Citation
Rohingya Refugee Crisis and Forest Cover Change in Teknaf, Bangladesh

Material Information

Title:
Rohingya Refugee Crisis and Forest Cover Change in Teknaf, Bangladesh
Creator:
Hassan, Mohammad
Publisher:
MDPI
Publication Date:
Language:
English
Physical Description:
Journal Article

Notes

Abstract:
Following a targeted campaign of violence by Myanmar military, police, and local militias, more than half a million Rohingya refugees have fled to neighboring Bangladesh since August 2017, joining thousands of others living in overcrowded settlement camps in Teknaf. To accommodate this mass influx of refugees, forestland is razed to build spontaneous settlements, resulting in an enormous threat to wildlife habitats, biodiversity, and entire ecosystems in the region. Although reports indicate that this rapid and vast expansion of refugee camps in Teknaf is causing large scale environmental destruction and degradation of forestlands, no study to date has quantified the camp expansion extent or forest cover loss. Using remotely sensed Sentinel-2A and -2B imagery and a random forest (RF) machine learning algorithm with ground observation data, we quantified the territorial expansion of refugee settlements and resulting degradation of the ecological resources surrounding the three largest concentrations of refugee camps—Kutupalong–Balukhali, Nayapara–Leda and Unchiprang—that developed between pre- and post-August of 2017. Employing RF as an image classification approach for this study with a cross-validation technique, we obtained a high overall classification accuracy of 94.53% and 95.14% for 2016 and 2017 land cover maps, respectively, with overall Kappa statistics of 0.93 and 0.94. The producer and user accuracy for forest cover ranged between 92.98–98.21% and 96.49–92.98%, respectively. Results derived from the thematic maps indicate a substantial expansion of refugee settlements in the three refugee camp study sites, with an increase of 175 to 1530 hectares between 2016 and 2017, and a net growth rate of 774%. The greatest camp expansion is observed in the Kutupalong–Balukhali site, growing from 146 ha to 1365 ha with a net increase of 1219 ha (total growth rate of 835%) in the same time period. While the refugee camps’ occupancy expanded at a rapid rate, this gain mostly occurred by replacing the forested land, degrading the forest cover surrounding the three camps by 2283 ha. Such rapid degradation of forested land has already triggered ecological problems and disturbed wildlife habitats in the area since many of these makeshift resettlement camps were set up in or near corridors for wild elephants, causing the death of several Rohingyas by elephant trampling. Hence, the findings of this study may motivate the Bangladesh government and international humanitarian organizations to develop better plans to protect the ecologically sensitive forested land and wildlife habitats surrounding the refugee camps, enable more informed management of the settlements, and assist in more sustainable resource mobilization for the Rohingya refugees.
Acquisition:
Collected for University of Florida's Institutional Repository by the UFIR Self-Submittal tool. Submitted by Mohammad Hassan.

Record Information

Source Institution:
University of Florida Institutional Repository
Holding Location:
University of Florida
Rights Management:
Copyright Creator/Rights holder. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

Article RohingyaRefugeeCrisisandForestCoverChangein Teknaf,Bangladesh MohammadMehedyHassan 1, *,AudreyCulverSmith 1 ,KatherineWalker 1 ID MunshiKhaledurRahman 2 andJaneSouthworth 1 1 DepartmentofGeography,UniversityofFlorida,3141TurlingtonHall,P.O.Box117315, Gainesville,FL32611-7315,USA;audreyculver@u.eduA.C.S.;kqwalker@u.eduK.W.; jsouthwo@u.eduJ.S. 2 DepartmentofGeography,VirginiaTech,115MajorWilliamsHall,220StangerStreet, Blacksburg,VA24061,USA;mkrahman@vt.edu Correspondence:mehedy@u.edu;Tel.:+1-352-745-9364 Received:6March2018;Accepted:25April2018;Published:30April2018 Abstract:FollowingatargetedcampaignofviolencebyMyanmarmilitary,police,andlocalmilitias,morethanhalfamillionRohingyarefugeeshaveedtoneighboringBangladeshsinceAugust2017,joiningthousandsofotherslivinginovercrowdedsettlementcampsinTeknaf.Toaccommodatethismassinuxofrefugees,forestlandisrazedtobuildspontaneoussettlements,resultinginanenormousthreattowildlifehabitats,biodiversity,andentireecosystemsintheregion.AlthoughreportsindicatethatthisrapidandvastexpansionofrefugeecampsinTeknafiscausinglargescaleenvironmentaldestructionanddegradationofforestlands,nostudytodatehasquantiedthecampexpansionextentorforestcoverloss.UsingremotelysensedSentinel-2Aand-2BimageryandarandomforestRFmachinelearningalgorithmwithgroundobservationdata,wequantiedtheterritorialexpansionofrefugeesettlementsandresultingdegradationoftheecologicalresourcessurroundingthethreelargestconcentrationsofrefugeecampsKutupalongBalukhali,NayaparaLedaandUnchiprangthatdevelopedbetweenpre-andpost-Augustof2017.EmployingRFasanimageclassicationapproachforthisstudywithacross-validationtechnique,weobtainedahighoverallclassicationaccuracyof94.53%and95.14%for2016and2017landcovermaps,respectively,withoverallKappastatisticsof0.93and0.94.Theproduceranduseraccuracyforforestcoverrangedbetween92.98.21%and96.49.98%,respectively.Resultsderivedfromthethematicmapsindicateasubstantialexpansionofrefugeesettlementsinthethreerefugeecampstudysites,withanincreaseof175to1530hectaresbetween2016and2017,andanetgrowthrateof774%.ThegreatestcampexpansionisobservedintheKutupalongBalukhalisite,growingfrom146hato1365hawithanetincreaseof1219hatotalgrowthrateof835%inthesametimeperiod.Whiletherefugeecamps'occupancyexpandedatarapidrate,thisgainmostlyoccurredbyreplacingtheforestedland,degradingtheforestcoversurroundingthethreecampsby2283ha.Suchrapiddegradationofforestedlandhasalreadytriggeredecologicalproblemsanddisturbedwildlifehabitatsintheareasincemanyofthesemakeshiftresettlementcampsweresetupinornearcorridorsforwildelephants,causingthedeathofseveralRohingyasbyelephanttrampling.Hence,thendingsofthisstudymaymotivatetheBangladeshgovernmentandinternationalhumanitarianorganizationstodevelopbetterplanstoprotecttheecologicallysensitiveforestedlandandwildlifehabitatssurroundingtherefugeecamps,enablemoreinformedmanagementofthesettlements,andassistinmoresustainableresourcemobilizationfortheRohingyarefugees. Keywords:Sentinel-2Aand-2B;randomforestRF;anthropogenicactivities;deforestation;ecologicalimpacts;carbonstorage;climatechange RemoteSens. 2018 10 ,689;doi:10.3390/rs10050689www.mdpi.com/journal/remotesensing

PAGE 2

RemoteSens. 2018 10 ,689 2of20 1.IntroductionTheworldisfacingthemostsevererefugeecrisisinhistorywithanaverageof28,300peopleperdayandevery20minforcedtoeetheirhomesduetowar,violence,orpersecutionfortheirrace,religion,ethnicityorpoliticalopinion,andthenumberisgrowingeveryday[1].TheRohingyapeopleareoneofthemoststatelessandwidelypersecutedminoritiesintheworld,facinganethniccleansingbytheBuddhistmajorityinMyanmar[2,3],forcingthemtoeeinsearchofrelativesafetyintheneighboringcountryofBangladesh.Sincetheearlynineties,theRohingyahavecontinuedtoeeinlargenumbersinFigure1fromtheRakhinestateofMyanmaracrosstheborder,mostlytotheTeknafregionofBangladesh,withmemoriesofgruesomeviolence,lossoflovedones,anddestructionofhomesandentirevillages[4,5].Thelatestwaveofviolence,however,hastriggeredthelargestRohingyainuxes;688,000refugeesareestimatedtohavecrossedfromtheRakhinestateintoTeknafsince25August2017[6].Themajorityofthispopulationsettledinmakeshiftcamps,replacingforestedhillssurroundingthetwoexistingrefugeecampslocatedinKutupalongandNayaparainTeknaf[7].Inadditiontosettlinginovercrowdedexistingcamps,refugeeshavesettledinspontaneoussitesinmoreoutlyingandremoteareaswithlittleaccesstoservicesandinfrastructure[6].Suchunprecedentedspeedandscaleoftherefugeeinuxandassociatedmakeshiftcampsexpansionhasalreadyresultedinthedegradationofprotectedforestanddestructionofcriticalwildlifehabitat,withwidespreadecologicalandenvironmentaldamageintheregion.Variousestimatessuggestthatapproximately4000acresofforestedhillsinthestudyareahavebeenclearedtoerectmakeshiftcampssinceAugust2017[711].Theforestlandlocatedinthestudyareaprovides,however,acriticalhomeforbothforest-dwellingandwetlandspecies,andasizeablenumberofbirdspecies[12].Itprovidesanimportantenvironmentforavastarrayofplants,includinganumberofmedicinalplantsthatareusedbythelocalcommunities[13,14],aswellasbeingasourceofsubstantialcarbonstorage[15].Additionally,thisenvironmentcontainsasanctuaryforwildAsianelephants,nestingsitesformanyshorebirds,andprovidesfoodandshelterformonkeys,snakes,bats,andotherwildanimals[16,17].Theprotectedforest,withitswildlifehabitatandothernaturalcapitalinthestudyarea,isbeingdestroyedanddegradedatanalarmingratemainlyduetoclear-cuttingforagriculture,ranchinganddevelopment,andloggingfortimber.However,degradationduetorapidconversionforrefugeecampsandmakeshiftsettlementsisthegreatestcatalystofenvironmentaldestructionoccurringatalargescaleinrecenttimes[18].Empiricalstudiessuggestthatdeforestationdrivenbyanthropogenicactivitiescanhavemultiplenegativeimpactsontheenvironmentincludinglossofwildlifehabitat[19],soilerosionanddesertication[20],watercycledisruption,lossoftraditionallivelihoods,andincreasedecologicalrisksfromforestfragmentation.Changesinforestcoverfurtheraffectthecapacityofforestbiomasstostorecarbon,disturbinglocalclimatebymodulatingthediurnaltemperaturevariation,andthusincreasingrisksofglobalclimatechange[21,22].Hence,periodicalforestcoverchangeanddriversofsuchchangemustbemonitoredanddocumentedtosupportpoliciesandmanagementpracticestoprotect,conserve,andsustainablyuseresourceswhilemaintainingecosystemfunctionsandforestsbiodiversity[22,23].Remotelysensedsatelliteobservationdataarewidelyusedtomonitorlocal[24],regionaltogloballandcoverchangeandvegetationhealthmonitoring[25]duetotheirhighspatialresolutionandtemporalfrequenciesandopenavailabilityontheinternet.Satelliteremotesensingmayalsobeusedforterrestrialcarbonquanticationandmonitoringofclimatevariability,using,forexample,NASA'sMOD17algorithm[26,27]anddatafromtheadvancedveryhighresolutionradiometerAVHRR[28].Themonitoringofrefugeecampexpansionandtheresultantdegradationoftheenvironmentandforestcoverchangeusingearthobservationdatasuchasaerialphotography,LiDAR,andhigh-resolutionsatelliteimageryi.e.,Quickbird,IKONOShasbeenseenquiteofteninpractice[2932].Differentmethodologieswithvaryingdegreesofresources,costs,accuracy,expertiseandtechnologyhavebeenappliedintrackingrefugeecampsandassessingenvironmentalimpactsonthecamps'surroundings.Forexample,Lodhietal.[33]monitoredlandcoverchangesand

