risk posed byhazards and (2) the preparation phase: the knowledge and capacities de-veloped by governments, professional response and recovery organiza-tions, communities, to effectively anticipate and to respond to theimpacts of likely, imminent or current hazard events or conditions. Inthese exampleswe try to summarize the information required to completea qualitative volcanic risk analysis for the island, show how to find and tostore data, outline the advantages of organizing the available data, visual-ize them in a GIS environment and the application of free e-tools.

Indeed, we believe that the availability of a well-organized databaseat the beginning of an unrest phase could become a very useful tool fordecision-makers and for the scientists that have to provide assessment.

3.1. Mitigation phase

The mitigation phase is a moment of relative calm during thevolcanic activity in which long-term volcanic hazard and risk assess-ment become feasible.

During this phase, the research, compilation and interpretation ofdifferent types of data should be carried out. Furthermore, availableinformation should be organized and stored in the database and com-pleted by further fieldwork, library searches and monitoring whereverdata is lacking.

Once uploaded, the data must be filtered before being used as inputsfor spatial and temporal analysis and for defining eruptive scenarios. Theresults obtained from the aforementioned analyses will become a usefultool for institutions such as Civil Protection when developing their emer-gency plans.

We assume that the Canary Islands Civil Protection Organizationneeds to know what impact the most likely eruption scenarios onEl Hierro will have on the population and the other exposed elements(property, infrastructures, communication networks, etc.). This informa-tion is useful for defining territorial planning, emergency plans, and foreducational programs. For this analysis different data layers will have tobe superimposed in order to reach a final risk assessment. Fig. 6 showsthe different data and steps to be performed in a GIS environment.

The first step is to obtain the Digital Elevation Model (DEM) ofEl Hierro and general information about the volcanic area and theeruptive event (see GroupVolcano). The DEM of this area can be freelyobtained from the website of Spanish Instituto Geográfico Nacional(IGN, http://centrodedescargas.cnig.es/CentroDescargas/index.jsp).

The second step involves the collection of volcano-structural data vianew fieldwork measurements and bathymetric information, as wellas the analysis of geological maps, orthophotos and aerial photographs,and remote sensing tools (Becerril et al., 2013). Once the wholevolcano-structural elements have been assembled, they can be geo-referenced on the DEM. These data represent the starting point toevaluate the susceptibility and to simulate different eruptive scenariosto go forward in the reduction of the risk.

In fact, the susceptibilitymap, i.e. the spatial distribution of new ventopenings, is based on the analysis of the aforementioned volcano-structural data. One of the tools that facilitates this type of analysis isQVAST (Bartolini et al., 2013) (the main input parameters required arespecified in the Supplementary material 2). All necessary data forconducting the susceptibility analysis is contained in GroupSusceptibility.In our example, we need to collect as much data as possible to improvethe accuracy of the spatial probability of a new vent opening. In thecase of El Hierro, data referring to past vent locations, dykes, eruptive fis-sures and faults are used. For example, to compile information related to

Fig. 6. Data layers in a GIS environment.

139S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

vents, we refer to the table VENT (Fig. 3) in the GroupSusceptibility thatcontains information about all known emission centers on the island.This table contains the information about the name of the volcano edifice,its coordinates (X, Y, Z), and additional description details. Also, the ge-ometry of the elements is defined and permits to visualize the vents aspoint features and use them inQVAST to obtain the smoothing parametervalue, that is, themost important parameter in the kernel density estima-tion to determine the shape of the probability density function (seeBartolini et al., 2013). This procedure needs to be also carried out withthe other volcano-structural elements and the final susceptibility map iscomputed assuming a non-homogeneous Poisson process. All the PDFsfor each volcano-structural data are combined in a weighted sum andthe result is the final susceptibility map. The map obtained for the

onshore distribution of future volcanic eruptions of El Hierro Island isshown in Fig. 7a. The total susceptibility map is available in Becerril etal. (2013). The result is a GeoTIFF raster file (map) in which each pixelhas a value that represents the probability that itwill host a newemissioncenter.

