Multivariate statistical analysis for priority area localization in air quality management plans

Transcription

1 Mutivariate statistica anaysis for priority area ocaization in air quaity management pans C. Trozzi & R. Vaccaro Techne Sr., Via N. Zabagia 3, Roma, tay Abstract n the paper the methodoogy estabished to ocate priority areas for remediation actions in air quaity management pans is reported. For cassification of areas within the regions an innovative methodoogy based on mutivariate statistica anaysis techniques was deveoped. Particuary automatic cassification criteria such as custer anaysis combined with simpe statistica procedures were used. n the methodoogy environmenta sensibiity and poutants oads are evauated and combined to obtain a cassification of the areas based on remediation action priority within the regiona territory. For priority areas ocaization two methodoogy are introduced: the first one takes into account overa sources, the second one keeps distinct areahe sources from point sources. Two different regiona case studies are discussed and evauated. Finay, baances and future studies direction are described. ntroduction The foing activities are necessary in the frame of the preparation of an Air Quaity Management Pan: emissions inventory, use of modes to estimate emissions in particuar topics (road transport, airports, forests, ports and navigation ines, forest fires). air quaity and meteoroogica data anaysis, projection of emissions without emissions reduction measures, cassification of territory and anaysis of priority in remediation actions, panning of measures to reduce emissions, definition of scenario of reduction,

2 projection of emissions in the pan scenario, use of air quaity dispersion and photochemica modes in the actua situation and in future projection with and without pan appication. These activities were recenty reviewed [], whde specific aspects were discussed in the past aso in Air Poution conference series [2, 3, and 41. n ths paper we discuss one of most important tasks of air quaity management: cassification of territory and anaysis of priority in remediation actions. The probem One of the most important probems in air quaity management is the ocation of the areas where the remediation actions are priority. From monitoring data we know if air quaity in seected ocations meets the nationa and internationa air quaity standards and from spatiay resoved emissions inventories we can have an idea of the contribution of the different sources to air poution. n optima situation we have, with the use of air quaity modes, an approximate frame of air poution distribution over the temtory. The deveoped methodoogy was an attempt to define automatic criteria to obtain a cassification of the areas based on priority in remediation action. The methodoogy The methodoogy was deveoped aong the foing ines: regiona territory cassification based on environmenta sensibiity of areas; regiona territory cassification based on poutants oads; regiona territory cassification based on remediation action priorities. For cassification of areas within the regions an innovative methodoogy based on mutivariate statistica anaysis techmques was deveoped. Particuary automatic cassification criteria such as custer anaysis combined with simpe statistica procedures were used. For statistica anaysis, and particuary for custer anaysis, the Statistica software from StatSoft was used [5]. Case studies tay Ministry of Environment in 1991 (Decree of May 20, 1991) introduced prescriptions and rues for the eaboration of regiona air quaity management pans. n December 1993 the Ministry of the Environment fmanced the reaization of such pans. n 1999 Liguria, and in 2000 Umbria, regiona administrations have competed the preparatory studies for the adoption of a regiona air quaity management pan according to the Decree of May 20, n the framework of Liguria and Umbria Regiona Air Quaity Management Pans, experimentation of automatic cassification of the areas within the territory, was carried out. Liguria region is a thin pan strip between sea and mountains with very compex orography. The Region contains one of main tay urban and industria area, Genova, and incudes the main nationa port and two other main ports,

3 three very arge power pant3, an integrated stee industry, chemica and petrochemicas industries, an airport and severa hghway sections. Umbria region is an interna area at the very heart of the taian peninsua, with 92 townships and no great urban aggomerations (Perugia, the capita, has ony mhabitants). The region has a good industria deveopment with iron and stee, metas and metaurgica, cement and chemicas industries. n the region there are aso two arge power pants. Environmenta sensibiity For the cassification of territory based on environmenta sensibiity the regions were subdivided in 1 km x 1 km squared grid and for each squared area sensibiity was evauated using the foing indicators: resident popuation, forests and other vegetation areas, beds avaiabe for tourists, biotopes. Resident popuation and beds avaiabe for tourists were avaiabe on municipa eve from STAT, the Nationa Statistica nstitute. Forests coverage was avaiabe from the maps of European Union CORNE Land Cover programme, an inventory of and cover in 44 casses presented as a cartograph product, at a scae of 1: tay CORNE and cover maps were avaiabe from the Centro nterregionae di coordinamento e docurnentazione per e informazioni territoriai. Biotopes map was avaiabe from European Union hded project Bioitay: Biotopes nventory of tay. A biotope is defined foing the Cozinci of the European Comrnunities Directive 92/43/EEC of 21 May 1992 on the conservation of natura habitats and of widfauna andfora as a site which, in the biogeographica region or regions to which is beongs, contributes significanty to the maintenance or restoration at a favorabe conservation status of: a natura habitat type of Community interest (means those which are: in danger of disappearance in their natura range, or have a sma natura range foing their regression or by reason of their intrinsicay restricted area, or present outstanding exampes of typica characteristics of one or more of the five foing biogeographica regions: Apine, Atantic, Continenta, Macaronesian and Mediterranean) a specie of Corntnunity interest (means species which are: endangered or vunerabe or rare or endemic and requiring particuar attention by reason of the specific nature of their habitat andor the potentia impact of their expoitation on their habitat andor the potentia impact of their expoitation on their conservation status). The foing procedure was used for data evauation on 1 km X 1 km grid: municipa data on resident popuation and on tourists beds avaiabe were assigned to 1 km x 1 km squared grid using CORTNE and cover cass "Urban fabric" through topoogic queries;