PAGE 3

RemoteSens. 2018 10 ,689 3of20environmentalimpactsusingLandsatimagerywithunsupervisedclassicationintheSiranValleyinnorthernPakistanresultingfromaninuxofAfghanrefugeesintheearly1990s.Sprhnleetal.[32]observedtheimpactsofinternallydisplacedpersonIDPcampsonwoodresourcesinZalingei,Darfurusinghighresolutionsatelliteimagerybetween2003and2008.TheirresearchfoundthatincreasingIDPcampscausedaconsiderabledecreaseinwoodyvegetationsurroundingthecamparea.Langeretal.[34]monitoredthelong-termenvironmentalimpacts,includingdeforestation,ofrefugeecampsusingLandsatdatainLukole,Tanzania.Theauthorsobservedthatthedevelopmentofcampscausedsignicantdegradationofthenaturalvegetationsurroundingthecampsarea.AlthoughthereexistsawidespreadconsensusthatdeforestationcurrentlytakingplaceintheTeknafregionofBangladeshislargelydrivenbyrefugeesettlementexpansionandrelatedinfrastructuredevelopment,todate,nostudyhasbeencarriedouttoquantifytheactualrateofforestdegradationresultingfromcampexpansionsinthearea.Hence,tollthisknowledgegap,thisstudyaimstoquantifysettlementexpansioncoupledwithforestcoverdestructionusingSentinel-2Aand-2Bimagerybetweenpre-inuxandpost-inuxofRohingyarefugeesin2017intheTeknafareasituatedinthesouthernmosttipofBangladesh.Sentinel-2Aand-2Bimageryhasahighspatialresolutionof10mblue,red,greenandnear-infraredbandsandisopenlyandfreelyavailableontheinternetandupdatedfrequently.Suchhigh-resolutiondataisoptimalforconductinglargeareavegetation,landcover,andenvironmentalchangemonitoring. Figure1.NumberofRohingyarefugeeinuxesinBangladeshfrom1991to2017cumulativenumberofrefugeesincludingpre-inuxpopulation.Datafortheperiodof1991wereretrievedfromUnitedNationsHighCommissionerforRefugeesUNCHR[35],andrefugeepopulationdatafortheyearof2017wereobtainedfromInterSectorCoordinationGroupISCG[36]. 2.MaterialsandMethods 2.1.StudyAreaThestudyareaTeknafisapeninsulainthesouthernmostterritoriesofBangladesh,locatedonaquitenarrowstripoflandbetweentheRiverNaftotheeastandtheBayofBengaltothesouthandwestinFigure2.Geographically,itextendsfrom2045011.1500Nto2117027.5100Nlatitudeandfrom9220015.8300Eto926038.4900Elongitude.Maximumextensionisabout65kminthenorthsouthdirectionand10kmintheeastwestdirection.Theareaexhibitsadiversephysiographysuchasundulatinghillock,piedmontplains,tidaloodplains[12],andacontinuouslineofsandybeachesthatstretchestoCox'sBazaralongtheBayofBengal-kmlongreportedlythelongestbeachintheworld.Thiscoastalbeachisbackedbyfoothills,whichareforestedinpatcheswithelevationranging

PAGE 4

RemoteSens. 2018 10 ,689 4of20between100mand250m.Thepiedmontplainscomprisegentlyslopinglandatthefoothillsandaregenerallyoccupiedbyhumansettlements.Themaximumelevationoftheseplainsis10mandtheyaccountfor31%ofthetotallandinthearea,mainlyonthewestern,eastern,andsouthernsidesofthehills,rollingnorthtosouththroughthepeninsula.Thesandybeachescompriseanareaof3155haor9.03%ofthetotalareaandlieonthewestsideofthepeninsulaalongtheBayofBengal.ThehighestmeanelevationinTeknafisestimatedat55mand31mintheTeknafandWhykonunionsrespectively,andthelowestaverageelevationisintheSabrangandNhillaunionsat5m.Theareahasasubtropicalclimatecharacterizedbyarelativelyhighamountofannualrainfallaverageannualrainfallis4000mmwiththemostrainoccurringinJulymm,andtheleastrainoccurringinJanuarymm.Theaverageannualtemperatureinthestudyareais78.98F.1C;thewarmestmonthisMay.2CandthecoolestmonthisJanuary,withanaveragetemperatureof58.82F.9C.Thephysiographytogetherwiththeclimatepatternsoftheregioncontributetotheoccurrenceofdenseforestlandandthenumberofendangeredanimalsforwhichthepeninsulaishome.Although41%oftheareaiscoveredwithforest,thisnumberhasdecreasedovertimeduetoextensiveanthropogenicactivitiesinrecentdecades[18].Withintheforestedarea,11,615haisadesignatedwildlifesanctuaryseeFigure2shadedarea,whichharborsthemanyendangeredspeciessuchasthewildAsianelephantandnumbersofshorelineandoffshorebirds.Thetotalareaselectedforthisstudyis555squarekilometers,whichcomprisestenunions,onePaurashavaand147villages,withatotalcumulativepopulationof440,009people[37].Theeconomyintheregionismainlydominatedbyprimaryeconomicactivitieswithalargeproportionofthepopulationrelyingonbetelleafandbetelnutgardening,saltproduction,andseashing.Asaresult,63.27%oftotalpopulationinthestudyareaareengagedintheagriculturalsector,25%areemployedintheservicesector,followedbyonly11.57%workingintheindustrysector[37].Duetotheundulatinglandscapecharacteristicsandsandysoilqualitywithmuchoftheareacoveredbyforest,traditionalagriculturalpracticessuchasricearehighlylimitedintheregion.Asmallerandbetter-offproportionofthepopulationengageinshrimpcultivationinthelowerriseofthebankoftheNafRiver,whichissubjecttoinundation.Thestudyareaalsohousesnearly900,000Rohingyarefugees,settlingindifferentlocationsbothinestablishedhostcommunitiesandspontaneouscampsthathavesprungupintheregionsinceAugust2017[38].Althoughtherearesome100spontaneousrefugeecampsacrossthestudyarea,threemainsitestogetherwiththeirexpansionareashousethemajorityofRohingyarefugeeswithatotalpopulationof771,000,accountingfornearly88%ofrefugeessettledintheregion[36,38].Amongthesethreesites,theKutupalongBalukhalicampexpansionsitehasthemaximumrefugeeconcentrationi.e.,apopulationdensityof533refugeesperhectarewithatotalpopulationof713,000Rohingya,makingitthelargestrefugeesettlementintheregion[6].Nayaparistheoldestregisteredcampandsecondlargestrefugeesettlementsiteinthestudyarea,whichextendsfromJadimurainthesouthtoLedainthenorthandhostingapproximately37,000refugees.TheUnchiprangsite,locatedbetweenNayaparaandKutupalongcamps,isthethirdlargestsite,accommodatingapproximately22,000RohingyarefugeesTable1. Table1.ThreemajorconcentrationsofRohingyarefugeecampsites,theirgeographicallocations,respectivepopulationsizes,andtheoptimalthresholdofeachbufferzonestudiedhere.Thebuffersizeforeachsitewasdeterminedbasedonthecamp'saerialexpansionin2017. CampsStudiedZoneBufferGeographicLocation TotalRefugee Population AverageElevation inBufferZone Kutupalongand BalukhaliExpansion 10kmfrompreexisting refugeecamp Latitude:21 12 0 36.70 00 N Longitude:92 9 0 52.41 00 E 713,00023m Unchiprang 3km,centroidexistingcamp Latitude:21 5 0 6.13 00 N Longitude:92 12 0 29.64 00 E 22,00027m NayaparaandLeda expansion 4kmfrompreexistingrefugee camp Latitude:20 57 0 19.15 00 N Longitude:92 15 0 5.09 00 E 37,00054m

PAGE 5

RemoteSens. 2018 10 ,689 5of20 Figure2.Locationandelevationofthestudyarea,includingthegeographicalsettingofeachrefugeecampwithtotalrefugeepopulationasof25March2018.Thecamps'geographicallocationsandnumbersofRohingyarefugeescirclesizerepresentscomparativepopulationsizeofeachareaineachcampwerederivedfromInterSectorCoordinationGroupISCG[36].TheinsetmaptoprightcornershowsBangladeshwiththreesides;west,northandeast,borderedbyIndiaandonlyasmallborderwithMyanmarinthesoutheastwherethestudyarea,Teknaf,islocated.