Once the susceptibility map has been drawn up, eruptive scenariosfor hazard assessment can be computed. In Becerril et al. (2014), differ-ent eruptive scenarios such as lava flows, ashfall and pyroclastic densitycurrents (PDCs) were considered and enabled a qualitative volcanichazard map to be generated. Here, we show (Fig. 8) a lava flow simula-tion probability map using VORIS tool (Felpeto et al., 2007) that requiresdifferent types of input parameters that can be found in VERDI, relatedto the past eruptive activity (see Supplementary material 2). The

a) b)

Fig. 7. Susceptibility map of the future volcanic eruptions at El Hierro obtained with QVAST tool (Bartolini et al., 2013): a) onshore spatial probability calculated only using geological dataand b) onshore spatial probability calculated using geological data and adding monitoring data on seismic activity during the first days of unrest.

140 S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

probabilistic lava flowmodel applied is based on the assumption that to-pography and flow thickness play major roles in determining lava paths(Felpeto et al., 2007 and references therein). Input parameters requiredby the model include a Digital Elevation Model (DEM), maximum flowlengths and thickness of the flow. In the case of our example, simulationwas run over a DEM with a cell size of 50 m. We assumed flow lengthsof 15 km and 3 m of thickness, corresponding to the average value of in-dividual flows measured in the field according to Becerril et al. (2014).All the information obtained during fieldwork, which also comes fromthe bibliography or from the devices, is vital in determining these param-eters. The simulationswere run for all cells in the DEMand the sumof the5000 iterations provided amapwith the probability for any particular cellof being covered by a lava flow (Fig. 8).

Once the distribution of the eruptive scenarios has been developed,Civil Protection is then able to evaluate the most likely eruptive scenar-ios for the island and their impact on the population and other exposedfeatures. For this, relevant data on elements such as population andtransport networks must be obtained for analysis. For example, popula-tion data for El Hierro can be downloaded from the website of theInstituto Nacional de Estadística (INE, http://www.ine.es/). Data ontransport networks can be obtained from the IGN website andOpenStreetMap (http://downloads.cloudmade.com/), the latter a toolused by public administrations, NGOs and even Civil Protection bodiesto manage in the aftermath of disasters such as the Haiti earthquake.

The acquisition of this information allows evacuation routes andeven a preliminary evaluation of general losses due to volcanic hazardssuch as lava flows to be calculated.

3.2. Preparation phase

Entry into the preparation or unrest phase means that the volcanicsystem has reawakened. During this phase, monitoring data plays an im-portant role and is essential as a support for decisionmaking and, a short-term hazard assessment becomes necessary. For this reason, VERDI data-base contains monitoring information distributed in different groups andtables. GroupHazard contains precursor data such as deformation, seismicactivity and groundwater monitoring for short-term hazard assessment;the MONITORING table summarizes all the monitoring networks withina volcanic area. Furthermore, in this phase the communication, alert,and evacuation play an important role in the disaster management.

The 2011–2012 eruption on El Hierro was preceded by three monthsof unrest. From July 2011 onwards a dense multi-parametric monitoringnetwork including seismic and magnetic stations and GPS recorderswere deployed throughout the island by the InstitutoGeográficoNacional(IGN). Data recorded during this unrest episode contributed to the under-standing of the reawakening of the volcanic activity on the island. Ingeneral, this monitoring network assisted authorities in emergency

Fig. 8. Lava flow probability map using VORIS tool (Felpeto et al., 2007).

141S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

management (López et al., 2012) and prepared them for the eruption thatfinally started on 10 October 2011, 2 km off the southern coast.

During an unrest phase, the updating of the possible eruptivescenarios computed during the emergency planning is necessary, main-ly because the arrival of new data such as seismic information canchange previous susceptibility analysis and, consequently, eruptionforecasts. This may involve a change in the direction taken by the crisismanagement. For better understanding how the real-time monitoringinteract with the VERDI database and with the different GIS tools, wegive an example using the seismic information obtained during the2011 unrest in El Hierro Island, and the QVAST tool to evaluate the vol-canic susceptibility. During the unrest phase, starting on July 2011,seven new seismic stationswere installed in different parts of the islandand data was transmitted on real time to the IGN National SeismicNetwork, where earthquake locations and local magnitudes were calcu-lated (López et al., 2012; Domínguez Cerdeña et al., 2014). The informa-tion about the seismic activity can be downloaded from the website ofthe Instituto Geográfico Nacional (IGN, http://www.ign.es/ign/layoutIn/sismoFormularioCatalogo.do).

