Požari-u-Dalmac.

7/29/2019 Poari-u-Dalmac.

1/21

Fire is the uncontrolled movement of flames overthe surface of the ground. It varies in type, origin anddegree of damage. For fire to occur, a certain temperat-ure, flammable mass and oxygen are required. If one ofthese is eliminated, the fire is quenched. Fires are a ma-

jor global problem; over ten million hectares of forestsarelost globally every year (Mladineo, 2001).

In the summer months, the Split-Dalmatia County is

highly exposed to the risk of fire. In addition to thedamagecausedbyfireandthecostofdealingwithit,isaparticular concern when people are directly or indir-ectly endangered. We need to know the dangerous cir-cumstances which lead to open fire, and what securitymeasures should be taken to reduce the risk as far aspossible. In addition to fire prevention measures, theonly effective way to reduce the damage caused by fireis to detect potential outbreaks and respond with rapid,appropriate intervention.

The degree of fire hazard depends on how well

equipped fire-fighting groups are to respond, the in-tensity of the flames, the terrain, vegetation, weather

and other conditions. In this paper, we will show that itis possible to analyze futurefire risk situations, based onpastdata.

The aim of this paper is to create an analysis of thefire risk for 2011 based on data from previous fire sea-sons in the Split-Dalmatia County. GIS has proved to bethemost appropriate tool for analyzing process overlap,and the results are presented cartographically. Thisanalysis has been created entirely in ArcInfo. It was car-ried out accordingto thefollowingsteps:

1. After reviewing the literature and previous researchon this topic, the most important factors in the out-break of fire were selected. The availability of datawas taken into account. Analyzed data on temperat-ure and humidity were obtained from the MHI (Met-eorological and Hydrological Institute). Data onvegetation were taken from theEPA(EnvironmentalProtection Agency) and downloaded from satelliteimages from NASA Terra satellite MODIS, data on re-lief were obtained from the SGA (State Geodetic Ad-ministration), while anthropological data were

obtained from the NPRD (National Protection andRescue Directorate), and are based on CRSU (Central


2/21

Poar je nekontrolirano, stihijsko kretanje vatre popovrini. Razlikuje se po vrsti, nainu postanka i teta-ma. Za nastanak poara potrebna je odreena tempera-tura, goriva masa i kisik; ako se jedno od toga ukloni,poar prestaje. Poari su veliki svjetski problem, jer se unjima godinje u svijetu izgubi vie od deset milijunahektara uma (Mladineo, 2001).

Podruje Splitsko-dalmatinske upanije u ljetnim

mjesecima izrazito je izloeno velikim opasnostima odizbijanja poara. Osim teta i trokova posebno je zabri-njavajue to stradavaju ljudi bilo oni izravno angaira-ni u gaenje poara ili oni koji su stjecajem okolnosti biliugroeni. U svakom pogledu treba imati znanja o opas-nimokolnostima u poaruotvorenogtipa, odnosno kojemjere sigurnosti treba poduzeti kako bi se opasnost sve-la na to manju mjeru. Osim preventivnih protupoar-nih mjera, jedini efikasan nain smanjenja tete kojuuzrokuju poari otvorenog prostora je pravovremenouoavanje poara u nastajanju, te brza i odgovarajua

intervencija.

Stupanj ili razina opasnosti prilikom poara ovisi osnazi vatrogasne grupe, intenzitetu gorenja, reljefnomizgledu terena, sastavuraslinja,meteorolokimi drugimuvjetima. Zatoe se u ovomradupokazatikakose moena temelju dosadanjih podataka napraviti analiza rizi-kaodpoarazabudue situacije.

Svrha je ovog rada da se uz pomo dosadanjih po-dataka o prostoru Splitsko-dalmatinske upanije napra-vi analiza rizika od poara. GIS se pokazao kao naj-primjereniji alat za analizu postupcima preklapanja, te

su rezultati prikazani kartografski. Analiza je u potpu-nosti napravljena u ArcInfou. Analiza je provedena usljedeim koracima:1. Nakon uvida u literaturu i dosadanja istraivanjana

tu temu, odreeni su imbenici koji su najvaniji priizbijanju poara. Trebalo je voditi rauna i o dostup-nosti podataka.Analizirani su podaci o temperaturi ivlanosti zraka dobiveni od DHMZ-a (Dravni hidro-meteoroloki zavod), podaci o vegetaciji dobiveni odAZO-a (Agencija za zatitu okolia) i preuzeti sa sate-litskih snimaka NASA-inog satelita Modis Terra, po-

daci o reljefu dobiveni su od DGU-a (Dravna


3/21

Register of Spatial Units) and CBS (Central Bureau of

Statistics) data.2. Using the DC-log (National Centre) web application,

we analyzed data from the fire season between June1 and October 1 2010. The data were georeferenced.Fire threat categories were created (in which 1 cor-respondedtoaverylowriskand5averyhighriskoffire).