PAGE 6

RemoteSens. 2018 10 ,689 6of20 2.2.DataandImageProcessingToachievetheobjectivesofthisresearch,diversedatasetswererequiredincludinggeospatial,socioeconomic,demographic,andbiophysicaldataTable2.ThemainlandcoverdatawerederivedfromSentinel-2AandSentinel-2Bhigh-resolutionmultispectralsatelliteimagerywithaspatialresolutionof10minthevisibleandNIRbands.Weacquiredtwotime-seriesSentinelsatelliteimagesforthestudyarea,includinganimageofthepre-inuxperiodi.e.,priorto25August2017Sentinel-2A,andoneimagerepresentinglandcoverfeaturesforthepost-inuxperiodi.e.,after25August2017Sentinel-2B.Duetothemonsoonclimatepatternsintheregion,itisdifculttoobtaincloud-freeimagesfortheareaduringthemonthsoftherainyseason,MarchtoNovember.Hence,wedownloadedapre-eventcloud-freeSentinel-2Asatelliteimage,acquiredforDecember2016,andaSentinel-2Bpost-eventimageacquiredforDecember2017,fromtheUnitedStatesGeologicalSurveyUSGSGlobalVisualizationViewerGloVisavailableforfreedownloadathttps://glovis.usgs.gov/.Alldownloadedimageswerevisuallyinspectedandonlythethreevisiblebandsbands2,3and4andonenear-infraredbandband8ofSentinel-2Aand2BwerestackedInFigure3andsubsettothestudyareageographicboundary.Inaddition,averyhighresolutionQuickBirdimagewasacquiredfor2010,consistingoffourmultispectralbandsred,green,blueandnearinfraredat2.4mspatialresolution,andonepanchromaticbandwith0.6mspatialresolution.Themultispectralbandswerepan-sharpenedwiththepanchromaticbandtoproduceamultispectralimageat0.6mspatialresolution.Thishigh-resolutionQuickBirdimagewasmainlyusedfortrackingpreviouslyestablishedcamplocations,identifyinglandcoverfeaturesinthearea,andaccuracyassessmentforthepre-inuxthematicmaps.Inaddition,acontinuoussurfaceelevationmapwasgeneratedusingaShuttleRadarTopographyMissionSRTM30-mspatialresolutionimageacquiredforSeptember2014.Trainingsamplesandothergeospatialinformationwerecollectedduringaeldvisittothestudyarea,includingtheKutupalongandBalukhairefugeecampsites,inDecember2017,usingahandheldGlobalPositioningSystemGPSetrexHCxlegendandahigh-denitionContourGPScamera[39].OuremployedContourGPScameracaptureshigh-denitionHDspatialvideoFigure4withdataaccuracywithin10mhorizontallyand30mvertically,allowingforeasyaccessoftrainingdata,siteobservations,andoverallreal-timedocumentationoftheconditionsexperiencedbydwellersincampsandthesurroundingenvironments[40].Later,thesespatialdataweredigitizedinGoogleEarthandaGIStodeterminethecamps'aerialexpansionandselecttrainingsitesforlandcoverclassication.Theprocessisrapidandcanberepeatedinstudysitesthroughtimetotrackspatial-temporaldynamicsofthecommunitiesandtheirimpactsontheenvironment.AsrecommendedbyJensen[41],thenumberoftrainingpixelsshouldatleastbeequalto10timesthenumberofvariablesusedintheclassicationmodelforanonparametricclassicationapproach;wethereforeextractedapproximately200samplesfromourGPSimbeddedspatialvideocameraacrossthestudyareas.SincemostGPS-basedgroundtrainingdatacollectedwerebiasedtolocationsacrossfrom/closetotheroadnetwork,anadditional300stratiedrandomsampleswerealsotakenasreferencedata[24].AsseveralstudiessuggestthatnonparametricmachinelearningclassierssuchasRFrequiredalargenumberofreferencedatatoattainthemostfavorableoutcome[4245],wecombinedgroundtrainingdataandrandomlychosensamplestoobtainsufcientinformationonthelandcover,whilemaintainingthespatialdistributionofourreferencedatathroughoutthestudyarea.Later,largeportionsofthesesampleswereusedasinputvariablesfortheRFclassierandsomewereretainedandusedforvalidationforthethematicmapof2017.Referencedatafortheoutputsof2016approximately500weremanuallycollectedusingGoogleEarthimagery,QuickBirdhighresolutionimagery,andSentinel-2Aimagery.Usingthesetrainingsamples,weextractedthespectralpropertiesofselectedbandsandindicesandlaterusedtheseasinputvariablesfortheRFmodelcalibrationandpredictions.

PAGE 7

RemoteSens. 2018 10 ,689 7of20 Figure3.Falsecolorcompositeimagesofthethreecampstudysitesbetweenpre-andpost-Augustof2017.Thecolumnsarealignedtorepresentthethreedifferentcampsi.e.,campsA,BandCwhiletherowsrepresentthetwo-timesteps.TherstrowofimageswastakeninDecember2017,depictingthepost-Augustrefugeeinux:A1KutupalongBalukhalicamps;B1Unchiprangcamp;andC1NayaparaLedacamparea.ThesecondrowofimagesA2,B2andC2arefromDecember2016,showingthesamecampsandsurroundingareasastheA1,B1andC1images,respectively,priortothemassinuxofRohingyarefugeesinAugustof2017.Theyellowpolygonsdrawnontheimageshighlightthegrayspectralreectancerepresentingthethreerefugeecampsites,andshowalarge-scalephysicalexpansionintherstrowofimagesDecember2017comparedtothesecondrowofimagesDecember2016.Inthesefalsecolorcompositeimages,reddepictsforestland;darkblueindicateswater,andbrownrepresentssoil/nonforest.

PAGE 8

RemoteSens. 2018 10 ,689 8of20Further,awiderangeofgeospatialinformationsuchasthecamps'geographicallocations,roadnetworksandotherphysicalfeatureswereusedintrackingthenewestsettlementexpansionsanddeterminingcampboundaries,especiallyofthosethatarosespontaneouslyoutsideofthetwopreexistingregisteredcampsduringthepost-inuxperiodinthestudyarea.Thevastmajorityofthisinformation,suchasthespatialdistributionofrefugeecampsandtheirterritorialexpansion,wasderivedfromthedatabanksofvariousnationalandinternationalorganizationssuchasISCG,UNCHR,andArcGISonlinemaps.Administrativeboundaries,roadnetworks,existingland-useandotherphysicalfeatureswerecollectedfromdetailedland-usesurveysforthemasterplanoftheregionin2011. Figure4. RefugeecampsonhillslopesofBalukhaliexpansionsitecapturedbyhigh-denitionspatialvideocamera.Insetimageshowslocationofground-truthtrainingsamplesleftandgeographicboundaryofBalukhalicamp.Source:eldvisitanddatacollectionon28December2017. Table2. Descriptionofthecollectedremotesensing,geospatialandsocioeconomicdata. DataTypesYear/AcquisitionDateProducerScale Sentinel-2A,2B 6December2015 30December2016 8February2017 15December2017 EuropeanSpaceAgency DownloadedfromUSGSgloballand coverFacilities https://glovis.usgs.gov/ 10mBands2,3,4,8 QuickBird6February2010DigitalGlobe0.6m SRTM23September2014https://earthexplorer.usgs.gov/30m TrainingSample28December28,2017 FieldvisitingusinghandheldGPSand spatialvideocamera PopulationCounts Populationand HousingCensuse-2011 BangladeshBureauofStatistics CommunityReport: Cox'sBazar RefugeeCounts ISCG,IOM,UNCHR, OCHA UnitedNationsUN SituationReport: RohingyaRefugeeCrisis PhysicalFeatures, RoadNetworkData MasterPlan-2011UrbanDevelopmentDirectorateUDDTheNormalizedDifferenceVegetationIndexNDVIiswidelyusedtomonitorvegetationchangedynamicsfromregionaltoglobalscalesandhasbeenshowntobeausefulmeasureofvegetativelandcoverchange[46].Forthisstudy,wecomputedNDVIvaluesusingbands4and8ofSentinel-2AandSentinel-2Bimagery,andproducedNDVImapsofthestudyareaforfourtimesteps.Ashighrainfall