During the unrest phase, the volcanic susceptibility can change con-siderably depending on the variation of the seismic activity. Therefore, ifwe consider from the first moments of the seismic unrest, we can seehow this information conditions changes in the susceptibility map. Forsimplicity, we consider the beginning of the unrest on 19 July which

was accompanied by an important increase in seismicity, mostly locatedon the north of the island. Seismic activity alternated relatively calmperiods with high energy periods and most of the earthquakes werelocated in the El Golfo area (Fig. 9) at 10–15 km depth (see López et al.,2012 for further details).

Firstly, the information of the seismic activity needs to be stored inthe SEISMICITY table located in the GroupHazard in the SHORT-TERMhazard subgroup (Fig. 4), containing monitoring data. In this case, theSEISMICITY table needs to be linked to the IGN database to obtain datain real-time. Once we have obtained this information, we need to visu-alize the information of the seismic activity in ourGIS as point shapefilesand add this information in the QVAST tool to update the susceptibilitymap. In Fig. 9 it is shown how the seismic information is added to theQuantum GIS software (http://www.qgis.org) and the correspondingstructure of the SEISMICITY table. The new susceptibility map elaborat-ed through QVAST tool is shown in Fig. 7b. We can see how the suscep-tibility map changes compared to the previous one (Fig. 7a), adding aprobability of new vent opening also in the north of the island. Thisdetermines that the product extent of the eruptive scenario simulationswill change and, consequently, this should be taken into account to de-termine new emergency evacuation plans.

During an unrest phase economic losses may be estimated by usinginformation derived from spatial and temporal analysis. It is imperativethat data regarding possible costs, along with an a priori analysis of

MAGNITUDE

El Golfo Valley

Las Playas Valley

El JulanValley

Fig. 9. Seismic data at the beginning of the El Hierro unrest showed using the Quantum GIS software, and the corresponding structure of the SEISMICITY table.

142 S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

losses, are stored right from the beginning of the process (Baxter et al.,2005). In VERDI, we have added a support table (Table 1) for cost eval-uation when a hazard volcanic scenario is computed. In fact, eruptionmodels are able to provide, for an assumed eruption scenario, a detailedmap of the possible geographical distribution of the eruption products,with point by point estimates of the key parameters. Where theseparameters are known, it becomes possible to develop estimates ofthe eruption impact on buildings and infrastructure, and also on theiroccupants (Spence et al., 2009).

At the end of this phase, subsequently towhether or not the volcanicevent has occurred, there are two phases in the disaster managementcycle: the response and the recovery. During these phases, it is funda-mental to upload all the data obtained to the database (see GroupResult)in order to facilitate future risk evaluations.

4. Conclusions and final remarks

VERDI is a new design for a database for risk assessment. Its logicalstructure has been conceived in order to facilitate the interaction betweendata sets and to guarantee the maintenance and evolution of the system.It is essential that the database structure permits the exchange of stan-dardized information and the updating of data in order to prevent redun-dancy and repetitiveness. The VERDI database design aims tomake scientific research easier and to promote information-sharing for

Table 1GroupCosts: SCENARIO_IMPACT table.

Table Field

SCENARIO_IMPACT scenarioImpact_idscenarioImpact_typescenarioImpact_popscenarioImpact_facilityscenarioImpact_buildingscenarioImpact_landUsescenarioImpact_transportscenarioImpact_electricityscenarioImpact_waterSystempopulation_cdvolcano_cd

volcanic surveillance, susceptibility, hazard and vulnerability. Its structureis linked to a spatial database in a GIS environment, which is used to cre-ate susceptibility, hazard, vulnerability, and risk maps.