3. Finally, these data were synthesised to create a finalmap of the risk of fire in the Split-Dalmatia County.The importance of each factor inthe outbreak of firewas weighted. How the procedure for processingeach of the elements included in the analysisdiffered will be explainedin detail later.

In geographical terms, the Split-Dalmatia County in-cludes a group of large islands (Bra, Hvar, Vis, andolta) and some smaller ones, the coastal flysch zoneandthehinterland.

Coastal flysch zone is the most appropriate area for

agricultural activities is the coastal flysch zone, there-fore the earliest villages developed here, rather than bythe sea, in elevated situations, where the flysch meetsthe limestone slopes of the coastal mountains. Sincemost of the population is focussed in the coastal area,and in the summer months a large number of touristsarrive, there is anincreased risk of fire.

Sub-Mediterranean Hinterland is a predominantlymountainous and sparse Karst area, where small-scaleanimal husbandry is common. The main agriculturalactivities take place in well cultivated inhabited areas

viticulture, growing maize, vegetables and tobacco. Thegreatest risk to this region in terms of the outbreak of

fire is represented by abandoned agricultural land

which is no longerworked.Central Dalmatian Islands includes a group of large is-

lands (Bra, Hvar, Vis and olta) and some smaller ones.The islands of Central Dalmatia are more densely popu-lated than the Kvarner islands. Agricultural productionand small-scale animal husbandry are predominant.There is a greater risk of fire during the summermonths, due to the large number of tourists visiting theislands.

When analyzing the risk of fire, 9 factors were selec-ted as being the most important in the outbreak of fire(Fig. 1).

The most important climate indicators to take intoconsideration are temperature and relative humidity.Dry and wet periods of the year are of special import-

ance for fire prevention and the recovery of burnedareas. The Split-Dalmatia County climate is moderatelywarm and humid with hot summers (Cfa), while thecoastal area and islands have a Mediterranean climatewith hot, dry summers (Csa). Hot summers are the res-ult of the intense daily warming of the low relief, whichis mostly bare, while the soil is mostly porous and dry(egota, 1996).

Inorder tocreatea firethreatmap,temperature andrelative humidity data from the MHI (Meteorologicaland Hydrological Institute) was collected from measur-

ing stations in ibenik, Split and Makarska and on Hvarand Vis and processed, With the help of Analysis tool


4/21

geodetska uprava), dok su antropoloki podaci

dobiveni od DUZS-a (Dravna uprava za zatitu ispaavanje), a bazirani su na SRPJ-u (Sredinji regis-tar prostornih jedinica) i podacima DZS-a (Dravnizavod za statistiku).

2. Koritenjem web aplikacije dnevnik DC-a (Dravnogcentra) obraeni su podaci iz poarne sezone od 1.lipnja do 1. listopada 2010. Podaci su georeferencira-ni. Napravljene su kategorije ugroenosti od izbija-nja poara (tako da je 1- vrlo niska opasnost, 5- vrlovisoka opasnost od izbijanjapoara).

3. Nasamom kraju, tipodaci supreklopljeni radi izradekonane karte rizika od poara na podruju Splitsko-dalmatinske upanije. Vanost svakog imbenika priizbijanju poara je ponderirana. Kako je postupakobrade podataka drugaiji za svaki od elemenataukljuenih u analizu, oni e se posebno objasniti da-lje u tekstu.

Splitsko-dalmatinska upanija prirodno-geografski

obuhvaa skupinu velikih (Bra

, Hvar, Vis, olta) i ma-njihotoka, obalnu flinu zonu i Zagoru.

Za poljoprivredno iskoritavanje najpogodnija jeobalna flina zonapa su se najstarija seoskanaselja razvilau toj zoni, ne uz more, nego na viem poloaju, na kon-taktu flia i vapnenakih padina primorskih planina.Kako je u obalnoj zoni teite naseljenosti, te u ljetnimmjesecima dolazi velik broj turista, poveava se rizik odpoara.

Submediteranska zagora je preteno brdovit i siroma-an krki kraj s tradicionalnim sitnim stoarstvom;

glavne su ratarske djelatnosti u dobro obraenim i na-seljenim poljima vinogradarstvo, uzgoj kukuruza i

povra te duhana. Najveu opasnost u toj regiji od izbi-

janja poara predstavljaju naputena poljoprivrednazemljita koja se vie ne obrauju.

Srednjodalmatinski otoci obuhvaaju skupinu velikih(Bra, Hvar, Vis, olta) i manjih otoka. Otoci srednje Dal-macije gue su naseljeni od kvarnerskih otoka. Nasrednjodalmatinskim otocima prevladava poljoprivre-dna proizvodnja te sitno stoarstvo. Ljetni mjeseci sunajugroeniji zbog velikog broja turista koji dolaze naotoke.

