INU- 42 THE WORLD BANK POLICY, PLANNING AND RESEARCH STAFF FILECOPY Public Disclosure Authorized
Infrastructure and Urban Development Department Environment Department
Report INU 42 Public Disclosure Authorized THE USE OF SATELLITE IMAGES FOR URBAN PLANNING
A Case Study from Karachi, Pakistan Public Disclosure Authorized
by Marie-Agnes Bertaud A
May 1989
Public Disclosure Authorized TECHNICALNOTE
This is a document publishedinformally by the World Bank.The views and interpretationsherein are thoseof the author and should not be attributedto the World Bank,to its affiliatedorganizations, or to any individualacting on their behalf. Copyright 1989 The WorldBank 1818H Street,N.W.
All RightsReserved First PrintingMay 1989
This is a documentpublished informally by the World Bank. In order that the informationcontained in it can be presentedwith the least possibledelay, the typescript has not been preparedin accordancewith the proceduresappropriate to formal printedtexts, and the WorldBank accepts no responsibilityfor errors. The World Bank does not acceptresponsibility for the views expressedherein, which are those of the authorand shouldnot be attributedto the World Bankor to its affiliatedorganizations. The findings,interpretations, and conclusionsare the resultsof researchsupported by the Bank;they do not necessarilyrepresent official policy of the Bank. The designationsemployed, the presentationof material,and any maps used in this documentare solely for the convenienceof the reader and do not implythe expressionof any opinionwhatsoever on the partof the WorldBank or its affiliatesconcerning the legal statusof any country,territory, city, area,or of its authorities,or concerningthe delimitationsof its boundariesor nationalaffiliation. This reportwas preparedby PADCO,Inc. Ms. Marie-AgnesBertaud was the mainauthor, assisted by Mr. CharlesPill of PADCO.Ms. Lorna Paceand Ms.Jennifer Allen providedmuch neededsupport in the operation of the World Bank'sremote sensing facilities. Guidance,comments and assistancewas providedby Ms. Jane Prattand Messrs.Lynn Holstein,Per Ljung,Wayne Luscombe, Steve Mayo, Glenn Morgan and PhilipeVillagram. TheWorld Bank
The Use of Satellite Images for Urban Planning
A Case Study from Karachi, Pakistan
TECHNICALNOTE
PREFACE
This report Is one elementof ongoing researchin the application of new technologiesfor urbanplanning and management,the analysisof urban land use and reconnaissanceof physicalInfrastructure.
Its preparationwas jointly sponsored by the Infrastructureand Urban DevelopmentDepartment and the EnvironmentDepartment of the World Bank, and supported by a World Bank research project (RPO 674/66). The report also draws heavilyon PADCO'swork on an UNCHS- executedand UNDP-fundedproject (PAK/86/029)to supportthe Karachi DevelopmentAuthority in Pakistan: *Strengthening the Karachi Master PlanningProcess, 1986-2000." In the Karachiproject, PADCO is associated with Pakistan Environment Planning & ArchitecturalConsultants, Ltd. (PEPAC).
THE USE OF SATELLITEIMAGES FOR URBANPLANNING Tableof Contents
PageNo INTRODUCTION 1
PART1. REMOTESENSING AND SATELLITEIMAGERY 3 A. FromAerial Photography To SatelilteImages 3 B DifferentRemote Sensing Satellite Systems 5 1 LANDSAT4 5 2 LANDSAT5 5 3 SPOTI 5
PART2. VARIOUSTYPES OF MEDIAAND WAYS OF PROCESSING 11 ANDINTERPRETING SATELLITE IMAGES A ProcessingSPOT Satelilte Images From Panchromatic 11 PhotographicPrints B Processingof SPOTSatellite Images From Multispectral 16 PhotographicPrints C ProcessingSatellite Images from Digital Tapes 20 D CostComparison Between Different Satelilte Image 20 ProcessingTechniques E CostDifference Between SPOT Satellite Images and 24 AerialPhotos PART3. USINGSATELLING IMAGES FOR URBAN PLANNING 25 AND URBANMANAGEMENT. KARACHI CASE STUDY A KarachiUrban Development from 1943to 1988 27 B ComputerizedMapping and GIS System 31 C MappingLand Use and Road Networks from Satellite 33 ImagesStarting an UrbanDatabase Useof SatelliteImages for MappingPurposes 33 ConvenientScales of PhotographicPrints made 33 from SatelliteImages VisualInterpretation of UrbanFeatures 33 MapsDerived from SatelliteImages 38 OtherMap Overlays 38 UrbanMapping and Urban Data Integration 38 D Establishinga HousingTypology Using "SPOT" 48 SatelliteImages and a set of AerialPhotographs