VERDI has been conceived to be used as a source for modeling soft-ware packages such as QVAST (Bartolini et al., 2013), HASSET(Sobradelo et al., 2014) and VORIS (Felpeto et al., 2007). New geologicalhazardmodels related to volcanic systems such as landslides, lahars andtsunamis could be included in order to complete geo-risk databases.VERDI also helps to identify the basic information required to conducthazard and risk assessment. We thus suggest that all the informationincluded in VERDI should be available for each volcanic area. We alsobelieve that it is important that information is stored in the same struc-ture and format.

A future role for VERDI will be the publication of an interactivewebsite that will enable registered users to access and share the infor-mation in the database, thereby allowing VERDI to become more dy-namic and to continue developing. However, we cannot ignore theinherent limitations of available data and the effect that this may haveon the interpretation of the compiled information. It is therefore vitalto acknowledge that both data and interpretations are dynamic, thatis, they have to be subject to continuous revision and updating. Forthis reason, VERDI needs to be freely available to all scientists interestedin volcanic risk assessment since only contributions from all will allowVERDI growing and evolving into the useful tool we envisage.

Info Type

Primary key AutoNumberType of scenario simulation (lava, pdc, ashfall, …) TextPopulation affected by eruptive scenario IntegerFacility affected by eruptive scenario IntegerBuilding affected by eruptive scenario IntegerLand use affected by eruptive scenario IntegerTransport affected by eruptive scenario IntegerElectricity network affected by eruptive scenario IntegerWater system affected by eruptive scenario IntegerForeign key POPULATION table IntegerForeign key VOLCANO table Integer

143S. Bartolini et al. / Journal of Volcanology and Geothermal Research 288 (2014) 132–143

Acknowledgments

This research has been partially funded by the European Commis-sion (FP7 Theme: ENV.2011.1.3.3-1; Grant 282759: VUELCO) and theMINECO grant CGL2011-16144-E. We thank Xavier Castelltort for theirsuggestions and Jorge Pedro Galve for thorough and helpful views. TheEnglish text was corrected by Michael Lockwood.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jvolgeores.2014.10.009.

References

Annen, C., Wagner, J.J., 2003. The impact of volcanic eruptions during the 1990s. Nat. Haz-ards Rev. 4 (4), 169–175.

Barde-Cabusson, S., Gottsmann, J., Martí, J., Bolós, X., Camacho, A.G., Geyer, A., Planagumà,Ll, Ronchin, E., Sánchez, A., 2014. Structural control of monogenetic volcanism in theGarrotxa volcanic field (Northeastern Spain) from gravity and self-potential mea-surements. Bull. Volcanol. 76, 788.

Barreca, G., Bonforte, A., Neri,M., 2013. A pilot GIS database of active faults ofMt. Etna (Sicily):a tool for integrated hazard evaluation. J. Volcanol. Geotherm. Res. 251, 170–186.

Bartolini, S., Cappello, A., Martí, J., Del Negro, C., 2013. QVAST: a newQuantumGIS plugin forestimating volcanic susceptibility. Nat. Hazards Earth Syst. Sci. 13 (11), 3031–3042.

Baxter, P., Cole, P., Spence, R., Neri, A., Zuccaro, G., Boyle, R., 2005. The impacts of pyroclas-tic surges on buildings at the eruption of the Soufriere hills volcano, Montserrat. Bull.Volcanol. 67, 292–313.

Becerril, L., Cappello, A., Galindo, I., Neri, M., Del Negro, C., 2013. Spatial probabilitydistribution of future volcanic eruptions at El Hierro Island (Canary Islands, Spain).J. Volcanol. Geotherm. Res. 257, 21–30.

Becerril, L., Bartolini, S., Sobradelo, R., Martí, J., Morales, J.M., Galindo, I., 2014. Long-termvolcanic hazard assessment on El Hierro (Canary Islands). Nat. Hazards Earth Syst.Sci. 14, 1853–1870.

Bertolaso, G., De Bernardinis, B., Bosi, V., Cardaci, C., Ciolli, S., Colozza, R., Cristiani, C.,Mangione, D., Ricciardi, A., Rosi, M., Scalzo, A., Soddu, P., 2009. Civil protectionpreparedness and response to the 2007 eruptive crisis of Stromboli volcano, Italy.J. Volcanol. Geotherm. Res. 182, 269277.