Pri analizi rizika od poara odabrano je 9 imbenikakoji su najvaniji pri izbijanju poara otvorenog tipa(slika1).

Najvaniji klimatski pokazatelji koji se uzimaju urazmatranje su temperatura i relativna vlanost zraka.Suna i vlana razdoblja u godini posebno su vana za

vatrogasnu preventivu i sanaciju izgorjelepovrine.Klima Splitsko-dalmatinske upanije umjereno jetopla i vlana s vruim ljetom (Cfa), dok priobalni dio iotoci imaju sredozemnu klimu sa suhim vruim ljetom(Csa). Vrua ljeta uvjetovana su jakim dnevnim zagrija-vanjem reljefno niskoga kraja, koji je uz to jo i ogolio, atlo jenajee propusno i suho (egota,1996).

Za izradu karte ugroenosti obraeni su podaci odDHMZ-a za temperaturu i relativnu vlanost zraka smjernih postaja ibenik, Split, Makarska, Hvar, Komia.Uz pomo Analysis toola (proximity; point distance) izrau-

nata je udaljenost svakog poara do najblie mjernepostaje.S obzirom natodase zna tono vrijeme izbijanja


5/21

(proximity; point distance) the distance of each firefrom the nearest measurement station was calculated.Thus, since we knew the time of each fire, we could as-sign it a temperature and relative humidity value (forthe exact time and the nearest measurement station).This was done because there were not enough monitor-ing stations for a more precise analysis. In this way, weobtained the exact values for each fire location, andtransferred them from a point feature to a polygon featureusing the Kryging method (using existing data to ap-

proximatean unknown value at a particular location).

Air temperature is one of the most important cli-mate elements. Air temperature is the degree to whichairis heated. Airtemperature is lowest just after sunrise,and highest just after its zenith. In Croatia, this isbetween 13.00 and 14.00. In the Split-Dalmatia County,high summer temperatures are predominant. Hightemperatures dryout thesoil andvegetation,sothat the

area is more at risk.

Using the Kriging method, we produced a map ofthe average temperatures for the Split-DalmatiaCounty. The data were transferred to a raster in orderto continue the analysis (Fig. 2.). How each fire wasgeoreferenced is shown in the histogram (Fig. 2),which shows the temperatures at which most firesbreak out. This information was important because itallowed us to identify risk zones for each element. El-even fires were recorded at a temperatureof 25C, and10 fires at a temperature of 32C. We noted that fires

were recorded at several different temperatures, in-dicating that fires occur independently of temperat-ure; nonetheless their number is dependent ontemperature increases. Once temperatures had beendivided into classes, the reclassification of data andaddition of new values was possible (histogram Fig.2). These values were reclassified using 3D Analysttools (raster reclass; reclassify) so that the class inwhich the most fires were recorded was given thehighest value (5), and the class in which the least fireswere recorded the lowest value (1) (Fig. 3).


6/21

poara, mogli smo svakom poaru pridodati vrijednosttemperature i relativne vlanosti zraka za tono to vrije-me i za najbliu mjernu postaju. To je napravljeno zbogtoga jer je premalo mjernih postaja za bilo kakvukvalitetniju analizu. Na taj nain dobili smo egzaktne vri-

jednosti za svaku lokaciju poara, a te vrijednosti preba-ene su iz point feature u polygon feature metodomkrigiranja (koja uz pomo postojeih podataka aproksi-mira nepoznatevrijednosti na nekoj lokaciji).

Temperatura zraka jedan je od najvanijih kli-matskih elemenata. Moe se rei da je temperaturazraka stupanj njegove zagrijanosti. Temperatura jenajnia neto nakon izlaza sunca, a najvia neto na-kon njegova najvieg poloaja. Kod nas je to izmeu13 i 14 sati. Na podruju Splitsko-dalmatinske upani-je prevladavaju visoke ljetne temperature, koje isuu-ju tloi vegetaciju tako da je prostor vie izloen riziku.

Metodom krigiranja dobili smo kartu prosjene

temperature za Splitsko-dalmatinsku upaniju. Tesmo podatke prebacili u raster da bismo dalje mogli

raditi analizu (slika 2). Kako je svaki poar georefe-renciran, iz histograma (slika 2) se moe vidjeti prikojim temperaturama zraka je izbilo najvie poara.Taj podatak nam je bitan zato to e se na temelju to-ga napraviti zone ugroenosti za svaki od elemenata.ak 11 poara je zabiljeeno pri temperaturi od 25C,te 10 poara pri temperaturi od 32C. Vidimo da supoari zabiljeeni pri svakoj temperaturi, to znai dase javljaju neovisno o temperaturi, ali da njihov brojipak ovisi o porastu temperature. Nakon to je na-

pravljena podjela temperature u razrede, moe seobaviti reklasifikacija podataka i pridodavanje novihvrijednosti (histogram slika 2). Te vrijednosti rekla-sificirane su s alatima 3D Analysta (raster reclass; recla-

sify) takodaonaj razred u kojem jezabiljeeno najviepoara ima vrijednost 5, a onaj razred u kojem je naj-manje poara ima vrijednost 1 (slika 3).