PAGE 9

RemoteSens. 2018 10 ,689 9of20variabilitymayleadtovaryingphenologicalconditionsofvegetationcanopygreenness,abetterunderstandingofrainfallpatternswithphenologicalconditionsandsubsequentimpactsonvegetationhealthiscrucialforchoosingappropriatesatelliteimagedatesforthistypeofanalysis.Evenasmallchangeinrainfallparametersinlongtermvegetationhealthmonitoringcanleadtounpredictableoutcomesandincorrectconclusions.Consideringthesensitivityofvegetationhealthtobroaderclimaticandlocalweatherconditions,weobservedNDVIonlyinthedryseason,calculatedforDecember2015,December2016,February2017andDecember2017.ChoosingimagesbetweenDecemberandFebruaryisofobviouspracticality,asitiswinterandthedryseasoninBangladesh,allowingfortheacquisitionofcloud-freeimagesandanalysiswithlessvariabilityingreennessofvegetationduetotherainymonthsofmonsoonclimates.TheNDVIspectralindicesvalueswerelaterusedasexplanatoryvariablesintheRFmodelcalibrationandfurtherassociatedwithouranalysis.Demographicdatasuchasnumberofrefugeesovertimeisrequiredforthestudyoftheimpactofcampexpansionandassociatedanthropogenic-inducedecologicalchangessuchasdiminishedforestcover.Assuch,incomingrefugeenumbersbothinregisteredandnon-registeredcampsandlocalpopulationdatawereassembledfromvariousnationalandinternationalagenciessuchastheBangladeshBureauofStatisticsBBS,UnitedNationsHighCommissionerforRefugeesUNCHR,InternationalOrganizationforMigrationIOMandInterSectorCoordinationGroupISCG.Therefugeecountpopulationdatawaslaterusedbothforlandcoverclassicationandlandcoverchangedriveranalysis.Furthermore,westudiedvariousonlineresourcesincludingnewspaperreports,blogsandotherinformationprovidedbycharityandhumanitarianorganizationswhoareactivelyworkingintrackinganddocumentingthismanmadesocialandecologicalcatastrophethatisunprecedentedinscaleinthestudyarea. 2.3.ImageClassicationAhumaninterpretationofanimagecaneasilycategorizeitintoclassesofinterest,butitisgenerallydifculttoachievethesameresultusingcomputer-derivedimageclassicationtechniques.Recentdevelopmentsinneresolutionremotesensingimageryandimprovedimageclassicationalgorithmsmakepossiblethedataminingandmonitoringofabroadrangeoftargetfeaturesontheground.RandomforestRF,amachinelearning,non-parametricalgorithm[47]hasbecomeanefcientandpopularmodelforremotesensingapplicationssuchaslandcoverimageclassication[42,43,48],andhasproventobeadesirablealternativetothetraditionalparametricbasedimageanalysis[44].Duetoitsabilitytoimplicitlydealwithmissingvaluesonpreviouslyunseenhighdimensionaldataandcomplexrelationsamongvariablescoupledwithhighclassicationaccuracy[44,45],applicationsofRFhavebeenseenwidelythroughoutdifferentdisciplinesandresearchobjectives[4244,48].ThelandcoverclassicationmethodemployedforthisstudywasthereforeRF,chosenforitspowerfulmachinelearningensembleandnon-parametricstatisticallearningtechnique.TheRFmodelandassociatedpackageswereloadedandexecutedinRStudioopensourcestatisticalcomputingandgraphicssoftwarehttps://www.r-project.org/.Agreatvarietyoflanduseandlandcovertypescanbeobservedinthisregion,suchasforest,homesteadvegetation,agriculturalland,constructionandopenarea,urbanland,andothers.Sincetheaimofthisstudywastoexaminetherefugeecampsexpansiondrivenforestcoverchange,weadoptedasimpleclassicationsystem,partlyderivedfromAnderson's[49]rst-orderhierarchicalclassicationsystem.Initially,sixrepresentativelandcovercategoriessuchasforest,urban,camp,water,agriculture,andsandsweregeneratedusingexpertknowledgeofthestudyareaandtheobservationsfromtheeldsurveyundertakeninDecember2017.Sincenonparametricbasedregressionmodelsperformwelloncetheyarecalibratedwithsufcientvariables[44],wedeveloped16variablesincludingthreetopographicvariablesi.e.,elevation,slopeandaspect,and13spectralvariablesderivedfromSentinel-2Aand2Bimagery.ThetopographicvariablesweremainlyextractedfromDigitalElevationdataofSRTMShuttleRadarTopographyMissionimagery,andthespectralvariableswereextractedfromthevisibleandNIRbandsofSentinel-2Aand-2Bimagery.ThespectralreectanceofthesebandswereusedtocomputeindicessuchasNDVINormalizedDifferenceVegetationIndex,

PAGE 10

RemoteSens. 2018 10 ,689 10of20MSIMoistureStressIndex,SAVISoilAdjustedVegetationIndex,TasseledCapKauthThomasGreenness,Brightness,andWetnessandPCAprincipalcomponentanalysisscores1.InRFmodels,severalparametersarerequiredtobespeciedpriortomodelexecution,suchasthenumberofinputvariablesrandomlychosenateachsplitmtry,thenumberoftreesintheforestntree,andthenodesize[45].AsalargenumberoftreesisrecommendedwhenusingRFalgorithmstostabilizethemeansquarederrorovermanyiterations,ineachiterationprocess,theRFalgorithmwasruntogrow1000treeswith5to6ofthe16explanatoryvariablesrandomlyselectedateachnode,aspotentialvariablesonwhichtobasethesplit.Ifthenumberofvariablesistoolarge,asisoftenthecaseformultisourcestudies,aRFcanbeappliedonlytothosevariableswhichhavebeenidentiedasthemostimportantandwhichcontributemosttoincreasedaccuracy[44,5052].Sincewehavealimitednumberofvariables,wecalibratedourmodelbasedonthecompletedatasetandgeneratedtwotime-periodlandcovermapswithsixlandcoverclasses.LatertheselandcovermapswereimportedinArcGISandconvertedtopolygonshapelesformappingandfurtheranalysis.Sinceourgoalistoestimatecampoccupancyrateandsubsequentforestcoverchange,wemergedsixlandcoverclassesintothreebroadcategories:forest,nonforest,andcamps.Afterconsideringthelandscapedynamicsintheregionandstressorsonvegetationcoverfrommultiplesources,wecreatedmultipleringbufferzonescenteringonthepreexistingrefugeecampsinthethreesites,therebyallowingustonarrowouranalysistothemostaffecteddeforestationzonesofthestudyareaandwithgreaterspatialandtemporaldetails.TheextensionofeachbufferzonewasdeterminedbasedontherefugeesettlementexpansionseeTable1.Alllandcoverclassesandconversionmatrixesfromthethreeperiodthematicmapsthenwereaggregatedandquantifiedintoeachbufferzoneandanalyzedforeachsiteindividually. 3.ResultsToquantifytherefugeecamps'expansion,westudiedtwotime-stepsoflandcovermapsofthestudyarea,includingthepre-influxbefore25August2017period,andthepost-influxafter25August2017period.Themapsdesignatethreemajorlandcoverclasses:forest,nonforest,andcampspresentedinFigure5andthefindingsaredescribedbelow. 3.1.AccuracyAssessmentGiventhatrandomforestsclassicationaccuracyisestimatedinternallyduringthebootstrappingprocess,70percentoftrainingdataareusedforthetreegrowingprocess,andtheremaining30percentofdatapointsareusedtoestimateout-of-bagOOBerror,asseparatevalidationdoesnotrequireanunbiasedestimateofthetesterror[4244].UsingtheOOBerrormatrixwithRF,weachievedmodestlyhighoverallclassicationaccuracyforthethreelandcovermaps,withlowerOOBerror,suchas3.67and3.19forthe2016and2017time-stepsrespectively.Inaddition,weinspectedeachlandcovermapandindependentlycross-validatedeachlandcoverclassusing150stratiedrandomtrainingpointsforeachperiod.Forthepre-inuxesmapof2016,weusedhigh-resolutionGoogleEarthimagerycombinedwithQuickBirdfalsecolorcompositeimages.Whileforthepost-inuximageDecember2017,weusedSentinel-2Btruecolorcompositeimagerywith10-mresolutionand50referencedatapointssincehigh-resolutionimageryfromGoogleEarthforthelaterdatewasunavailableinthisstudyarea.Usingthesedataasabasis,producer'saccuracy,useraccuracyandkappacoefcientwerecalculated.Ourindependentvalidationalsoreportedhighoverallclassicationaccuraciesof94.53%and95.14%withoverallkappastatisticsof0.93and0.94forthelandcovermapsof2016and2017respectively.Whileproduceraccuracyanduseraccuracyforforestcoverwasintherangebetween92.98.21%and96.49.98%,respectively.Campareaalsoreportedhigherproduceranduseraccuracy,between91.67%to96.61%and96.49%,respectively.Themostimportantvariablesaccordingtothevaluesofmeandecreaseaccuracywerevariedbut,ingeneral,theNIRandredbands,NDVI,elevation,andtasseledcapgreennessandbrightnesswereofgreatestimportance.Ontheotherhand,slopeandaspectweretheleastpromisingindicatorsinourclassicationmodel.Asvariouspreviousstudieshavesuggested[44,5052],ifthenumberofvariablesistoolarge,asis

PAGE 11

RemoteSens. 2018 10 ,689 11of20oftenthecaseformultisourcestudies,aRFcanbeappliedtoonlythosevariableshavebeenidentiedasthemostimportantandthosewhichcontributemosttoincreasedaccuracy.Sincewehavealimitednumberofvariables,wedidnotexcludethelesspromisingvariablesasindicatedbythevariableofimportanceselectiongraph.Hence,wecalibratedourRFmodelbasedonthecompletedatasetwith16explanatoryvariables. Figure5.Landcovermapsforthestudyareaclassiedintothreemajorlandcoverclasses,includingforestgreen,refugeecampfuchsia,andnonforestgray,attwotime-stepsrepresentingpre-inux:ADecember2016andpost-inux:BDecember2017.Thepre-inuxmapAshowstworefugeesettlementcamps;however,inthepost-inuxlandcovermapB,manyadditional,spontaneouscampsarevisiblewithforestedlandreplacedbycontinuousexpansionofrefugeesettlements.