Blakely, R.J., Christiansen, R.L., Guffanti, M., Wells, R.E., Donnelly-Nolan, J.M., Muffler, L.J.P.,Clynne, M.A., Smith, J.G., 1997. Gravity anomalies, Quaternary vents, and Quaternaryfaults in the southern Cascade Range, Oregon and California: implications for arc andbackarc evolution. J. Geophys. Res. 102 (B10), 22513–22527.

Blong, R., McKee, C., 1995. The Rabaul Eruption 1994: Destruction of a Town. NationalHazards Research Centre, Macquarie University, Sydney.

Bonadonna, C., Macedonio, G., Sparks, R.S.J., 2002. Numerical modelling of tephra falloutassociated with dome collapses and Vulcanian explosions: application to hazardassessment in Montserrat. In: Druitt, T.H., Kokelaar, B.P. (Eds.), The Eruptionof Soufrière Hills Volcano, Montserrat, from 1995 to 1999. Geological Society ofLondon, Memoir, London, pp. 517–537.

Connor, C.B., Stamatakos, J.A., Ferrill, D.A., Hill, B.E., Ofoegbu, G.I., Conway, F.M., Sagar, B.,Trapp, J., 2000. Geologic factors controlling patterns of small-volume basalticvolcanism: application to a volcanic hazards assessment at Yucca Mountain, Nevada.J. Geophys. Res. 105, 417–432.

Connor, C.B., Hill, B.E., Winfrey, B., Franklin, N.M., LaFemina, P.C., 2001. Estimation ofvolcanic hazards from tephra fallout. Nat. Hazards Rev. 2, 33–42.

Connor, L.J., Connor, C.B., Meliksetian, K., Savov, I., 2012. Probabilistic approach to model-ing lava flow inundation: a lava flow hazard assessment for a nuclear facility inArmenia. J. Appl. Volcanol. 1 (3), 1–19.

Costa, A., Macedonio, G., Folch, A., 2006. A three-dimensional Eulerianmodel for transportand deposition of volcanic ashes. Earth Planet. Sci. Lett. 241, 634–647.

Cova, T.J., 1999. GIS in emergency management. In: Longley, P.A., Goodchild, M.F.,Maguire, D.J., Rhind, D.W. (Eds.), Geographical Information Systems: Principles,Techniques, Applications, and Management. John Wiley and Sons Ltd, New York,pp. 845–858.

Crosweller, H.S., Arora, B., Brown, S.K., Cottrell, E., Deligne, N.I., Ortiz, N., Hobbs, L.K.,Kiyosugi, K., Loughlin, S.C., Lowndes, J., Nayembil, M., Siebert, L., Sparks, R.S.J.,Takarada, S., Venzke, E., 2012. Global database on large magnitude explosive volcaniceruptions (LaMEVE). J. Appl. Volcanol. 1.

De la Cruz-Reyna, S., 1996. Long-term probabilistic analysis of future explosive eruptions.In: Scarpa, R., Tilling, R.I. (Eds.), Monitoring and Mitigation of Volcanic Hazards.Springer-Verlag, Berlin, pp. 599–629.

Domínguez Cerdeña, I., del Fresno, C., GomisMoreno, A., 2014. Seismicity Patterns Prior tothe 2011 El Hierro Eruption. Bulletin of the Seismological Society of America 104.http://dx.doi.org/10.1785/0120130200.

Felpeto, A., Martí, J., Ortiz, R., 2007. Automatic GIS-based system for volcanic hazardassessment. J. Volcanol. Geotherm. Res. 166, 106116.

Folch, A., Costa, A., Macedonio, G., 2009. FALL3D: a computational model for transport anddeposition of volcanic ash. Comput. Geosci. 35, 1334–1342.

Geyer, A., Martí, J., 2008. The new worldwide collapse caldera database (CCDB): a tool forstudying and understanding caldera processes. J. Volcanol. Geotherm. Res. 175, 334–354.