Relativna vlanost zraka je broj koji u postocima

pokazuje odnos izmeu koliine vodene pare kojastvarno postoji u zraku (u odreenom trenutku) i


7/21

Relative humidity is a number expressed as a per-centage that indicates the relationship between theamount of water vapour actually present in the air (at agiven time) and the maximum amount of water vapourthat the air at that temperature could absorb before be-coming saturated. If the relative humidity is near 0%,the air is dry, and if it is close to 100%, the air is humid.Relative humidity is greater in winter than in summer,

and greater onislands thanon the coast.Relative humidity is an important factor in affectingthe occurrence and spread of fire, and it manifests itselfin the form of air humidity, ground moistness and themoistness of combustible materials. If combustible ma-terials are dry, they will burn faster and reach a temper-ature of 200C swiftly. If air humidity is low, the dry airabsorbs more easily water vapours from combustiblematerial and the ground where the fire is burning.

The process of making maps is the same as for airtemperatures. The histogram shows that when the rel-

ative humidity is low, there are more reported fires (Fig.4).These data were also reclassified(Fig. 5).

Analysis of the relief was based on a digital heightmodel (x, y, z). Given the large number of points, theclassic production of TIN (Triangulated Irregular Net-work) models was not possible. Instead, a model wasbuilt using 3D Analyst tools; Terrain (create terrain, addthe feature class to terrain, build terrain). Subsequently,these data were transferred to raster (3D Analyst tool;conversion; from terrain, terrain to raster). From that

format, wecreated aspect and slope in3DAnalyst.

Aspect has a significant influence on the occurrenceof fire because microclimate and vegetation elementsare determined by the points of the compass. North-fa-cing slopes are much less exposed to sunlight. South-and southwest-facing slopes are exposed to sunlight forlonger periods, and xerophyte plants thrive on suchslopes. It is known that dry habitats and the plant struc-

ture in such environments favours the occurrence offire, and it should be emphasised outthat the sun's heat


8/21

maksimalne koliine vodene pare koju bi zrak na tojtemperaturi mogao primiti da bi bio zasien. to je re-lativnavlanostzrakablie0%toznaidajezraksui,aako je blie 100% znai da je zrak vlaniji. Relativnavlanost zraka vea je zimi nego ljeti, vea je na otoci-manegonaobali.

Relativna vlanost zraka je bitan imbenik kojiodluuje o mogunosti nastanka i irenja poara, ajavlja se u obliku zrane vlage, vlanosti stanita ivlanosti gorivog materijala. Ako je gorivi materijal

suh, gori bre i bre se postie temperatura od 200C.Ako je zrana vlanost manja, suhi zrak lake upijaoslobaajuu vodenu paru iz gorivog materijala i iztlana kojemutraje procesgorenja.

Proces izrade karte isti je kao i kod temperaturezraka. Iz histograma je vidljivo da to je relativna vla-nost zraka manja, vie je zabiljeenih poara (slika 4).Ti podacisutakoerreklasificirani(slika5).

Analiza reljefa bazirana je na digitalnome modeluvisina (x, y, z). S obzirom na veliki broj toaka, klasina

izrada TIN (nepravilna mrea trokuta) modela nije bilamogua. Umjesto toga napravljen je model s alatom 3DAnalysta Terrain (create terrain; add feature class to terrain;buildterrain). Nakon toga, ti podaci su prebaeni u raster(3D Analyst tool; conversion; from terrain; terrain toraster). Iz tog formata dalje se izrauju u 3D Analystuorijentiranost i nagibpadine.

Orijentiranost padine znatno utjee na pojavu po-ara jer su mikroklimatski i vegetacijski elementi

odreeni stranama svijeta. Sjeverne padine puno sumanje izloene sunevu zraenju. June i jugoza-padne padine due su vremena izloene sunevuzraenju, pa se na tim ekspozicijama razvijaju ksero-fitne biljke. Poznato je da suha stanita i biljna struk-tura na takvim tlima pogoduje pojavi poara, a uz totrebanaglasitidasunevatoplina tijekom dana utjeena pojavu vjetra koji pue iz doline prema vrhu. Kaoto je vidljivo iz slike 6, najvie poara zabiljeeno je

na jugoistonim, junim i jugozapadnim padinamazato to su te padine pod jaim utjecajem suneva


9/21

during the day causes winds to arise, blowing from thelower towards the higher ground. As shown in Fig. 6,most fires were recorded on southeastern, southernandwestern slopes. These slopes are under the strong influ-ence of solar radiation, as well as having different ve-getation, and the direction they face makes them dryerthan others. With reclassification, we obtained fire riskzones (Fig. 7).