4 biotopes were evauated on 1 km X 1 km squared grid through topoogic query on Bioitay sites of regiona, nationa or community interests. forests and other vegetation were evauated on 1 km X 1 km squared grid through topoogic query on CORNE and cover map "Forests"; n Figure 1 Liguria and in Figure 2 Umbria case studies are reported. Liguria Region Environmenta sensibiity ' high urbanzat~on, 1 high tounsm andior urbanuat~on, few forests 1 forests wth urban~zzation andor tounsm hgh tounsm, few forests ' 1 forests - on density (between 20% and 40%) H forests - density (between 40% and 75%) forests -high dens~ty (greater than 75%) others Figure 1 - Liguria territory cassification based on Environmenta Sensibiity Umbria Region Environmenta sensibiity i -high high vegetaton no biotopes high - biotopes no vegetaton very high - vegetation and b~otopes very h~gh - urban areas Figure 2 - Unbria territory cassification based on Environmenta Sensibiity W n Liguria case study the areas with no popuation and tourists beds avaiabe are cassified in empiricay way. Particuary, using vegetation

5 coverage of the grid units, three casses are defined: forest - density (coverage of the singe units between 20% and 40%), forest - medium density (coverage between 40% and 75%) and forest - hgh density (coverage greater than 75%). The other grid units were cassified using custer anaysis on popuation, forests and tourism (no data on Biotopes was avaiabe at the time of eaboration). n Umbria case study automatic territory cassification (custer anaysis) based on a four previous introduced variabe, was carried out. Poutants oads For the cassification of territory based on poutants oads area, ine and point sources from regiona emissions inventories were taken into account. Emissions of the foing main five poutants were used: carbon monoxide (CO), voatie organic compounds (VOC), nitrogen oxides (NOx), fme particuate or suspended partices with diameter ess than 10 micron (PM1 O), suphur oxides (SOX). The nomencature used for the inventories fos the guideines coming from the European Commission CORNAR working group. CORNAR nomencature incudes about 200 activities grouped in 11 groups [6]. This nomencature was enarged to take into account ocay reevant activities. The sources are spit in point sources, area sources and inear sources. The fixed sources, for which the tota annua emissions of one poutant are arger than a fixed threshod vaue, are considered point sources. Linear sources correspond to the main communication ways (road, river, raiway, and seaway) and generay a the highways and a the main extra-urban roads are incuded. A the other sources are defined as area sources. The poutant oads methodoogy uses mutivariate statistica anaysis for cassification of areaine sources and simpe statistica methods for autonomous cassification of point sources. The area and ine emissions were evauated on 1 km X 1 km squared grid using EC CORhTE Land Cover maps andior others avaiabe and use maps, whe point sources are individuay treated. For area and ine emissions cassification was used the same methodoogy as in environmenta sensibiity case; the methodoogy is based on mutivariate statistica anaysis techmques, and particuary on automatic cassification criteria such as custer anaysis. The custer anaysis was appied using as variabes the vaues of 5 (poutants) X 11 (groups of activities) emissions and as cases the grid vaues. Point sources are individuay treated using simpe statistica methods for cassification. Particuary an indicator was buid for each point source k, using the emissions E of the poutants j, as: Pk = Z; (Ejk /( + 4*qSc0)) where 6,,co = 1 if j=co, 0 otherwise Mapnfo "Natura Break" agorithm was used for cassification [7]; the agorithm distributes the vaues so that the of each cass is as cose as possibe to each of the cass vaues in that cass. n F:igure 3 Liguria and in Figure 4 Umbria case studies are reported.

6 1 Poutants oads area and ine sources i 1, very high 1 ugh - hph i - B - 1 Figure 3 - Liguria poutants oads territory cassification based on area and ine sources Figure 4 - Liguria Poutants oads territory cassification based on point sources n Tabe 1 area and ine sources emissions in different oad casses are reported. As an indicator of municipaity poutant oad the number of grid areas in the different casses can be counted ad a rank of municipaities in function of the number of grid areas in very high, high, -high and areas casses can be perforned.

