ChinaFocus:AIalgorithmmapsoutdetailedcarbonemissionsforcitymanagers_GuangmingOnlineBEIJING,April23(Xinhua)--AteamofChinesescientistsareusingartificialintelligence(AI)techtoaccuratelymapoutcarbonemissionsinbigcities.Thisisanattemptthatmayhelpurbanmanagersmakemoreevidence-basedplansinfightingglobalwarming.ItispartofChina'seffortstopeakitscarbondioxideemissionsby2030andachievecarbonneutralityby2060.TheexperimentaltoolofresearchersfromtheAerospaceInformationResearchInstitute(AIR)undertheChineseAcademyofSciencesisarefittedcarbon-monitoringvehiclewhichtakesajoy-rideregularlyatbustlingzonesinthesouthernChinesecityofShenzhen.Atanexperimentalday,awhitecylinderroseslowlyontopofavantostartsensor-enabledcarbonemissionmonitoringinthecity.Thecarbonmonitorwassynchronizingthecollectionofcarbondioxideemissionconcentrationontheroad.Differentfromordinarycarbonmonitors,thevehiclewasalsomountedwithpanoramiccamerastocapturestreetscenes.Thefirst-handdatawastransmittedtoalaboratoryatAIRinBeijing,whereabrand-newdeeplearningalgorithmisbeingtrained.Drivenbytheteam'sdeeplearningalgorithm,thosecamerascoulddiscerncarbonemissionandsinksources,suchasmovingcars,buildingsandcarbon-fixingvegetation,inrealcomplexscenarioses.ThentheAImodelcanproduceestimatesonthecarboncontributions,eitherpositiveornegative,ofthosesources."Takeasinglesiteasanexample.TheAIanalysisrevealedthatsmallcarscontributethemosttoemissions,followedbytrucks,"saidWangLi,aresearcherfromAIR."Thesurroundingroadsandbuildingsalsocontributetoemissions."Currently,Shenzhenhassixcarbonmonitoringtowers,butthat'sinadequatetomakeadetailedemissionmapandthebuildingofmoretowerswillbecostly,makingalternativeandcomplementarymonitoringmethodsnecessary,saidZhangYonglin,aresearchassistantinWang'steam.ItoccurredtoZhang,whohasacomputersciencebackground,thattheAIcouldlendahandandhedevelopedthealgorithm."Youhaveatleast100elementsatonesinglesite,andtheobjectskeepchanging,"saidZhang."OnlyAIcancalculateviableresultsviaaninversemodeling."Now,atotalof650kilometersdrive-aroundjourneyhasbeencompleted,obtainingmorethan100,000recordsofstreetimagesandroadtrafficcarbonemissionsinShenzhen.Aftertraining,theAImodelcanproduceahigh-definitionmapwithinadistanceof100metersanditsaccuracyforroadtrafficcarbonemissionintensityhasexceeded92percent,accordingtoZhang.Themodelisprojectedtoprovideafineguideforcitymanagers,likewhatcausesacertainblocktoseehighercarbonemissionsandpossiblesolutionsbehindit.That'sanelaborateworktheremote-sensingcarbonsatellitesandmonitoringtowerscannotdo."Theresultscanremindpolicy-makerstotakemeasuressuchasincreasingefficiencyofaccessincertainroadsections,orplantingmoretreesatablock,"saidZhang."Sometimes,toomuchgreeneryisnotallbeneficialforreducingcarbonconcentrationwhenitresultsinapipelineeffect,"headded."That'swhattheAImapcantellyou."InMarch,theMinistryofEcologyandEnvironmentsetagoalinbuildingamodernizedecologicalenvironmentmonitoringsystemby2035viastepped-upeffortsfordigitalandintelligentupgrading.NewtechnologiesincludingAIandInternetofThingswillbeappliedtoestablishanintelligentmonitoringsystem,accordingtotheministry.Editor:DisclaimerTheviewsandopinionsexpressedinthisarticlearethoseoftheauthor's,GMW.cnmakesnorepresentationsastoaccuracy,suitability,orvalidityofanyinformationonthissiteandwillnotbeliableforanyerrors,omissions,ordelaysinthisinformation.