Hills with steep slopes facilitate the faster spread offire. On steeper ground, fire rapidly progresses up aslope, because warm air rises, drying combustible ma-terial as it goes. It is rare for fire tospread downa slope(in such cases, it must be fuelled by a strong, downwardwind). A fire that breaks out at the foot of a slope, whenthere is no wind, causes currents of warm and cold airand the fire spreads by advancing up the slope. Al-though slope plays a more important part in the actualspread of fire, rather than its outbreak, according to the

histogram results it is still included in the analysis be-cause of the interesting data (Fig. 8) recorded. We can

see from the results obtained in the histogram that anumber of fires occur in areas of less inclination, and asthe angle of inclination increases, the number of firesrapidly decreases. An estimated 40% of fires were recor-ded on slopes between 04. This can be linked to thefact that a large number of fires broke outon agricultur-al land, i.e. on flat surfaces. They were probably the res-ult of burning weeds inthe afternoon.

With reclassification we obtained risk zones (Fig. 9).

The most important vegetative parameters usedwere the CORINE (Coordination of Information on theEnvironment) Land Cover and Normalized DifferenceVegetation Index(hereinafter: NDVI).

CORINE is a programme for the coordination of in-formation about the environment and natural re-

sources, launched by the European Community. CLC isthe identificationand meaningfulcategorizationof land


10/21

zraenja. Uz razliku u vegetaciji, takoerjebitnodajeta strana padine sua od ostalih jer je dui dio danaosunana. Reklasifikacijom smo dobili zone ugroe-nosti od poara (slika 7).

Strmi nagib padine pogoduje brem irenju poa-ra. Na nagnutom terenu vatra bre napreduje uz pa-dinu, jer se topli zrak penje i pred sobom isuuje

gorivi materijal. Rijetka je pojava da se poar iri nizpadinu (tada je poar potpomognut jakim vjetromkoji pue niz padinu). Poar koji izbije u podnoju pa-dine i u vrijeme bezvjetra, gorenjem izaziva strujanjetoplog i hladnog zraka pa irenje poara napreduje uzpadinu.

Iako je nagib vaniji pri samom irenju poara,prema rezultatima histograma ipak je ukljuen u an-alizu radi zanimljivih podataka (slika 8). Iz dobivenihrezultatau histogramu vidimoda se broj poarajavljana prostorima manjeg nagiba, te da se poveanjem

tognagibabroj poarabrzo smanjuje.ak40% poarazabiljeeno je na nagibu izmeu 0 i 4. To moemo

povezati s time da je veliki broj poara izbio na poljo-privrednim zemljitima koja su na zaravnjenim povr-inama, vjerojatno kao posljedica potpaljivanja koro-va u poslijepodnevnim satima. Reklasifikacijom sudobivene zone ugroenosti (slika9).

Najvaniji vegetacijski pokazatelji kojima smo se ko-ristili su CORINE (COoRdination of INformation on the

Environment) Land Cover i Normalized Difference Ve-getation Index normirani indeks razlike u vegetaciji(daljeu tekstuNDVI).

CORINE je program za koordinaciju informacija ookoliu i prirodnim resursima pokrenut od Europskezajednice. CLC je identifikacija i smislena kategoriza-cija pokrova zemljita, koja ukljuuje definiranu no-menklaturu kodiranja i stvaranja kvalitetne baze

podataka, potrebne za nadgledanje prirodnih resur-sa, njihovo organiziranje i upravljanje njima na


11/21


12/21


13/21

cover, which includes defined nomenclature and cod-ing, to create a high-quality database for monitoring,organizing and managing natural resources at the re-gional and national levels. Data on the status of landcover provide new insights into changes in natural re-sources in different fields, such as agriculture, forestry,

regional spatial planning, the inventorisation of naturalresources and environmental monitoring. With thecreation of CORINE Land Cover, data on the actual landcover for Croatia, based on standards applied in Europe,was obtained.

It is evident from the histograms (Fig. 10) that mostfires recorded broke out on agricultural land, growingplots, meadows, transitional forest areas, undergrowthand garrigue. Through reclassification into threatenedareas, it is evident that the Split-Dalmatia County is avery large area, highly compromised in terms of veget-ation (upto73%) (Fig. 11).

NDVI is based on the reflection properties of veget-ated areas in comparison to clouds, water and snow onthe one hand, and rocks and barren land on the other.Vegetated areas have relatively high reflectance in thenear-infrared section of the spectrum, and low reflect-ance in the visible section. Clouds, water and snow havea higher reflectance in the visible section than in thenear-infrared section of the spectrum. Rock and bareland have an equal degree of reflectance in both sec-tions of the spectrum. NDVI can be used to determinethe presence of vegetation and its condition (wet or

dry). In this analysis, NDVI was used only for determin-ingthepresence of vegetation.