CONCLUSION 69 ANNEXES Annex I COMPUTEREQUIPMENT NEEDED FOR PROCESSING SATELLITE IMAGESFOR MAPPING AND FOR URBANMANAGEMENT SYSTEMS 71 A HardwareNeeded to ProcessSatellite Images B SoftwareNeeded to ProcessSatellite Images C HardwareNeeded for Mappingand Data ManagementSystems D SoftwareNeeded for Mappingand Data ManagementSystems E RegisteredTrademarks
Annex II HOW TO ACQUIRESPOT SATELUTE IMAGES 77
Annex III KARACHIHOUSING TYPOLOGY PLANSDATA AND PHOTOGRAPHSOF DIFFERENTDWELLING UNITS 81
GLOSSARY 99
REFERENCES 107 FIGURES
Figure1 1 ChronologicalLaunch and RetirementHistory of the LANDSATand 4 SPOTSeries of Satellitesfrom 1972to 1989 Figure1 2 Off-NadirViewing 7 Figure1 3 SPOT Off-NadirRevisd Capabilities 8 Figure 1 4 StereoscopicViewing Capabilities 9 FIgure1 5 DiagramIllustrating the SpatialResolution of 10 Three DdferentSatelide Images Figure2 1 'SPOT' SatelliteImage PanchromaticFilm, Scale 1 400,000 12 Karachi,Pakistan Figure2 2 Identfflcationof DdferentPatterns of Development 13 From "SPOr PanchromaticPhotographic Print at Scale 1 100,000,Karachi, Pakistan Figure2 3 'SPOT' PanchromaticPhotographic Print at Scale 1 24,000 14 Karachi,Pakistan Figure2 4 "SPOT"Panchromatic Photographic Pnnt at Scale 1 6,000 15 Karachi,Pakistan Figure2 5 Identdicationof DdferentUrban FeaturesFrom 17 'SPOT' MuitispectralPnnt, Karachi,Pakistan Figure2 6 PhotographicPrint made from "SPOTrMultispectral Film 18 Image beforeEnhancement, South Jakarta,Indonesia Figure2 7 PhotographicPrint made from "SPOT' MultispectralFilm 19 EnhancedImage, South Jakarta,Indonesia Figure2 8 RemoteSensing Satellite Image Data Processingand 22 Their ApplicationsIn Regional& Urban Planning Figure2 9 "SPOT" SatelifteImage Products- ProcessingCost 23 and ResultingOutput Figure2.10 Cost Comparisonfor Mappingwith AerialPhotos and 24 SPOTSatellite Images - iii -
Figure3.1 KarachiOld City Map - 1885 26 Figure3.2 KarachiRegion 28 Figure3.3 KatchiAbadi or SquatterSettlement - Karachi 30 Figure3.4 GeographicInformation System 32 Figure3.5 Use of SatelliteImages in Urban Planningand Urban 35 Management Figure3.6 Identificationof DifferentUrban Featuresfrom "SPOT" 36 PanchromaticPhotographic Print - Karachi,Pakistan Figure3.7 DifferentTypes of ResidentialDevelopment, 37 Karachi,Pakistan Figure3.8 ThematicLand Use Map with Major Road Network, 39 Karachi,Pakistan Figure3.9 KarachiRoad Network 40 Figure3.10 KarachiWater-Supply Network 42 Figure3.11 KarachiSewerage Network 42 Figure3.12 KarachiDrainage Network 43 Figure3.13 Land Use Type and Area Attributesare Listedon Map 44 Figure3.14(A) Attributes or Data Listed on "AUTOCAD"Map are 45 DirectlyImported to Databaseand LOTUSSpreadsheet. AUTOCADData Figure3.14(B) Attributes or Data Listed on "AUTOCAD"Map are 46 DirectlyImported to Databaseand LOTUSSpreadsheet. dBASEProgram Figure3.14(C) Attributes or DataListed on "AUTOCAD"Map are 47 DirectlyImported to Databaseand LOTUSSpreadsheet. LOTUS1-2-3 Figure3.15 Type A - SquatterSettlement 49 'SPOT"Panchromatic Satellite Image Figure3.16 Type B - High Density 50 *SPOT"Panchromatic Satellite Image Figure3.17 Type C - MediumDensity 51 "SPOT"Panchromatic Satellite Image Figure3.18 Type D - Low Density 52 "SPOT"Panchromatic Satellite Image Figure3.19 Identificationof DifferentHousing Types 54 on Aerial Photographat Scale 1:4,000 Figure3.20 DifferentResidential Types of Development 55 Figure3.21 DifferentResidential Types of Development 56 Figure3.22 DifferentResidential Types of Development 57 Figure3.23 SpaceAccommodation per PersonAmong Different 60 HousingTypes Figure3.24 Plot Size Comparisonof SelectedDwelling Units 61 Figure3.25 SquatterSettlements Along Railwaysand on 63 SwampLand Figure3.26 KatachiAbadies Locations and Projected 64 Expansionby the Year 2000 Figure3.27 HousingTypology and Land Use Map - 65 Karachi1988 Figure3.28 KatachiPopulation Distribution by 67 HousingType - iv -