PAGE 12

RemoteSens. 2018 10 ,689 12of20 3.2.RefugeeCampExpansionandForestCoverChangeSurroundingtheKutupalongBalukhaliAreasThetimeserieslandcovermapsgeneratedinthisstudysuggestthattherewereapproximately146hectaresoflandoccupiedbyrefugeecampsinDecember2016,mainlyinKutupalong,whichisoneoftwogovernment-runrefugeecampsinthestudyarea.Therecentmilitarycrackdown,however,triggeredamuchlarger-scalerefugeeinfluxinthisregion,resultinginspontaneoussettlementexpansionacrosspreexistingsites.Asaresult,landoccupiedbysettlementsexpandedveryrapidlyacrosstheKutupalongrefugeecamps,increasingfrom146hectaresto1365hectaresbetweenDecember2016andDecember2017,withatotalgrowthrateof835percent.Theforestcoverwithinthe10kmbuffercreatedaroundthecenterofthepreexistingrefugeecampsinKutupalongshowsadownwardtrend,from11,800hectaresto9740hectarestotalforestloss2060hectareswithanetdeclinerateof18%.Theforestlossduringthisshorttimeperiodisdrivenmainlybytheever-increasingspatialexpansionoftherefugeecampsandassociatedanthropogenicactivities,suchascuttingdownforestfortimber,fuelwoodandothersubsistenceneeds.Asaresult,nonforestrelatedactivitieshaveshownanincreasewithanetgainof842hectaresaroundtheKutupalongcampsseeTable3.Thisvastcampexpansionassociatedwithlargescaleforestcoverdeclinetookplacemainlyinthesouth,westandsouthwestdirectionsextendingfromthepreexistingrefugeecampsinKutupalong.Forexample,inthesoutherlydirectionfromKutupalong,refugeecampexpansiontotaled324hectares,andinthewestandsouthwestcampsexpanded184and605hectares,respectively.Meanwhile,forestcoverinthisthree-directionradiusdeclinedby202,364,and940hectaresrespectively.ThequanticationofdegradationofforestcoverinthesetrajectoriesarealsosupportedbytheNDVImapgeneratedinthisstudy.Thefour-periodNDVIgraphsindicatethatNDVIvaluesdeclinedsignicantlybetweenFebruaryandDecemberof2017within7kmsouthwestandwithin4kminawesterlydirectionoftheKutupalongcampFigure6.Thelandcoverclassconversionmatrixalsosuggeststhatforesttononforestconversionincreasedsubstantially,assuch,1882hectaresofforestlandtransformedintononforestlandsurroundingtheKutupalongcampbetween2016to2017Figure7.Additionally,foresttocampandnonforesttocampconversionratewas763and536hectares,respectively.Thehighestforestconversionagaintookplaceinasouthwestdirectionfromthecamps;theforesttocampandforesttononforestconversionrateswere471and629hectaresrespectively.Hence,majorcampexpansionandlossofforestresourcessurroundingtheKutupalongcampoccurredmainlyinasouthwesterlydirection,accountingfor605hectaresofrefugeecampswith940hectaresforestdegradationbetweenDecember2016andDecember2017.Thislargescaleofcampexpansionstretches8kmtowardthesouth-southwestfromthepreexistingrefugeecampsinKutupalong,reachingThangkhaliandfurthersouth-southwesttoHakimpara,Jamtoli,andBagghonacampsseeFigures2and5.Amongtheothertrajectories,thenortherlydirectionaccountsforthehighestforestcoverdecline,estimatedat137hectaresbetweenthe2016and2017timeperiodsstudiedhere. Table3.Areainhectaresandspatialchangesinlandcoverclassesandoverallnetgainandlossesbetween2016and2017inthreestudysites:KutupalongBalukhali,Unchiprang,andNayaparaLeda. KutupalongBalukhali2016ha2017haNetChangebyClasshaGrowth/DeclineRates% NetChangeinThreeCamp Areasbetween2016 Camp14613651219835 CampArea: +1356ha Forest: )]TJ/F192 8.9664 Tf 8.469 0 Td [(2283ha Nonforest: +928ha Forest11,8009740 )]TJ/F195 8.9664 Tf 7.375 0 Td [(2060 )]TJ/F195 8.9664 Tf 7.375 0 Td [(18 Nonforest7550839284211 Unchiprang Camp0323232 Forest14991452 )]TJ/F195 8.9664 Tf 7.375 0 Td [(47 )]TJ/F195 8.9664 Tf 7.374 0 Td [(3 Nonforest949964151.5 NayaparaLeda Camp29133105359 Forest28602684 )]TJ/F195 8.9664 Tf 7.374 0 Td [(176 )]TJ/F195 8.9664 Tf 7.374 0 Td [(6 Nonforest925996718

PAGE 13

RemoteSens. 2018 10 ,689 13of20 Figure6.NDVIinthetwomostaffecteddirectionsi.e.,SouthwestandWestinthe10kmbufferaroundKutupalongandsurroundingrefugeecampareafrom2015to2017.TheY-axisrepresentsNDVIrangingbetween+1and)]TJ/F195 8.9664 Tf 1.02 0 0 1 207.472 541.423 Tm [(1.TheX-axisrepresentsthedistanceofeachbuffermincrements.HigherNDVIvaluesindicatehealthy,greenvegetationwhilelowervaluescorrespondwithstressed,depletedvegetationorbarrenland.Theabovegraphindicatesthatvegetationgreenness/healthwaspersistentinthepre-inuxperiodi.e.,pre-august2017;however,vegetationhealthandbiomassdeclinedsignicantlyinthepost-inuxperiodinthetwomostaffecteddirectionsofSouthwestandWestoftheKutupalongBalukhalicamps. 3.3.RefugeeCampExpansionandForestCoverChangeSurroundingtheNayaparaLedaandUnchiprangAreasThesecondlargestsettlementsitestudiedhereistheNayaparaLedacamp,whichisinhabitedbynearly34,000Rohingyarefugees.BetweenDecember2016andDecember2017,thelandcovermapshowsthatlandoccupiedbyrefugeecampsexpandedfrom29hectaresto133hectares,whiletotalforestcoverdeclinedatapproximately176hectareswithina4-kmbufferzoneSeeTable3.DuetohigherelevationonthewestsideofthecampandtheNafRivertotheeast,therefugeecampcouldonlyexpandinthenorthandsouthdirections.Assuch,thecampsexpandedtowardthenorthandsouthbycreatinganarrowswathofsettlementsbetweenhilly-forestedlandandtheNafRiver.AnothersitestudiedhereisUnchiprang,whichaccommodatesapproximately23,000refugees.ThisrefugeecampwascreatedspontaneouslybylevelingforestedhillsaftertheAugust2017massinuxofRohingyarefugees.Thelandcoverclasseswithina3kmbufferofthiscampscentersuggestthat47hectaresofforestcoverwaslostbetweenDecemberof2016and2017.Thisdegradationwascausedfromincreasingcampexpansionhaandothernonforestactivitiesha.Foresttocampconversionandforesttononforestaccountfor19and158hectaresofforestlossrespectively,whilenonforesttocamptransformationaccountsforonly14hectaresconversionintheUnchiprangcamparea.Overall,thethreelargestrefugeesitesstudiedhereshowbothincreasingpopulationandsettlementexpansionintheforestedareaswhichdegradedsubstantiallyoverthestudyperiod.

PAGE 14

RemoteSens. 2018 10 ,689 14of20 Figure7.Landcoverclassconversionmapwithmajorlandconversionclassesi.e.,foresttocampmagenta,foresttononforestoff-white,andnonforesttocamporange,depictinglandcoverconversionandnonconversionbetweenDecember2016andDecember2017atthreerefugeecampsites: A KutupalongBalukhali, B Unchiprang,and C NayaparaLedaexpansionsites.