NDVI wascalculated using theformula:

NDVI = (NIRVS)/ (NIR+VS),

in which NIR is the value of reflectance in the near-in-frared section of the spectrum, and VS the value of re-

flectance in the visible part of the spectrum. Thecalculation of NDVI for a pixel is always in the range ofvalues from 1to+1, however,green plants donot give avalue of 0.Zeromeans no vegetation, and as the value +1is approached, the area is greener and richer in vegeta-tion,indicating a high density inthetree canopy.

For our analysis, we used a photograph taken by theModis Terra satellite of the Rome tor Vergeta zone, res-olution 250 m, in GeoTiff format. Since it was impossibleto conduct the analysis throughout 36 days, because wedid not have the personnel or computing power, the

analysis was conducted as follows. The date chosen wasthat on which we recorded the highest number of firesduring the season. This date was 22 August, when therewere five fires in the Split-Dalmatia County. Since foot-age comes in 3 bands (RGB), we first needed to transfervalues to NDVI values. This was done by exporting eachband separately in grid format, and then reloading andreclassifying RGB values according to Table 1. In thisway, we obtained the value of NDVI for the Split-Dalma-tia County and could proceed to the analysis of histo-grams (Fig. 12).

Red Green Blue NDVI153 204 255 0.0225 175 100 0.05255 225 150 0.15225 255 175 0.25152 255 152 0.35

102 255 102 0.4551 204 51 0.550 153 0 0.650 102 0 >0.75255 255 255 No data

As expected, when the NDVI value was high, thenumber of fires in the area was larger. A high NDVIvalue indicates more vegetation, so it was clear that inareas with lower NDVI values, there could also be fires,in these cases consuming undergrowth and garrigue.

Reclassification of the data was calculated and the firerisk wascalculated accordingto NDVI values(Fig. 13).


14/21

regionalnoj i nacionalnoj razini. Podaci o stanju po-krova zemljita daju novi uvid u stanje i promjeneprirodnih resursa na razliitim poljima poput poljo-privrede, umarstva, lokalnog prostornog planiranja,inventarizacijeprirodnihresursa i praenja okolia.

Izradom CORINE Land Covera dobiveni su podacio stanju stvarnog pokrova zemljita za Hrvatsku kojise temelje na standardima primijenjenim u Europi. Izhistograma (slika 10) je vidljivo da je najvie poarazabiljeeno na poljoprivrednim zemljitima, uzgoj-

nim parcelama, travnjacima, prijelaznim umskimpodrujima, makiji i garigu. Reklasifikacijom u zoneugroenosti vidljivo je da je Splitsko-dalmatinska u-panija jako velik prostor koji je vrlo ugroen to se ti-e vegetacije (ak 73%) (slika 11).

NDVI se temelji na svojstvima refleksije podrujate njegove krajnje vrijednosti ukazuju na prisutnostoblaka, vode i snijega s jedne strane, te stijena i golog

zemljita s druge strane. Podruja pod vegetacijomimaju relativno visoku refleksiju u bliskom infracrve-

nom dijelu spektra, a nisku u vidljivom. Oblaci, vode isnijeg imaju viu refleksiju u vizualnom nego u bli-skom infracrvenom dijelu spektra. Stijene i gola zem-ljita imaju podjednaki stupanj refleksije u oba dijelaspektra. NDVI se moe koristiti da se utvrdi prisut-nost vegetacije i njezino stanje(vlana/suha). Priovojanalizi koriten je samo za utvrivanje prisutnostivegetacije.

NDVIserauna po formuli:

NDVI= (BICVID)/ (BIC+VID),

gdje je BIC vrijednost refleksije u bliskom infracrve-nom dijelu spektra, a VID vrijednost refleksije u vid-ljivom dijelu spektra. Proraun za NDVI za odreenipiksel uvijek je u rasponu vrijednosti od 1 do +1, me-utimzelene biljke ne dajuvrijednost0. Nulaznai danema vegetacije, i to se vie pribliava vrijednosti +1to je povrina zelenija i bogatija vegetacijom, te poka-zuje velikugustoulia u kronjama.

Za potrebe analize preuzeli smosnimkusa satelita

Modis Terra, iz Rome tor Vergeta zone, rezolucije250m, u formatu GeoTiff. S obzirom na to da je bilo


15/21

Fires are closely linked to the proximity of roads,railwaysandtowns.Deliberate or accidental, thehumanfactor is to blame in most cases of outbreaks of fire(Nodilo, 2003). In creating these zones, Analysis Tools(Proximity, Buffer) wereused.