TABLES
Table 3 1 AverageDensities Among DifferentTypes of 53 ResidentialDevelopment Table 3 2 HousingTypes Classification 58 Table 3 3 HousingTypes and Their Characteristics 59
ANNEXTABLES Table I 1 List of Some Image ProcessingSoftware for MicrocomputerSystems Table 111 SPOr SatelliteImage Productsand ProcessingCosts 1
INTRODUCTION THE USE OF SATELLITEIMAGES FOR URBANPLANNING
0.01 Effective urban planning requires access to accurate and continually updated informationconcerning the changingconditions of urbanareas. Originally, remote sensing using satellitemedia was not very suitablefor urbanapplications because of the relativelypoor resolution provided80m x 80m. The new generationof remotesensing systemswith resolution equal to 1Om x 1Om in panchromaticmode and 20m x 20m in multispectralmode has now made urban applications possible. In addition, advancementsin micro-computers,related hardware and softwarehave made this technologymore affordable and accessibleto urbanmanagers, planners, and engineers. 0.02 In most Third World cities, accurate up-to-date data on current land use are non-existent.This makes planning urban infrastructureand shelter programs difficult, If not impossible,since there is no knowledgeof the existingurban area. Further, the high costs and time requiredfor traditionaldata collectionefforts have madesystematic updating of urbandatabases beyondthe reach of many planningauthorities.
0.03 Landuse changesInduced by rapid urbangrowth amplify the needfor up-to-dateland use Inventories.Many citiesin the developingworld have urbanpopulation growth ratesin excess of 6 percent,doubling their populationabout everyten years.They needto monitor moreclosely the patternof urban growth to Improveplanning and activitiesto accommodatethis growth.
0.04 The lack of adequate spatial data has been a serlous constralnt on many recent Bank-financedprojects. For example,recently the Bankparticipated in the financingof a drainage masterplan for Karachi.Work on the projectwas seriouslydelayed due to the lack of adequate maps showingthe extent of the built-uparea and existingdrainage systems. The only data were aerial photoswhich could not be made easilyavailable to the project.Similar delays In obtaining accuratespatial data confront practicallyevery other Bank-financed (and other donor agencies, for that matter) urban projects.
0.05 New technologicaladvances now providea relativelyinexpensive solutlon to some of these problems.Satellite Images, such as LANDSAT,have been generatingland resourcedata over the past 16 years for such purposesas agriculture,forestry, mining,etc.; but the resolution was not sharpenough for most urbanplanning purposes. PADCO first applied LANDSATdata for urban planningin Egypt (the EgyptianNational Urban Pollcy Study in 1980-1982),but data were limitedto gross land usesand built-upurban areas. Nevertheless, this Informationwas a valuable tool In developingstructure plans for smalltowns where mapdata were Inaccurate.Unfortunately, this applicationwas limited becausethe processingcosts were still relativelyexpensive.
0.06 Since 1986,SPOT I, a satellitefinanced by France,Belgium and Sweden,has been sending Imageswith eight times the spatialresolution of LANDSATand other civilian satellites.It producesdata with a ground resolutionequal to 10 meters in panchromaticmode (black and white image) and 20 meters in multispectralmode (false color Image).Ten-meter resolution Is generally adequate for urban spatial planning and macro land managementas It roughly correspondsto most urbanphysical features of concern:street widths and lengths, building and plot size, etc.