PAGE 15

RemoteSens. 2018 10 ,689 15of20 4.DiscussionInthisstudy,weutilizedaRFmodelbasedonadevelopeddatasetof16explanatoryvariablesandhighresolutionmultispectralsatelliteimageryfromSentinel-2Aand-2Btoclassifytwotime-steplandcovermapstoobservethegrowthofrefugeesettlementsandassociatedforestlossintheTeknafregionofBangladesh.Thelandcovermapsoftwotime-stepswiththreebroadlandcategoriesderivedfromtheRFmodelshowpromisingresultswhichareechoedbythehighoverallclassicationaccuracy.Thendingsalsohighlightthatbyusingexpertknowledgeandaniterativeanalysisprocess,theproductionofasatisfactorylandcovermapwithadesirableoutcomeispossible.AlthoughRFmodelsdonotrequireseparatevalidation,weemployedindependentcrossvalidationinGISusinghighresolutionQuickBirdimagesandGoogleEarthforthepre-inuxperiodsi.e.,2016landcovermaps.However,forthelandcovermapof2017,wehadtorelyonatruecolorcompositeSentinel-2Bimageandground-truthtrainingsamplessincehighresolutionGoogleEarthimageryfortheareawasunavailableafterJanuary2016.Evenwiththeincreasingavailabilityofsophisticateddataanalysistoolsandimprovedspatialandspectralresolutionofremotelysenseddata,monitoringforestcoverchangeremainschallenging.Thisisespeciallytruefortropicalareas,astheyareconstrainedbymultiplefactorsincludingpersistentcloudcover,highrainfallandtemperaturevariability,andthespatialandspectralresolutionofimagerywithlowavailabilityofuseableopticalimageryduringtherainyseasonespeciallyduringthemonsoon,allofwhichpresentachallengeforinterannualanalysisandtimelydetectionofnewlychangedareas.Ouranalysismayhavebeenimprovedifweabletousesatelliteimageryfordatesclosertothemajorrefugeeinuxevent,i.e.,betweenthemonthsofSeptemberandAugustof2017andonward;however,persistentcloudcoveroverthestudyarearestrictedustodryseasoni.e.,Decemberimageanalysis.Therearemultiplespontaneousrefugeecampswithinandoutsideofthestudyarea,someofwhichareinhostcommunitiesandothersconsistingofrefugeesresidingintemporaryshelters.Inaddition,alargequantityofrefugeecampswereidentiedduringoureldvisitassettlementsinmoreremoteforestedlocations,whicharenotdetectableusingmoderateresolutionimageryduetotreeshade.Further,someareasinthenorthandnortheastsectionsofthestudyareawereclassiedasforestwhichshouldbehomesteadvegetation.Hence,ourresultsmayhavesufferedfromtheselimitations.However,thetrenddocumentedhereshowsthattheareahasbeenexperiencingaradicallandcoverchangewithforestdegradationduetothesuddenexpansionofrefugeecampsandtheirassociatedactivities.Suchanunprecedentedmassexodusofrefugeesputsextremepressureonthesocial,economic,andecologicalfabricoftheentirenation,whichisalreadyafictedbyoverpopulationandpoverty,coupledwithincreasingfrequencyofclimaticandenvironmentalhazards.TheareawherethegreatestconcentrationofRohingyarefugeesaresettlinginmakeshiftcampsisahighlysensitiveecologicalregion,containingaprotectedforestforendangeredanimals.Themagnitudeandrateoftherecentinuxhascreatedenormouspressureonthenaturalresourcesandhasalreadysubstantiallyalteredthelocallandscape.Withthelargeinuxesofrefugeeentrants,thereisahugedemandforrewood,whichisincreasingdaily.Thenewarrivalsdemand750,000kgoffuelwoodeveryday[53],andmeetingthisneedisputtingextremepressureoncuttingprotectedforestsandsocialforestrytrees[11].Althoughvariousestimatessuggestrefugeeshavebeenthecauseofthestrippingawayof4000acresofforestland[7,9,10],ouranalysisusingsatelliteimageryestimatesapproximate5650acresofforestedlandsurroundingthreeresettlementcampswerelostsinceDecember2016.Ifthecurrentpatternofrefugeecampexpansionpersistsandthevolumeofforestclearingcontinues,wefearthattheareawillsoonbecomeabarrenlandandthehillyforestwillceasetoexist.Inaddition,thewipingoutofvegetativecoverandtheherbaceouslayerfromthehills,alongwithrampanthill-cutting,maytriggerlandslidesduringthemonsoonseason,whichmaycostthelivesofmanyrefugeesandresultinfurtherenvironmentaldegradationFigure8.Wildlifeintheareahasalsobeenaffectedbytheencroachment.Themostdramaticimpactisalossofhabitatforthousandsofspeciesthediversearrayofanimalsandplantsthatliveintheforest.Duetospaceshortageinthearea,manyrefugeeshavebeencampingtwotothreekilometersinthedeepforests,blockinganelephantcorridor.Suchexpansionintotheforestedlandcausesnotonlyecologicaldamage

PAGE 16

RemoteSens. 2018 10 ,689 16of20butalsoputsthelivesofRohingyarefugeesatfurtherriskfromwildlifeencounters,assixrefugeeshavealreadybeentrampledtodeathandnumbersinjuredbywildelephantsasofJanuary2018.Astherefugeecampscontinuetogrowinsizeandnumber,therewillbefurtherecologicalandsocioeconomicimpacts.Manyrefugeecampshaveexpandedintoforestedareasoralongmountainousregionsandarebuiltnexttootherbuilt-upareaswhichwilladditionallyimpactpreexistingcommunities.AlthoughlocalBangladeshipeoplearebeingsympathetictotheplightoftheRohingyarefugees,thecontinuedinflux,however,hasfueledconcernsamongthelocalpopulacewhofearthattherefugeeswilldrasticallyalterthelandscapeofthecoastaldistrictandpopulationconfigurationintheregion.Alongwithoureldinvestigation,variousreportsindicatethepriceofdailyessentialcommoditiesincludingrice,vegetables,andoilsamongothershavesoaredsincethecrisiserupted.Localtransportationcostshaveclimbed,makingconditionsmoredifcultfordailywagers,andmanyfearlosingtheirjobs,astherefugeesarewillingtodothesameworkforlowerwages.Inaddition,iftherefugeecrisiscontinuesandisnotsolvedimmediately,itmayadverselyimpactthetourismindustryintheregion,astheforestandsandybeachesarethemainattractionsandmasstouristdestinationsofthenation[54].Thereisalsoapotentialriskofariseinlocalandinternationalterrorismactivitiesasthesevulnerablepeoplescanbeaneasytargetofvestedgroups[55].Thiswillputthecountyinfurtherdangerofhomegrownterrorismwhichmaydestabilizethewholeregionaswell.Inaddition,theareamayseetheproliferationofasyntheticdrugcalledYaba,whichisimportedfromMyanmarthroughtheborderbyRohingyarefugees,intensifyingtheriskofhumantrafckingandprostitution.Thereisalsofearofanincreaseintheincidenceandtransmissionofinfectiousdiseasesthatmayspreadoutintheregion,suchaswater-bornepathogens,ascrucialgroundwatersuppliesaredepletedandcontaminated[56]. Figure8.PhotographsofrefugeecampsattheBalukhalisite.AMakeshiftsettlementsonrazedhillsandthehillbase,showingtheecologicaldamageandpotentialthreatofmudslidesduringthemonsoonseason.BAmakeshiftsettlementinsidethedegradedforestedland,depictingdestructionofhabitatandnaturalenvironment.Source:Collectedduringeldsurveyon28December2017.Theconsistent,methodicalandescalatingpatternofkillings,torture,rape,andarsonagainsttheRohingyaminoritygroup,whowerepreviouslylivinginadenialstateoverthelastdecades,isaclearindicationofethniccleansingandanactofgenocidecommittedbyMyanmar[3].ThelatestevidencepresentedbyMdecinsSansFrontiresMSFshowsthatnearly9000Rohingyahavebeenkilled[57]and354villagesburnedinthenorthernRakhineState[58]sinceviolenceeruptedinAugust2017atthehandsofthelocalmilitantsandtheBurmesearmy,intensifyingtheuncertaintyofthefateoftheseminoritygroups.ThemassinuxofrefugeestoTeknafwithinsuchashortperiodoftimehasmadethisregionofBangladeshtheworld'sfastestgrowingrefugeecrisisinrecenthistory.Inadditiontothehumanitariancrisisthathasensued,anenvironmentalcatastropheisoccurringinthearea.HencethegovernmentofBangladeshshouldtakeimmediateandconcretemeasurestorelocatetherefugeesorrepatriatethemusingdiplomaticchannelswithMyanmarandbroader,international