Gogu, R.C., Dietrich, V.J., Bernhard, J., Schwandner, F.M., Hurni, L., 2006. A geo-spatial datamanagement system for potentially active volcanoes-GEOWARN project. Comput.Geosci. 32, 29–41.

Gomes, A., Gaspar, J.L., Queiroz, G., 2006. Seismic vulnerability of dwellings at Sete CidadesVolcano (S. Miguel Island, Azores). Nat. Hazards Earth Syst. Sci. 6, 4148.

Johnson, B., Turnbull, K., Brown, P., Burgess, R., Dorsey, J., Baran, A.J., Webster, H.,Haywood, J., Cotton, R., Ulanowski, Z., Hesse, E., Woolley, A., Rosenberg, P., 2012.In-situ observations of volcanic ash clouds from the FAAM aircraft during the erup-tion of Eyjafjallajokull in 2010. J. Geophys. Res. 117, D00U24.

Kelfoun, K., Druitt, T.H., 2005. Numerical modeling of the emplacement of Socompa rockavalanche, Chile. J. Geophys. Res. Solid Earth 110 (B12) (19782012).

Kiyosugi, K., Connor, C.B., Zhao, D., Connor, L.J., Tanaka, K., 2010. Relationships betweenvolcano distribution, crustal structure, and P-wave tomography: an example fromthe Abu Monogenetic Volcano Group, SW Japan. Bull. Volcanol. 72, 331–340.

Lirer, L., Petrosino, P., Alberico, I., Postiglione, I., 2001. Long-term volcanic hazard forecastsbased on Somma-Vesuvio past eruptive activity. Bull. Volcanol. 63, 45–60.

Longley, P.A., Goodchild, M.F., Maguire, D.J., Rhind, D.W., 2005. Geographic InformationSystems and Science, 2nd ed. John Wiley and Sons Ltd, England, p. 517.

López, C., Blanco, M.J., Abella, R., Brenes, B., Cabrera-Rodríguez, V.M., Casas, B., Domínguez-Cerdeña, I., Felpeto, A., Fernández de Villalta, M., Del Fresno, C., García, O., García-Arias,M.J., García-Canada, L., Gomis-Moreno, A., González-Alonso, E., Guzmán-Pérez, J.,Iribarren, I., López-Díaz, R., Luengo-Oroz, N., Meletlidis, S., Moreno,M.,Moure, D., Peredade Pablo, J., Rodero, C., Romero, E., Sainz-Maza, S., Sentre-Domingo, M.A., Torres, P.A.,Trigo, P., Villasante-Marcos,M., 2012.Monitoring theunrest of ElHierro (Canary Islands)before the onset of the 2011 submarine eruption. Geophys. Res. Lett. 39.

Macedonio, G., Costa, A., Longo, A., 2005. A computer model for volcanic ash fallout andassessment of subsequent hazard. Comput. Geosci. 31 (7), 837–845.

Martí, J., Felpeto, A., 2010. Methodology for the computation of volcanic susceptibility:application to Tenerife Island (Canary Islands). J. Volcanol. Geotherm. Res. 195, 69–77.

Martí, J., Spence, R.J.S., Calogero, E., Ordonez, A., Felpeto, A., Baxter, P., 2008. Estimatingbuilding exposure and impact to volcanic hazards in Icod de los Vinos, Tenerife(Canary Islands). J. Volcanol. Geotherm. Res. 178 (3), 553–561.

Martí, J., Pinel, V., López, C., Geyer, A., Abella, R., Tárraga, M., Blanco, M.J., Castro, A.,Rodríguez, C., 2013. Causes and mechanisms of the 2011–2012 El Hierro (CanaryIslands) submarine eruption. J. Geophys. Res. Solid Earth 118 (3), 823–839.

Marzocchi, W., Sandri, L., Selva, J., 2008. BET EF: a probabilistic tool for long- andshort-term eruption forecasting. Bull. Volcanol. 70, 623–632.

Marzocchi,W., Sandri, L., Selva, J., 2010. BET VH: a probabilistic tool for long-term volcanichazard assessment. Bull. Volcanol. 72, 705–716.

Mastin, L.G., 2001. A Simple Calculator of Ballistic Trajectories for Blocks Ejected DuringVolcanic Eruptions. U.S. Geological Survey Open-File Report 01–45, p. 16.