0.07 SPOTI, in a continuousregular orbit, providesup-to-date Images of urban centers throughoutthe world.Change detection In the urbanenvironment can be conductedby purchasing satelliteImages at regularIntervals to monitorthe changesof urbanland use, estimatepopulation growth and evaluatethe need on new Infrastructure. 2
0 08 Imagesmedia can be accessedcommercially without the lengthysecurity clearances usuallyrequired for aerialphotography Furthermore, much of the requiredmapping data extraction can be undertakenboth In and outsidethe country concerned
0 09 The time requiredto acquiresatellite Images for any part of the world is about 15days to one month It shouldbe noted that for tropical countriesthe time to acquirean image may be severalmonths, especially If the order Is placedduring the monsoon,or rainy,season where cloud coverageis very dense o 10 One SPOTsatellite image or scenecovers an area of 60 km by 60 kin, an arealarge enoughto encompassmost metropolitansettlements in the world In some cases, SPOTsatellite can also be programmedto get a special Imagefocused on a particularzone of interest (for example,to capture on one image a metropolitanarea splitbetween two or three scenes,or to monitorareas devastated by flood, earthquake,fire, etc) SPOTsystem can also be programmed to produce 3-dimensIonalstereoscopic images useful for topography and for infrastructure planning purposessuch as roads and drainagedesign systems
0 11 For most urbandevelopment and managementpurposes, the satelliteimages provide an excellentsource of detailedland use data for highwayplanning and construction,urban mass transit, urban renewal, neighborhoodfacilities and infrastructureplanning, open space land programs,land assessments,housing stock assessment,environmental impact statements,and growth managementsystems
0 12 The processingof satelliteimages Is the most effectiveway to performsuch tasks at an affordablecost Satelite imagesprovide less expensiveviews of humansettlements than aerial photographyand these materialscan be much morereadily obtained
0 13 Film or transparencyof an image can be enlargedthrough photographicprocesses to scales of 1 100,000down to 1 20,000 One satellitepanchromatic image (on black and white film) and a photographicprint at scale 1 100,000cost only US$1,600,far less expensivethan a set of aerial photographsneeded to cover the same area (Cost refersto early 1989) Photographic prints derivedfrom satelliteimages made from transparenciesor films can be usedto make base maps and to do preliminary urban land use analysis without sophisticated equipment (micro-computer,digitizing tablet, and CADsoftware)
0 14 The purpose of this publicationis to providethe necessaryinformation concerning the use of remotesensing satellite images in urbanplanning and to assesstheir usefu ness PART 1 of this publication gives basic informationregarding remote sensing, listing different civilian satellitesystems, and their particularities PART2 provides Informationconceming the different satellte image media and ways of processingsatellite image data PART3 is a case study that illustratethe use of satelliteimage for urbanplanning purposes The casestudy demonstrateshow satelliteimages, complemented by aerialphotography, can be usedto conducta housingtypology study, evaluatingpopulation densities among differenttypes of residentialdevelopments 3
PART 1 REMOTESENSING AND SATELLITEIMAGERY
1.01 Remotesensing is the measurementof the reflectionof objects or terrainon Earthby recordingdevices or remotesensors fixed on airplanesor satellites.Over the years, a numberof sensingdevices have been used such as photographiccameras, multispectral scanners, thermal scannersand microwaveinstruments. Remote sensing has generateda great deal of information about the Earth -- its surface,environment and resources.
A. From Aerial Photography to Satellite Images 1.02 Remotesensing in urban areasdates back to 1858when Tournachon(later known as "Nadar")used a camera carried aloft in a balloonto study parts of the city of Paris. In his photographsof the village of Petit Bicetre,the houses could be seen clearly. Since then, the morphology or form of urban space has been used directly or as a surrogate for functional identificationand classificationof urban areas. Until World War II, remote sensing techniques remainedbasically the same as the first aerial photographs.Either a singlephotograph or a stereo pair was used for each area of interest. Photographywas essentiallyblack and white and of medium-to-largescale becauseof altitude limitations.
1.03 The second world war brought about a dramaticincrease in the awarenessof the potential of remote sensing in urban areas. In the sixties and early seventies,the conventional approachto remotesensing was dramaticallychanged by the introductionof color, color infrared, multiband,high altitude,and space photography.This was due to the increasingavailability of productsand data broughtabout by new technologiesoften developedfor non-civilianpurposes that were later declassHfiedfor general use. Declassificationof technologyand data provideda major stimulusto the developmentof non-militaryapplications. 1.04 The United StatesNational Aeronautics and SpaceAdministration (NASA) accepted a major responsibilityfor developingcivilian applications of remotesensing. The EarthResources ObservationSystem Program (EROS), initiated in 1965,involved the developmentof aircraft and spacecraftfor remote sensingsystems.