Roads are an important factor in the analysis, be-cause the transport network is extremely dense, whichin the case of fire is unfavourable. The human factor isthe most important in the outbreak of fire (Nodilo,2003), and the dense network of roads means that, dur-ing their daily travels, people sometimes throw cigar-ette ends from their vehicles, and this often causes fires.Looking atthe histogram (Fig.14) wesee thatmorethan50% of fires broke out within350mof a road.

Before developing risk zones, it was necessary tocreate a union of all values in a buffer (using an editor),

and then clip it toget a compact polygon withzones forreclassification. First, however, it was transferred from

the vector into raster (conversion tools; to raster, poly-gonto raster).

Given the fact that the railroad in the Split-DalmatiaCountycrosses onlya small partofthe terrain, itdid notmake sense to analyze the influence of railways on allfires. With the help of a histogram, it was determinedthatthe analysis would include fires within10 kmof the

railway line (Fig. 16). In the next step, we created a 10km radius buffer and showed the results in the histo-gram (Fig. 17). From this histogram we can see that intheimmediatevicinity of therailway, sixfires broke out.These data were transferred to raster and reclassifiedinto5 fire risk zones (Fig. 18).

As a final factor in creating the fire risk map, the dis-tance from settlements was considered. It is evident

from the histograms (Fig. 19) that the number of firesdecreases as the distance from settlements increases. It


16/21

nemogue provesti analizu za svih 36 dana jer nismoimali ni ljudstvo, ni raunalnu snagu, analizu smoproveli na sljedei nain. Datum je odabran po tometo je na taj dan zabiljeeno najvie poara u sezoni.Dana 22. kolovoza zabiljeeno je 5 poara u Splitsko-dalmatinskoj upaniji. Kako snimka dolazi u 3 pojasa(RGB), prvo je trebalo prebaciti te vrijednosti u vri-jednosti NDVI-a, i to tako da se prvo svaki pojas po-sebno eksportirao u format grid, te ih se uitalo ireklasificiralo po vrijednostima RGB-a prema tablici

1. Na taj nain dobivena je vrijednost NDVI-a za Split-sko-dalmatinsku upaniju i moglo se krenuti u anali-

zu histograma (slika12).

Kaotoseoekivalo, toje vrijednost NDVI-avia, toje broj poara na tom prostoru vei. Visoke vrijednostiNDVI-a znae obilatu vegetaciju, tako da je jasno da i namjestima s niom vrijednou NDVI-a moe biti poara,

jer to znai da gori makija i garig. Reklasifikacijom po-dataka izraunata je zona rizikaod poaraza vrijednostiNDVI-a(slika13).

Red Green Blue NDVI153 204 255 0,0225 175 100 0,05255 225 150 0,15225 255 175 0,25152 255 152 0,35102 255 102 0,45

51 204 51 0,550 153 0 0,650 102 0 >0,75255 255 255 Bez podataka

Poari otvorenog prostora usko su povezani s bli-zinom cesta, eljeznikih pruga i naselja. Svjesno ilinesvjesno ljudi su najvei uzronici izbijanja poara(Nodilo, 2003).Pri izradi ovih zona koriten jeAnalysis

tools (proximity; buffer).


17/21


18/21


19/21


20/21

Ceste su bitan faktor pri analizi zato to je promet-na mrea iznimno gusta, to u ovom sluaju nije po-eljno. Kako su ljudi najvaniji imbenik pri izbijanjupoara (Nodilo, 2003), gusta mrea prometnica znai

da pri svojim dnevnim migracijama ljudi bacanjemopuaka iz vozila esto uzrokuju poare. Na histogra-mu (slika 14) moemo vidjeti da je vie od 50% poaraizbilounutar 350m od prometnice.

Prije izrade zona ugroenosti, bilo je potrebnoureivanjem spojiti (union) sve vrijednosti buffera ujedan, te nakon toga izrezivanjem dobiti jedan kom-paktan poligon sa zonama koje e se zatim reklasifici-rati. No prije toga podaci su promijenjeni iz vektora uraster (conversion tools;to raster; polygon to raster).

S obzirom na to da eljeznika pruga Splitsko-dal-matinskom upanijom prolazi samo malim dijelom,nije imalo smisla provoditi analizu utjecaja pruge nasve poare, nego na samo jedan dio. Uz pomo histo-grama odreeno je da se radi analiza unutar 10 kmudaljenosti od eljeznike pruge (slika 16). U iduemkoraku napravljen je koridor (buffer) od pruge radijusa10 km i prikazan je histogram (slika 17) iz kojega semoe vidjeti da je u neposrednoj blizini eljeznice izbi-lo 6 poara. Ti podaci prebaeni su u raster i reklasifici-rani u 5 zonaugroenosti odizbijanja poara(slika 18).