Menoni, S., Costa, L., Galderisi, A., Margottini, C., 2011. Methodological framework for anintegrated multi-scale vulnerability and resilience assessment. ENSURE, Del 4.1.1.ENSURE Project, WP4, Deliverable 4.1.1, p. 96.

Newhall, C.G., Self, S., 1982. The volcanic explosivity index (VEI): an estimate of the explo-sive magnitude for historical eruptions. J. Geophys. Res. 87, 12311238.

Pareschi, M.T., Cavarra, L., Favalli, F., Gianni, F., Meriggi, A., 2000. GIS and volcanic riskmanagement. Nat. Hazards 21, 361379.

Patra, A.K., Bursikb, M., Dehne, J., Jonesc, M., Madankana, R., Mortone, D., Pavolonisf, M.,Pitmand, E.B., Pougetb, S., Singha, T., Singlaa, P., Stefanescua, E.R., Webleye, P., 2013.Challenges in developing DDDAS basedmethodology for volcanic ash hazard analysiseffect of numerical weather prediction variability and parameter estimation. ProcediaComput. Sci. 18, 1871–1880.

Phillipson, G., Sobradelo, R., Gottsmann, J., 2013. Global volcanic unrest in the 21st century:an analysis of the first decade. J. Volcanol. Geotherm. Res. 264, 183–196.

Pyle, D.M., 2000. Sizes of volcanic eruptions. In: Sigurdsson, H. (Ed.), Encyclopedia ofVolcanoes. Academic, San Diego (California), p. 263269.

Rout, D.J., Cassidy, J., Locke, C.A., Smith, I.E., 1993. Geophysical evidence for temporal andstructural relationships within the monogenetic basalt volcanoes of the Aucklandvolcanic field, northern New Zealand. J. Volcanol. Geotherm. Res. 57 (1), 71–83.

Scaini, C., Felpeto, A., Martí, J., Carniel, R., 2014. A GIS-based methodology for theestimation of potential volcanic damage and its application to Tenerife Island,Spain. J. Volcanol. Geotherm. Res. 278–279, 40–58.

Sheridan, M.F., Stinton, A.J., Patra, A., Pitman, E.B., Bauer, A., Nichita, C.C., 2005. EvaluatingTitan2D mass-flow model using the 1963 Little Tahoma Peak avalanches, MountRainier, Washington. J. Volcanol. Geophys. Res. 139, 89–102.

Sobradelo, R., Bartolini, S., Martí, J., 2014. HASSET: a probability event tree tool to valuatefuture volcanic scenarios using Bayesian inference presented as a plugin for QGIS.Bull. Volcanol. 76, 770.

Spence, R.J.S., 2004. Risk and regulation: can improved government action reduce theimpacts of natural disasters? Build. Res. Inf. 32 (5), 391–402.

Spence, R.J.S., Kelman, I., Calogero, E., Toyos, G., Baxter, P., Komorowski, J.C., 2005.Modelling expected physical impacts and human casualties from explosive volcaniceruptions. Nat. Hazards Earth Syst. Sci. 5, 1003–1015.

Spence, R.J.S., Gunesekara, R., Zuccaro, G., 2009. Insurance Risks from Volcanic Eruptionsin Europe. Willis Research Network, London, United Kingdom, pp. 1–26.

UNISDR, 2009. UNISDR Terminology of Disaster Risk Reduction, (Geneva, Switzerland).Venezky, D.Y., Newhall, C.G., 2007. WOVOdat Design Document: The Schema,

Table Descriptions, and Create Table Statements for the Database of WorldwideVolcanic Unrest (WOVOdat Version 1.0). US Geological Survey Open-File Report2007–1117, p. 184.

Witham, C., 2005. Volcanic disasters and incidents: a new database. J. Volcanol. Geotherm.Res. 148, 191–233.

Zuccaro, G., Cacace, F., Spence, R., Baxter, P., 2008. Impact of explosive eruption scenariosat Vesuvius. J. Volcanol. Geotherm. Res. 178 (3), 416–453.