1.05 In 1967, NASAencouraged by the US Departmentof the Interior, initiatedthe Earth ResourceTechnology Satellite (ERTS) program. This program resultedin the deploymentof five satellitescarrying a varietyof remotesensing systems designed primarily to acquireearth resource information.The chronologicallaunch and subsequentretirement history of some of the satellites are shown in Fig. 1.1.The ERTS-1satellite launched on July 23, 1972,was the first experimental system designedto test the feasibilityof collecting earth resourcedata by unmannedsatellites. Priorto the launch of ERTS-Bon January22, 1975,NASA renamedthe ERTSprogram Landsat, distinguishingit from the Seasatoceanographic satellite launched June 26, 1978.At this time ERTS-1was retroactivelynamed Landsat 1 and ERTS-Bbecame Landsat 2 at launch. Landsat3 was launched March 5, 1978; Landsat4 on July 16, 1982,and Landsat5 on March 1, 1984.A variety of mechanicalfailures prompted the retirementof some of the Landsatsatellites.'
'Remote SensingData Acquisition Alternatives. John R. JENSEN- IntroductoryDigital Image Processing - A RemoteSensing Perspective. Published by Prentice-Hall,Englewood Cliffs, New Jersey07632. 4
FIGURE1 1
CHRONOLOGICALLAUNCH AND RETIREMENTHISTORY OF THE LANDSATAND S P 0 T SERIESOF SATELLITESFROM 1972 TO 1989
1981 1982 1993 19Z14 1995 1968 1987 1998 1989 1972 1972 1974 1 1975 1976 1iIT7 1978 1979 1980
LAND17SATI JL9 8
X l Li -. ,_ | J l a LUX5 CSiT\k A21 X Iy I ti SAT 2~ ~ LkDST F4mJ 'aJ I71983
~~ I LA.'~~~~~DSAT7IaDAT M.~~-z~ 1978 P~ur~.d S~ 7 W 1984
a I~~~~~~:O
1. 19849
s~~~or SPT1 5
B. Different Remote Sensing Satellite Systems
1.06 Threesatellite sensor systems providing commercially available digital data are oper- ational. These Include the LANDSAT4, the LANDSAT5, and SPOT I. Dependingon available resources,the application,and the level of informationneeded, users have these systemsfrom whichto choose.Below is a briefdescription of the threesatellite systems and their possibleappli- cations for urban analysis.
1. LANDSAT4 1.07 The satelliteLANDSAT 4 was launchedin July 1982.This system originally had two sensorson board, Multi-SpectralScanner (MSS) and ThematicMapper (TM), but only the MSS sensor is still In service. (See definition of Multi-SpectralScanner and Thematic Mapper in Glossary.)Images from the LANDSAT4 MSSsensor have a ground projectedinstantaneous-field- of-view (IFOV) or ground resolutionequal to 56m x 79m. A satelliteimage or scene covers an areaof 185kmx 185kmwith 10 percentoverlap between west and east scenesat the equator.As latitude increases,so does the amount of overlap becauseof the curvature of the earth. The LANDSATsensor systems have the capability of coveringthe entire globe every 16 days.
1.08 Because of the large geographic area of a LANDSAT4 scene, this product is convenientfor country planning and for global analysis of regionaland country resources,to monitor climate, agricultureand other changesdue to drought, deforestation,etc. However,the low resolutionof LANDSATMSS images does not makethis product particularlyvaluable for urban planning.
2. LANDSAT5
1.09 LANDSAT5 was launchedin March 1984.This systemhas both Multi- Spectral(MSS) andThematic Mapper(TM) sensors operating on board.The MSSimages have a groundprojected instantaneous-field-of-view(IFOV) or ground resolutionequal to 56m x 79m and the TM images have a 30m x 30m ground resolution.Both satelliteimages or scenescover an area of 185kmx 185km.The period betweenrepetitive coverage is 16 days. 1.10 LANDSAT5 data has been generatedfor more than four years. Its resolution has provedadequate for naturalresource management, agriculture and forestryland use classification. LANDSAT5 has also beenused for urbanapplications, but with limited successdue to its relatively poor resolution.The advantage,however, of the LANDSAT4 and 5 imagesare that they can be used together with earlier scenesto providelonger term time serieson land use changes.