PAGE 17

RemoteSens. 2018 10 ,689 17of20platformsbyincludingtheUnitedNations,ASEAN,OIC,SARCandotherappropriateparties[59].WhilerepatriatingthemtothehandofMyanmar,allpartiesshouldbeincludedtoensuretheirsafetyintheirreturntotheirhomebyguaranteeingcitizenshipandequalaccesstoallcivicfacilitiesthatothercitizensenjoyinthecountry.IfrefugeesremaininBangladesh,theymustberelocatedtoothersafeplacesimmediatelyandalternativefuelsourcesprovidedfortheRohingyasothattheforestanditsresourcesarepreserved.Inaddition,thedataresultsfromthisstudymaybeusedtoprojectfuturegrowthofrefugeecamps,reducedeforestationandenvironmentalimpacts,andhelpplanmoreorganizedandstablelivingconditionsfortheRohingyarefugeesthroughspecialprotection,conservation,andsustainablepracticesoflandscape,wildlife,andimportanceoftourism. 5.ConclusionsAnunprecedentedinuxofRohingyarefugeesintosoutheasternBangladeshisputtingtheecologicallyfragileregiononthebrinkofanenvironmentaldisaster.BasedonremotesensingdataandanonparametriclandcoverclassicationtechniquesuchasRF,thisstudydocumentedlandcoverchangeandforestcoverdegradationresultingfromRohingyarefugeesettlementexpansionbetweenpre-augustandpost-augustinuxesofAugust2017.EmployingRFasanimageclassicationapproachforthisstudywithacross-validationtechnique,weobtainedahighoverallclassicationaccuracyof94.53%and95.14%for2016and2017landcovermaps,respectively,withoverallkappastatisticsof0.93and0.94.Theproduceranduseraccuracyforforestcoverrangedbetween92.98.21%and96.49.98%,respectively.Ourlandcovermapsproducedfromthisstudy,theevidenceobtainedfromgroundobservations,photosandhigh-resolutionspatialvideos,andtheanalysisofvariousonlinestudies,suggeststhatrefugeesaredestroyingtheforestecosystemsbytherampantandswiftclearingoftheforestedhills.Asmanyassevenreserveforests,totalingabout5650acres,havebeendamagedfromtheerectionofmakeshiftshelters,burningofrewood,andanthropogenicactivitiesrelatingtosubsistenceneedsoftherefugees.Asaresult,landsthatwereformerlyvegetatedandforestedarenowconvertedtorefugeecampsaspopulationsurgentlyseekshelterandsafetyinanareaunequippedandunpreparedtodealwiththecrisis.RemotesensingdataandRFlandcoverclassicationprovedefcientandvaluableinquantifyingtheeffectsofrefugeecampsandassociatedhumanactivitiesonthesurroundingenvironment,providingevidenceofthenegativecorrelationbetweencampexpansionandadverseimpactsonthenaturalsurroundings.Theresultsindicatedthatenvironmentaldestructionnamelylossofforestedlandandothervegetationthathousedendangeredanimals,biodiversityandecosystemsandtheirserviceshasoccurredatanalarmingrateinthelastvemonthsi.e.,August2017toDecember2017.Suchdegradationofthesecriticalecologicalresourcesmighttriggermultiplicativeimpactsontheenvironment,biodiversity,wildlifehabitatandoverallsocioeconomichealthoftheentireregion.Ifnomeasuresaretakennoworinthenearfuturetoprotectthevegetationcover,forests,andoveralllocalenvironment,therewillbelong-termandirreparabledamagethatmaycauselargerproblemsforthecountryaswell.Giventhespeed,numberofmigrants,andspontaneousnatureoftherecentrefugeeinux,thegovernmentofBangladeshandtheBangladeshipeopleareunpreparedtoaccommodatetheswellingrefugeepopulations,lackingthemeanstoeffectivelyplancampstructuresandlimitenvironmentalimpacts.Hencetheresultinggeographicalinformationandthematicmapwithemployedmethodologyproducedfromthisstudymayprovideausefultoolforpolicymakersandconcernedauthoritiestoassesstheenvironmentalimpactsoflargescalerefugeemovementsandconcentrationsinthecontextofeffectivecrisismanagement.Inaddition,theenvironmentaleffectsofforcedmassmigrationsshouldbestudiedanddocumentedsothatnationsandtheworldmaybetterprepareandamelioratetheadverseoutcomesoftheever-increasingphenomenaofrefugeecampsandtemporarysettlementsintheworld.

PAGE 18

RemoteSens. 2018 10 ,689 18of20 AuthorContributions:Inthisarticle,MohammadMehedyHassanwasresponsibleforresearchdesign,landcoverclassication,analysisandwritingthepaper.MunshiKhaledRahmancollectingelddataincludingtrainingsamplesandcaptureofhigh-denitionspatialvideo.AudreyCulverSmithcontributedtothewritingandediting.KatherineWalkercontributedtothepaperdiscussionsection,andJaneSouthworthprovidedguidance,andeditingofthepaper. Acknowledgments:PublicationofthisarticlewasfundedinpartbytheUniversityofFloridaOpenAccessPublishingFund. ConictsofInterest: Theauthorsdeclarenoconictsofinterest. References 1.UnitedNationsHighCommissionerforRefugeesUNCHR.StatisticalYearBook,FigureataGlance.Availableonline:http://www.unhcr.org/en-us/gures-at-a-glance.htmlaccessedon1February2018. 2.TheBritishBroadcastingCorporationBBC.MyanmarRohingya:WhatYouNeedtoKnowabouttheCrisis.Availableonline:http://www.bbc.com/news/world-asia-41566561accessedon1February2018. 3.HumanRightsWatch.RohingyaCrisis.Availableonline:https://www.hrw.org/tag/rohingya-crisisaccessedon1February2018. 4.CableNewsNetworkCNN.TheRohingyaCrisis.Availableonline:https://www.cnn.com/specials/asia/rohingyaaccessedon1February2018. 5.TIME.Myanmar'sCrisis,Bangladesh'sBurden:AmongtheRohingyaRefugeesWaitingforaMiracle.Availableonline:http://time.com/5031342/bangladesh-myanmar-rohingya-refugee-crisis/accessedon1February2018. 6.InternationalOrganizationforMigrationIOM.IOMBangladesh:RohingyaRefugeCrisisResponse.Availableonline:https://www.iom.int/sites/default/les/situation_reports/le/Bangladesh_SR_20180119-25.pdfaccessedon27January2018. 7.REUTERS.BangladeshCarvingOutForestLandtoShelterDesperateRohingya.Availableonline:https://www.reuters.com/article/us-myanmar-rohingya/bangladesh-carving-out-forest-land-to-shelter-desperate-rohingya-idUSKBN1CA0ZNaccessedon4February2018. 8.USATODAY.ElephantsandRohingyaMuslimRefugeesJostleforSpaceinBangladesh.Availableonline:https://www.usatoday.com/story/news/world/2018/01/18/elephants-and-rohingya-muslim-refugees-bangladesh/1043296001/accessedon4February2018. 9.bdnews24.com.NewlyArrivedRohingyaRefugeesDamagedTk1.5bnBangladeshForest,GovtSays.Availableonline:https://bdnews24.com/environment/2017/10/11/newly-arrived-rohingya-refugees-damaged-tk-1.5bn-bangladesh-forest-govt-saysaccessedon4February2018. 10.DailySun.DestructionofForestsbyRohingyas.Availableonline:http://www.daily-sun.com/post/265211/Destruction-of-forests-by-Rohingyasaccessedon4February2018. 11.DhakaTribune.RohingyaInux:15-Year-OldForestationProjectDestroyedin57Days.Availableonline:http://www.dhakatribune.com/bangladesh/2017/10/21/rohingya-inux-15-year-old-forestation-project-destroyed-57-days/accessedon4February2018. 12.Moslehuddin,A.Z.M.;Rahman,M.A.;Ullah,S.M.A.;Moriyama,M.;Tani,M.Physiography,Forests,andPeopleinTeknaf.In DeforestationintheTeknafPeninsulaofBangladesh ;Tani,M.,Rahman,M.,Eds.;Springer: Singapore,2018. 13.Karim,M.N.Localknowledgeofindicatorbirds:Implicationsforcommunity-basedecologicalmonitoringinTeknafgamereserve.InConnectingCommunitiesandConservation:CollaborativeManagementofProtectedAreasinBangladesh;Fox,J.,Bushley,B.R.,Miles,W.B.,Quazi,S.A.,Eds.;East-WestCenter:Honolulu,HI,USA,2008;pp.139. 14.Khan,M.A.S.A.;Mukul,S.A.;Uddin,M.A.;Kibria,M.G.;Sultana,F.TheuseofmedicinalplantsinhealthcarepracticesbyRohingyarefugeesinadegradedforestandconservationareaofBangladesh.Int.J.Biodivers.Sci.Manag. 2009 5 ,76.[CrossRef] 15.Pan,Y.;Birdsey,R.;Fang,J.;Houghton,R.;Kauppi,P.;Kurz,W.;Phillips,O.;Shvidenko,A.;Lewis,S.;Canadell,J.;etal.AlargeandpersistentcarbonsinkintheWorld'sforests.Science2011,333,988.[CrossRef][PubMed] 16.Alam,M.F.;Uddin,M.Z.;Hasan,M.A.EvaluationofPlantBiodiversityinTeknafWildlifeSanctuary,Bangladesh;LAPLAMBERTAcademicPublishing:Saarbrcken,Germany,2012.

