Quanti-litative Revolution in GIS

Quanti-litativeRevolutioninGIS

Pre-NextLabLicenceinMaths,ULg

LicenceinComputerSciences,ULg

Geospatialspecialization,ULg

ManyyearsinGIS(IonicSoftwareanditsmutations)

Alwaysindatamanagementandsemantic

Newtechnologiesenthusiast

NextLabageProjectin fortheSPW

Projectin forVirdata

Wrotea aboutPlay!2

evangelist(OKFN,...), speaker

Distributed +Distributed expert

co-founder| organizer

Andthelistgoeson...andon...

@Noootsab

@NextLab_be

QualitativeGISCoinedin1963

Rastervulgus:BigFatMultiBandImages

SPOT(seelater)IKONOS(greek:image)LANDSATALOSQuickBird:resolution64cm→2m44cm

Vectorvulgus...nah,vector

FeatureextractionLandmeters(rare)Fieldsurveying~GPS(GNS)

hence,Producerswere...

Govs!

thus,Userswere

Govs!

Ohyeah,researchersaswell...huh!?wait!

Okay,veryfeeeewprivates

inconsequence,AnalyseswerePerformedbyexperimentedfolks......ornerds

finally,Toolswere...Youdon'tevenknowthem!

ESRI,OracleSpatial,OGC,RedSpider(yeah),...

QualitativeGIS"Started"around2000

Actually,notreallybefore2010

GISRevolutionInducetheoriesfromdata

Theriverhasfloodedon3km

Theriverwillfloodifitrainsmorethan...

NewproducersGoogle,4²,OSM,GPS

old:Mappy,Michelin,...

NewUsersDevs,Lambda

Billioooooons

AnalysesRouting,AccurateGWR¹,...

Disasterprediction,floodprevention,crashprobability,...

¹GeospatialWeightRegression

Tools?Userfriendlyand/orOpenSource

OSMGeoServer

GoogleMapsOpenLayers

QuantilitativeGISCoinedby@Noootsab^^

WTF?!

Rasterstoday(f.i.Spot7)RedGreenBlue+NearInfraRedBandsOneshot:60x60km(3.600sqkm)Takes3.600.000sqkmofgeodataperDAYResolution2satellitesat and2othersat

Revolution:110minutes26daystocompletethegeoid(allpiecesofcrapcovered)1singlef*****gfilefora60x60kmtileisworthupto

Vector(mostlyposition)

Twitter,Facebook,...Foursquare,InstagramWazeGoogle(initswhole)ConnectedDevices

everywhereandeverytime

Presidente

Model-Driven

Deductive

Top-Down

Quantitative

LaggedTime

Commandante

Data-Driven

Inductive

Bottom-Up

Qualitative

RealTime

GISPutschMarcelino

DataLake

MachineLearning

Variety

Value

Velocity

VOLUME?Itwasthereforages!

Who,What,HowYouthoughtIwasjoking,huh!

SocrataEvanChanusingSpark

Customershave pointdata(manymillionsof

rows)PostGIS:point-in-polygonandotherdoes

Partitioningpointdatain (tiling,Z-curve)

Partitionsinto forquickanalysis

Adding toSparkforspeedingup

AzaveaRobEmanueleonGeotrellis

Run ofclimatedataagainstdaily

temperatureandprecipitationdataoutto2009Createsuitabilitymapsoverhigh-resrasterlayersspanningthe forUrbanForestry

Modeling.

GeoTrellisisproviding withgeospatialcapabilities.

Ingest,mosaicandpyramidrasterdatainto

forfast(sub-500ms)tilefetching

SnipsRandHindiinhisLabs

Tranquilien: seatingavailabilityinpublic

transportRiskContext:Determiningtheriskofbicycleandcar

Using technologiesthatcanscalelinearly

(Akka,Scala)Thankstothebottomup ofthearchitecture,the

clusterskeepscrunchingdatainaresilientmanner

Done!Thx&cuonTwitter

@noootsab

@NextLab_be