3. SPOT I 1.11 The SPOT I sensor system, developed by the French Centre National d'Etudes Spatiales(CNES), was launched in early1986. It has a groundprojected Instantaneous-field-of-view (IFOV),or groundresolution equal to 1Om x 1Om in panchromaticmode (black-and-white)and 20m x 20m In multispectralmode (falsecolor). The period betweenrepetitive coverage is 26 days in "nadir viewing (verticalviewing). Since the SPOTsatellite has a pointableHigh ResolutionVisible Sensor(HRV), it can be programmedto view an areathat is not directly below the spacecraft(i.e., off-nadir).For example,this capabilitywould allowa locationon the equatorto be observedduring sevendifferent passes if on the equatorand on elevenpasses if at a latitude of 45O(Figures 1.2 and 1.3).A given region can be revisitedseveral times each week as opposedto twice a month 6 as with other satellites This capability is usefulfor monitoringrapidly changing processesand events on Earth and for increasingthe possibilityof image acquisiion over areaswith frequent cloud cover 1 12 The SPOTsensors may also acquirestereoscopic pairs of imagesfor a given area (Figure 1 4) Two observationsmay be made on successivedays in that the two images are acquiredat angleson eiher sideof the vertical Thethree-dimensional stereoscopic imagery may be used to producetopographic maps with contour intervalsas low as 10and 20 meters It is also used in digial terrain modeling and the production of three dimensionalperspective views for terrain simulation,strategic planningand impactassessments 1 13 SPOT'sten-meter resolution provides information that was previouslyavailable only wih aerial photographs SPOT data are useful for urban and suburban land use mapping, monitoringof urban growth and change detection, impact of developmentand policies on the naturalenvironment and cultivatedlands, and map updating
1 14 As with all satelliteimages publicly available,SPOT resolution is not to be compared with low or mediumaltitude aerial photography SPOTimages cannot be used for cartographyas a uniquesource of information No cadastralmap can be made with such products until remote sensingsystems reach 1 or 2 meters resolution Satellitesystems are also vulnerableto technical problemsdeveloping on the spacecraft which may not be solved immediatelyAs for SPOT1, only one of the two recorders is now operational,this restricts the number of scenes that can be registered before down loading data to receiving stations on Earth Figure 1 5 is a diagram illustratingthe spatialresolution of the LANDSATMSS, the LANDSATTM and SPOTpanchromatic mode
115 The Soviet Union sells commercialsatellite images with a resolutionequal to five meters Theirphotographic products are recordedon film and reproducedon digital tapes So far, few of these photos have been made public, and it is unclearhow many scenesfrom around the globe will ultimatelybe available 7
FIGURE1.2 OFF-NADIRVIEWING
0Reg-ors Observed
i, J/f; SJ0/ , ~~~~~~~Satellite I CN b/fu / { / / ~~GroundTraclk
1Observable Str.ioW,
(Courtesyof SPOTImage Corporation)
If the HRVinstruments were only capableof viewingthe terraindirectly below the spacecraft(i.e., at nadir), the revisitfrequency for any given regionof the world would be 26 days. However,SPOT is pointable.This meansthat it can be used to view areasthat are not directly below the spacecraft (i.e., off nadir). This is very useful for collecting informationin the even of a natural or human-madedisaster when the satellitetrack is not optimunm,or for collecting stereoscopicimagery. 8
FIGURE 1 3 SPOT OFF-NADIRREVISIT CAPABIUTIES
One Pass Each On Davs D-10 D-5
D-5
CZneObsere
(Courtesyof SPOT Image Corporation)
Duringthe 26-day period separatingtwo succesiveSPOT satelliteoverpasses, a point on the earthcould be observedon sevendifferent passes d it is at the equatorand or 11 occasionsd at a altitudeof 45 degre A given region can be revisitedon dates separatedalternatively by. 1,4 and occasionally5 days 9
FIGURE 1 4
STEREOSCOPICVIEWING CAPABILITIES
pass on cay D pass on day E
/C) X~~~~
rete'ence sunzace
(Courtesyof SPOTImage Corporation) The observationscan be made on successivedays such that the two imagesare acquireda, angies or, either side of the vertical, resulting in stereoscopic imagery Such imagery can be used to produce topographic and planimetricmaps 10
FIGURE1 5 DIAGRAMILLUSTRATING THE SPATIAL RESOLUTION OF THREEDIFFERENT SATELLITE IMAGES
_2 ~ ~
t> - -,;;>^ -56m <-.
Source Jensen, 1986
79 m LANDSAT MSS Ground Resolution = 56 x LANDSAT TM Ground Resolution 330 x 30 m SPOT Panchromatic Band Ground Resolution = 10 xi10m 11
PART2 VARIOUSTYPES OF MEDIAAND WAYSOF PROCESSING AND INTERPRETINGSATELLITE IMAGES
2.01 Sensors fixed on satellites register spectral reflectanceof objects from Earth on digitaltape and send the scannedimage to receivingstations. Receiving stations on Earthcollect imagesas digital data. Radiometricand/or geometriccorrections of the remotelysensed data are needed to improve their interpretability.Subsequently, data may be enhancedalso for visual analysisusing computer applications such as digital image processing.Information extracted from satelliteimages at different processinglevels are madeavailable to governments,public agencies and privateenterprises in need of such information.