PAGE 19

RemoteSens. 2018 10 ,689 19of20 17.InternationalUnionforConservationofNature.ConservationofAsianelephantsinBangladesh;TheWorldConservationUnion:Dhaka,Bangladesh,2004. 18.Rahman,M.Z.LivelihoodsofRohingyasandTheirImpactsonDeforestation.InDeforestationintheTeknafPeninsulaofBangladesh ;Tani,M.,Rahman,M.,Eds.;Springer:Singapore,2018. 19.Liu,J.;Linderman,M.;Ouyang,Z.EcologicalDegradationinProtectedAreas:TheCaseofWolongNatureReserveforGiantPandas. Science 2001 292 ,98.[CrossRef][PubMed] 20.Islam,K.R.;Weil,R.R.LanduseeffectsonsoilqualityinatropicalforestecosystemofBangladesh.Agric.Ecosyst.Environ. 2000 79 ,9.[CrossRef] 21.Newell,G.R.;Stavins,R.N.ClimateChangeandForestSinks:FactorsAffectingtheCostsofCarbonSequestration. Environ.Econ.Manag. 2000 40 ,211.[CrossRef] 22.Morales-Hidalgo,D.;Oswalt,S.N.;Somanathan,E.Statusandtrendsinglobalprimaryforest,protectedareas,andareasdesignatedforconservationofbiodiversityfromtheGlobalForestResourcesAssessment2015. For.Ecol.Manag. 2015 352 ,68.[CrossRef] 23.Romijn,E.;Lantican,C.B.;Herold,M.;Lindquist,E.;Ochieng,R.;Wijaya,A.;Murdiyarso,D.;Verchot,L.Assessingchangeinnationalforestmonitoringcapacitiesof99tropicalcountries.For.Ecol.Manag.2015,352 ,109.[CrossRef] 24.Hassan,M.M.Monitoringlanduse/landcoverchange,urbangrowthdynamicsandlandscapepatternanalysisinvefastesturbanizedcitiesinBangladesh.RemoteSens.Appl.Soc.Environ.2017,7,69.[CrossRef] 25.Southworth,J.;Zhu,L.;Bunting,E.;Ryan,S.J.;Herrero,H.;Waylen,P.R.;Hill,M.J.Changesinvegetationpersistenceacrossglobalsavannalandscapes,1982. J.LandUseSci. 2014 11 ,7.[CrossRef] 26.Zhao,M.;Running,S.W.Drought-inducedreductioninglobalterrestrialnetprimaryproductionfrom2000through2009. Science 2010 329 ,940.[CrossRef][PubMed] 27.Neumann,M.;Moreno,A.;Thurnher,C.;Mues,V.;Hrknen,S.;Mura,M.;Bouriaud,O.;Lang,M.;Cardellini,G.;Thivolle-Cazat,A.;etal.CreatingaRegionalMODISSatellite-DrivenNetPrimaryProductionDatasetforEuropeanForests. RemoteSens. 2016 8 ,554.[CrossRef] 28.Zhao,X.;Heidinger,A.K.;Walther,A.ClimatologyAnalysisofAerosolEffectonMarineWaterCloudfromLong-TermSatelliteClimateDataRecords. RemoteSens. 2016 8 ,300.[CrossRef] 29.Bjorgo,E.Usingveryhighspatialresolutionmultispectralsatellitesensorimagerytomonitorrefugeecamps.Int.J.RemoteSens. 2000 21 ,611.[CrossRef] 30.Hagenlocher,M.;Lang,S.;Tiede,D.IntegratedassessmentoftheenvironmentalimpactofanIDPcampinSudanbasedonveryhighresolutionmulti-temporalsatelliteimagery.RemoteSens.Environ.2012,126,27.[CrossRef] 31.Lang,S.;Tiede,D.;Hlbling,D.;Freder,P.;Zeil,P.EarthobservationEO-basedexpostassessmentofinternallydisplacedpersonIDPcampevolutionandpopulationdynamicsinZamZam,Darfur.Int.J.RemoteSens. 2010 31 ,5709.[CrossRef] 32.Sprhnle,K.;Kranz,O.;Schoepfer,E.;Moeller,M.;Voigt,S.Earthobservation-basedmulti-scaleimpactassessmentofinternallydisplacedpersonIDPcampsonwoodresourcesinZalingei,Darfur.GeocartoInt.2016 31 ,575.[CrossRef] 33.Lodhi,M.A.;Echavarria,F.R.;Keithley,C.UsingremotesensingdatatomonitorlandcoverchangesnearafghanrefugeecampsinnorthernPakistan. GeocartoInt. 1998 13 ,33.[CrossRef] 34.Langer,S.;Tiede,D.;Lthje,F.Long-termMonitoringoftheEnvironmentalImpactofaRefugeeCampBasedonLandsatTimeSeries:TheExampleofDeforestationandReforestationduringthewholeLifespanoftheCampLukole,Tanzania. GI_ForumJ.Geogr.Inf.Sci. 2015 ,1.[CrossRef] 35.UnitedNationsHighCommissionerforRefugeesUNCHR.PopulationStatistics.Availableonline:http://popstats.unhcr.org/en/time_seriesaccessedon1February2018. 36.InterSectorCoordinationGroupISCG.SituationReport:RohingyaRefugeCrisis.Cox'sBazar.Availableonline:https://reliefweb.int/sites/reliefweb.int/files/resources/180225_weeklyiscg_sitrep_final.pdfaccessedon27March2018. 37.BangladeshBureauofStatisticsBBS.BangladeshPopulationandHousingCensus2011;CommunityReport:Cox'sBazar;StatisticsandInformaticsDivision,MinistryofPlanning,GovernmentofBangladesh:Dhaka,Bangladesh,2014.Availableonline:www.bbs.gov.bdaccessedon13February2018.

PAGE 20

RemoteSens. 2018 10 ,689 20of20 38.UnitedNationsOfcefortheCoordinationofHumanitarianAffairsUNOCHA.RohingyaRefugeeCrisis.Availableonline:https://www.unocha.org/rohingya-refugee-crisisaccessedon13February2018. 39.Mills,J.W.;Curtis,A.;Kennedy,B.;Kennedy,S.W.;Edwards,J.D.Geospatialvideoforelddatacollection.Appl.Geogr. 2010 30 ,533.[CrossRef] 40.Rahman,M.K.;Schmidlin,T.W.;Munro-Stasiuk,M.J.;Curtis,A.GeospatialAnalysisofLandLoss,LandCoverChange,andLandusePatternsofKutubdiaIsland,Bangladesh.Int.J.Appl.Geospat.Res.2017,8,45.[CrossRef] 41.Jensen,J.R. IntroductoryDigitalImageProcessing ,3rded.;PrenticeHall:UpperSaddleRiver,NJ,USA,2005. 42.Chrysa,I.;Mallinis,G.;Gitas,I.;Tsakiri-strati,M.EstimatingMediterraneanforestparametersusingmultiseasonalLandsat8OLIimageryandanensemblelearningmethod.RemoteSens.Environ.2017,199,154.[CrossRef] 43.Sesnie,S.E.;Finegan,B.;Gessler,P.E.;Thessler,S.;Bendana,Z.R.;Smith,A.ThemultispectralseparabilityofCostaRicanrainforesttypeswithsupportvectormachinesandRandomForestdecisiontrees.Int.J.RemoteSens. 2010 31 ,2885.[CrossRef] 44.Belgiu,M.;Dragut,L.Randomforestinremotesensing:Areviewofapplicationsandfuturedirections.ISPRSJ.Photogramm.RemoteSens. 2016 114 ,24.[CrossRef] 45.Hastie,T.;Tibshirani,R.;Friedman,J.TheElementsofStatisticalLearning:DataMining,InterfaceandPrediction;Springer:NewYork,NY,USA,2009. 46.Tucker,CJ.;Townshend,J.R.G.Strategiesformonitoringtropicaldeforestationusingsatellitedata.Int.J.RemoteSens. 2000 21 ,1461.[CrossRef] 47.Breiman,L.Randomforests. Mach.Learn. 2001 45 ,5.[CrossRef] 48.Hassan,M.M.;Southworth,J.Analyzinglandcoverchangeandurbangrowthtrajectoriesofthemega-urbanregionofDhakausingremotelysenseddataandanensembleclassier.Sustainability2017,10,10.[CrossRef]49.Anderson,J.R.;Hardy,E.E.;Roach,J.T.;Witmer,R.E.ALandUse/LandCoversClassicationSystemforusewithRemoteSensorData ;GeologicalSurveyProfessionalPaper;USGS:Washington,DC,USA,1976. 50.Qian,Y.;Zhou,W.;Yan,J.;Li,W.;Han,L.Comparingmachinelearningclassiersforobject-basedlandcoverclassicationusingveryhighresolutionimagery. RemoteSens. 2015 7 ,153.[CrossRef] 51.Hapfelmeier,A.;Ulm,K.VariableselectionbyRandomForestsusingdatawithmissingvalues.Comput.Stat.DataAnal. 2014 80 ,129.[CrossRef] 52.Akar,.;Gngr,O.ClassicationofmultispectralimagesusingRandomForestalgorithm.J.GeodesyGeoinf.2012 1 ,105.[CrossRef] 53.FinancialExpress.InuxofRohingyaRefugeesEntailsaSeriousBurdenontheEconomy.Availableonline:https://thenancialexpress.com.bd/views/views/inux-of-rohingya-refugees-entails-a-serious-burden-on-the-economy-1513519560accessedon4February2018. 54.Hassan,M.M.;Shahnewaz,M.MeasuringTouristServiceSatisfactionatDestination:ACaseStudyofCox'sBazarSeaBeach,Bangladesh. Am.J.Tour.Manag. 2014 3 ,32.[CrossRef] 55.Rahman,U.TheRohingyarefugee:AsecuritydilemmaforBangladesh.J.Immigr.Refug.Stud.2010,8,233.[CrossRef] 56.WorldHealthOrganizationWHO.MortalityandMorbidityWeeklyBulletinMMWBCox'sBazar,Bangladesh.VolumeNo8.26NovemberDecember2017.Availableonline:http://www.searo.who.int/bangladesh/nalmmwbvol8.pdf?ua=1accessedon4February2018. 57.MdecinsSansFrontiresMSF.Myanmar/Bangladesh:MSFSurveysEstimatethatatLeast6700RohingyaWereKilledduringtheAttacksinMyanmar.Availableonline:http://www.msf.org/en/article/myanmarbangladesh-msf-surveys-estimate-least-6700-rohingya-were-killed-during-attacksaccessedon30January2017. 58.HumanRightsWatch.Burma:40RohingyaVillagesBurnedSinceOctober.Availableonline:https://www.hrw.org/news/2017/12/17/burma-40-rohingya-villages-burned-octoberaccessedon30January2017. 59.Faruk,H.;Imran,A.;Mian,N.TheRohingyaRefugeesinBangladesh:AVulnerableGroupinLawandPolicy. J.Stud.Soc.Sci. 2014 8 ,226. 2018bytheauthors.LicenseeMDPI,Basel,Switzerland.ThisarticleisanopenaccessarticledistributedunderthetermsandconditionsoftheCreativeCommonsAttributionCCBYlicensehttp://creativecommons.org/licenses/by/4.0/.