7 abe 1- Average area and ine sources emissions in different oad casses (Mg) a) Liguria Region Low- Average Average-hgh Very hgh CO COV NO, PMo so, Areas N b) Umbria Region Low Low-Average way Average Average-hgh CO 2,9 34,O 47,2 102,4 229,8 COV 3, ,9 35,3 70,6 NO, 0,s 5,7 25,6 14,2 26,8 PMo 0?3 23 8,4 8,o 16,7 so, , ,1 Areas N emediation actions priorities he combination of environmenta sensibiity and poutant oads cassification roduce a regiona cassification based on priorities. Ths cassification can take nto account overa sources (integrated methodoogy) or areaine sources and eparatey point sources (area and ine methodoogy). n the area and ine methodoogy the contribution of point sources was taken into account separatey from area and ine sources. An exampe of the appication of ths method is reported in Tabe 2 and Figure 5. Figure 5 - Liguria (Genova area) remediation action priority cassification

8 Tabe 2 - Cassification criteria in Liguria case studv Priority Sensibiity Area and ine poutant oad Very very high urbanization, hgh tourism, few forests Very high Very high Average Low tourism andor urbanization, few forests Very hgh ~orests - hgh density urbanization, high tourism, few forests Forests - density Forests with urbanization andor tourism tourism andor urbanization, few forests Forests - hgh density urbanization, high tourism, few forests Forests density Forests - density Forests with urbanization andor tourism tourism andor urbanization, few forests Forests - hgh density urbanization, high tourism, few forests Forests - density Forests with urbanization andor tourism tourism andor urbanization, few forests Forests - hgh density urbanization, hgh tourism, few forests Forests - hgh density Others hgh Very hgh Very high Average-high Very high Average-hgh Average-high Average Average-high Average Average Average- Average- Others n this approach great attention was devoted at the probem of dispersion of poutants emitted at high quote. Consequenty cassification of territo~y in terms of priority in remediation actions was performed ony with area and ine sources whie priority in point sources was separatey stated. n the integrated methodoogy the contribution of a the sources was taken into account at the same time. n this approach great attention was devoted at the oca impact of sources and at the origins of poution. Consequenty cassification of territory was performed in terms of where the emissions are generated. An exampe of the appication of this method is reported in Tabe 3 and Figure 6. Tabe 3 - Cassification criteria in Umbria case study Priority Sensibiity Area and ine Point sources aoutant oad oad high very high high high high very high hgh Average very hgh urban high Low

9 Tabe 3 (fo)- Cassification criteria in Umbria case study Priority Sensibiity Area and ine Point sources poutant oad oad Average-hgh very hgh urban hghways Average Average- Very Low very hgh urban hgh - vegetation no biotopes very hgh urban high - vegetation no biotopes very hgh urban hgh - biotopes no vegetation very hgh - vegetationhiotopes high - vegetation no biotopes hgh - vegetation no biotopes high - vegetation no biotopes high - biotopes no vegetation very hgh - vegetatio~i'biotopes hgh - vegetation no biotopes high - biotopes no vegetation A1 the others hgh hghways highways highways hghways hgh very hgh hgh Zero high Zero 1 priority, B high 1 - hgh,, - ov, 1, S L ery Figure 6 - Umbria territory cassification based on remediation action priority

10 Concusion The methodoogy presented was an usefu support for priority area individuation in air quaity management pans. The methodoogy can be further improved for better representation of point sources. Future research wi be directed toward: improve the cassification of territory based on environmenta sensibiity through sub-municipa popuation data; use of oca indicator of air poutant dispersion characteristics; use of regiona photochemica modes resuts as an indicator of poutant oads; use of mutivariate correspondences anaysis instead of empirica methods for remediation actions priority anaysis. Acknowedgement The work has been reaized for Liguria and Umbria regiona abstration in the framework of regiona air quaity management pan. References Trozzi C., Vaccaro R., Regiona Air Quaity Management Pans in tay: baances and perspectives, Air & Waste ~tanagement Association 94th Annua Co~ference & Exhibition, June 24-28, 2001, Orando, Forida, USA Vaccaro R., Trozzi C., Crocetti S., Air quaity management panning in tay: Trento Apine area case study, Sixth nternationa conference on Air Poution, Air Poution 1998, September 28-30, 1998, Genoa, tay. Trozzi C., Vaccaro R., Estimate of voatie organic compounds emissions from vegetation in regiona air quaity management pans. Seventh nternationa Conference on Air Poution, Air Poution 1999, Stanford, Caifornia, USA, Juy 1999 Trozzi C., Vaccaro L., Photochemica modes in Air Quaity Management Pan of a compex coasta area, Eighth nternationa conference on Air Poution, Air Poution 2000, Juy 24-26, 2000, New Ha Cambridge University, UK. StatSoft, Stntistica, Voume : Statistics 11, 1998 European Conxnission, DGX, CORNAR Technica Annexes, Voume 1 Nomencature and sofiware, Luxembourg, EUR EN, 1995 Mapnfo Corporation, Mapnjb Professiona User Manua, Troy, New York, 2000