2.02 The techniquesto processsatellite images vary accordingto the type of media used (film negativeor transparency,and digital tape).The processingof a negativefilm is similarto the processingof ordinaryphotos. When usingdigital tapes, image processingis done on a computer to enhancean image or to obtain more contrastand increaseddefinition for improveddata inter- pretation.Below, ways of processingSPOT satellite images from panchromaticand multispectral films and digital tapes are discussed,together with some rules of how to visuallyinterpret satellite images.LANDSAT images (MSS or TM) can be processedin the sameways though the resolution is much lower.
A. ProcessingSPOT SatelliteImages from Panchromatic Photographic Prints
2.03 SPOT panchromaticimages have only one data band in the visible part of the spectrumproviding the equivalentof a blackand white imagewhen reproducedas a photographic print. A panchromaticfilm (black and white) at a scale of 1:400,000can be printed at a scale of 1:100,000down to 1:25,000without losing too much detail. (Enlargementof an image resultsin loss of clarity.)See panchromaticprints at variousscale, Figures2.1 to 2.3. A trial was conducted in which a digital panchromaticimage was edge enhancedusing high'passfiltering techniques to increasecontrast. Different urban patterns were usedto determinescale limits in which imagescan be reproduced.Figure 2.4 is one of these urbanpatterns where the imagewas clear enoughdown to scaleof 1:6,000.At this level, panchromaticimages start to lose some contextbut can be used to evaluateurban populationdensities in differenttypes of development.
2.04 In such prints (madefrom panchromaticnegative film), urbanfeatures can be iden- tified by different intensitiesof grey. For example,healthy vegetation produces a dark shade of grey and concretestructures are shown in light tones. Asphaltedroads will havea dark tone while low-densityurban areas will be composedof a mixed patternof light and dark grey where houses are surrounded by trees or some other kind of vegetation.Water (i.e. rivers) will usually be in mediumdark grey tones. However,the relativeturbidity of the water, its depth, the composition and color of the bottom will affect the eventualtone of suchareas. The tone of water in reflecting ponds is often unpredictabledue to these variables. 2.05 In general,the fundamentalmaterials found in urban environmentsmay be distin- guishedfrom one another.Different roofing materials, plaster, concrete, vegetation, asphalt, earth, etc., will produce different tones of grey. In Figure 2.2, the differences between a squatter settlementarea, an industrialarea, a low density and a high density area are readilyidentifiable. 12 FIGURE2 1 'SPOT"SATELLITE IMAGE PANCHROMATICFILM, SCALE 1 400,000 KARACHI, PAKISTAN
S~~~~~~~~~~~~ 4t1tkii ;# t * 13 FIGURE2.2 IDENTIFICATIONOF DIFFERENTPATTERNS OF DEVELOPMENT FROM "SPOT' PANCHROMATICPHOTOGRAPHIC PRINT AT SCALE 1:100,000 KARACHI,PAKISTAN
vs'~~~~~~~~~~~~~~~~~~~~~~~~~/k
L~~~~~~~~~~~~~~~~~~~~~~~~
A KatchNAbadh (SquatterSettlement) B High Density C MediumDensity D Low Density G PublicFacilities, Govermmernt& Defense H DevelopedLand I Agriculture K TreatmentPlantt 14
FIGURE2.3 "SPOT' PANCHROMATICPHOTOGRAPHIC PRINT AT SCALE 1:24,000 KARACHI.PAKISTAN
I~~~~~~~~~~~~~~F 15
FIGURE2.4 "SPOT"PANCHROMATIC PHOTOGRAPHIC PRINT AT SCALE 1:6,000 KARACHI,PAKISTAN 16
B. Processingof SPOT SatelliteImaaes from MultispectralPhotographic Prints
2.06 SPOT multispectralimages have three data bands (green, red and near infrared) which providea uniquerecord of spectralreflectance. The interpretationof a multispectralimage reproducedas a photographicprint is quite differentfrom aerial photographs.The "falsecolor" of these imagesare generallyinterpreted as indicatedbelow.'
2.07 Healthyvegetation is recordedin brighttones and dark red in the multispectralcolor print. Concreteor asphalt roofs in downtown urbanareas will often appear bluish grey. Squatter settlementsare more easyto distinguishin multispectralimage than in panchromaticmode where they may appearin the same grey shade as vegetation.Squatter areas in the multispectralmode have a bluish tone. In dense urban areas, such as slums, the rooftops of the small housesare close to one another giving a more uniform reflectionthan in other residentialareas where plots are largerand roofs are sometimesovershaded by trees. In multispectralmode, baresoil appears as yellowishand grey and non-forestedmountainous sites are easily identifiedby their wrinkled pattern of grey and brown tones. Whereforests or vegetationexist, these appear in red (Figurft 2.5).