Kao posljednji imbenik pri izradi karte zona rizi-ka od poara uzeta je udaljenost od naselja. Vidljivo jeiz histograma (slika 19) da se udaljavanjem od naseljasmanjuje broj poara. Ve je utvreno da je ovjekglavni uzronik poara, tako da je logino da se uda-

ljavanjem od njegova prebivalita broj poara sma-njuje. Pri izradi te procjene izraeni su koridori(buffer) oko naselja, i s obzirom na to da se na nekimmjestima preklapaju, uz pomo ureivanja (union,clip) taj problem je rijeen, kreiran je jedan poligon sazonama. Nakon to je prebaen u rasterski format,vrijednosti su reklasificirane i prikazane su zoneugroenosti (slika 20).

Kako bismo dobilizone ugroenosti od poara biloje potrebno sve imbenike koji su analizirani i rekla-

sificirani preklopiti. No prije toga trebalo ih jerangirati po vanosti pri samom izbijanjupoara.

U tu svrhu opet su posluili histogrami. Iz sva-kog histograma proitan je najvei broj poaraunutar nekog stupca. Sve vrijednosti svih imbe-nika su zbrojene, te je svaka vrijednost podijeljena

s decimalnom vrijednou tog zbroja da bi se do-bila vanost svakog imbenika u postotnom udje-lu (tablica 2).

imbenik Broj poara Rizik u %Temperatura zraka 11 6Relativna vlanost zraka 18 10Orijentiranostpadine 15 8

Nagib padine 20 11CORINE Land Cover 34 18NDVI 18 10Prometnice (ceste) 39 21eljeznika pruga 16 8Naselja 16 8Ukupno 187 100

Na taj nain minimalizirana je subjektivnaprocjena. U Spatial analyst tools (overlay; weightedoverlay) pridodani su svi imbenici (reklasificirani)te im se na temelju tablice 2 dodao postotni udiovanosti. To je bitno zato to svaki imbenik nemajednaku vanost pri izbijanjupoara.

Zoneugroenosti Brojpoara Povrina(u%)Zona1 (vrloniskaopasnost) 0 0,0Zona 2 (niska opasnost) 2 4,2Zona3 (srednjaopasnost) 24 51,7

Zona4(visokaopasnost) 40 44,0Zona5 (vrlo visokaopasnost) 0 0,1

Dobiveni rezultati (slika 21) usporeeni su s ge-oreferenciranim poarima iz sezone 2010. na teme-lju kojih se i radila procjena, te se iz tablice 3 moevidjeti da je najvie poara (ak njih 40) unutar zonerizika 4 (velika opasnost od izbijanja poara). Na te-melju toga moemo zakljuiti da je takva analizakvalitetna i da se na taj nain moe raditi procjenaugroenosti za svaki dan, s time da bi se vremenski

imbenici izmjenjivaliza svaki dan. Time bi se dobilai preciznijaprocjena.


21/21

has already been established that the human factor isthe main cause of outbreaks of fire, therefore it is onlylogical that the greater the distance from a settlement,thefewer thefires. In preparingthese estimates, bufferswere made around settlements, and since they over-lapped in some places, this problem was solved using

the editor (Union, Clip), by creating one polygon withzones. After being transferred to a raster format, valueswere been reclassified and vulnerability zones shown(Fig. 20).

In order to obtain fire damage zones, it was neces-sary to collate all the factors analyzed and reclassified.First, they needed to be ranked by importance in termsof fire outbreak.

For this purpose, histograms were again used. Fromeach histogram, the largest number of fires within acolumn was extracted. The values for all factors weretotalled, and each value was divided by the decimalvalue ofthe total toderivethe importance ofeachfactorexpressedas a percentage (Table2.).

In this way, we minimizedsubjective assessment. Allthe analyzed (reclassified) factors were incorporated inSpatial Analyst tools (overlay; weighted overlay) and,based on Table 2, we added a percentage of importance.This was important because not all elements were ofequal significancein causing fires.

Factors Number of fires Risk %Temperature 11 6Relative humidity 18 10Aspect 15 8Slope 20 11CORINE Land Cover 34 18NDVI 18 10Roads 39 21Railways 16 8Settlements 16 8Total 187 100

Risk zone Number of fires Area (%)Zone 1(very low risk) 0 0

Zone 2 (low risk) 2 4.2Zone 3 (medium risk) 24 51.7Zone 4 (high risk) 40 44Zone 5 (very high risk) 0 0.1

The results obtained (Fig. 21) were compared withgeoreferenced fires from the fire season 2010, andfrom Table 3 it is clear that most fires (up to 40) fellwithin the high risk zone 4. Based on these results, wecan conclude that the analysis was sound, and that itcan be used for daily assessment, using alternatefactors day by day. This would produce a more accur-ate estimate.

Date post:	04-Apr-2018
Category:	Documents
Upload:	ivana-miletic-photography
View:	212 times
Download:	0 times

Požari-u-Dalmac.

Documents