2.08 It should be notedthat for tropical countries,low densitytype of urbandevelopment may not be as easyto identify as in dry or semi-drycountries such as Pakistan.The vegetation may cover roads and roofs, and cloud cover may be a problem at times. When using films and photographicprints, imagesfrom tropical countriesshould be enhancedto betterdefine built-up areasfor visual interpretation.Two multispectralphotographic prints at a scale of 1:50,000made from the same image of south Djakarta,Indonesia, are shown in Figures2.6 and 2.7. Figure2.6 is a print made from a non-enhancedimage while in Figure2.7 the image on digital tape was enhancedprior to the reproductionof the film.
' False colors on hard copy imagesdepend on the operator'sdiscretion regarding the assignmentof colors to individualbands. 17
FIGURE2 5 IDENTIFICATIONOF DIFFERENTURBAN FEATURES FROM MULTISPECTRALPRINT AT SCALE1 24,000 KARACHI,PAKISTAN
io ~~~~~~~~~~Q 'A~~~~~~~~A ,s
* t~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-l
'4, ~ hiAba 1 ~~ Sqate Setlmet
A TratmehtAlant
(Squatter~Settlment~ ~ 4 W B HihDesit F Industry 18
FIGURE2 6 PHOTOGRAPHICPRINT MADE FROM"SPOT" MULTISPECTRAL FILM IMAGE BEFOREENHANCEMENT - SCALE 1 50,000 SOUTHJAKARTA, INDONESIA 19
FIGURE 2.7 PHOTOGRAPHIC PRINT MADE FROM "SPOT" MULTISPECTRAL FILM ENHANCED IMAGE - SCALE 1:50,000 SOUTH JAKARTA, INDONESIA
-~~~~~~~~~~4
*~~~~~~ a - e> - -
4 -- <_e:S t- v ~~~4t~ ~ X~ - A t- 4--t
_4; tF - 55i-,$-- i-, ~ ~ t @
r ~ - , 444 4- t
A S t--'!
A tr. a - '- ,,,. 71_
-' 4 C-2- - ?-fe, r- S
~~- -4- -6
g*41 ~~~- ' - \ 5 5.ti >>
4- *,,tT - -* t+ * --! - ,, A ' o, *f- 1* 4
* - .9 A IR4 ''a- i- '- '
_ ~~t -Ak tt' 17 - ' a''_,
-t -~ : -a
_ w
Jr, ., . 3 ~~~~~~~~~~~~~~-- -s-
_wil._ W-t~ -- - 4 4
_l | ' t 5 ~~~~~* A 20
C. ProcessingSatellite Images from DigitalTapes 2.09 When panchromaticand multispectralSPOT digital images are merged using an image processing system, the image can be enhancedto obtain a picture much like aerial photographsat scale 1:25,000.The color of the multispectralimage permitsone to distinguish urbanfrom non-urbanareas, residentialareas of varyingdensity as well as other land use types. From these images,maps can be producedwith still limited accuracy.The accuracyis a function of the quality and the number of ground control points (minimum9 to 12 points)that have been usedto correctthe image.Ground control pointscan be road crossingsor buildings(monuments) which are easyto identify in the image.Using distances between these points, geometricimage correctionsare made. 2.10 Processingsatellite images from digital tapes has proven effective for change detection in rapidlygrowing urban areas. Usingtwo sets of imagestaken of the same sceneat differenttimes, the interpretercan locateareas that havebeen "disturbed" by urbangrowth or other phenomena:flood, earthquake,fire, etc.
2.11 Remotesensing digital data may also be analyzedin correlationwith existingdigital maps and relatedto a geographicalinformation system (GIS) to improveits usefulness.A GIScan be used to relate the remotesensing data with other types of spatiallydistributed data such as soils, topography,existing maps, etc., to facilitatemore complex analysis.
2.12 Use of image processingsoftware to classifyland cover such as forests,and to dif- ferentiateagricultural and soiltypes has provento be a reliabletechnique. However, for classifying urbanfeatures the capacity of digital imageprocessing is limited.The majorproblem is that digital image processinguses the spectral signatureor reflectivevalue of a feature in order to classify it. Unfortunately,numerous man-madematerials have roughly the same spectral reflectance properties.Thus, the spectralsignature of urbanphenomena obtained by remotesensors is often not unique,presenting a high degreeof heterogeneity,and accurateidentification or classification require usingvisual image interpretationin addition to the spectral response.
2.13 Varioustechniques have beenused to measureand incorporateadditional elements of image interpretationinto the image analysisprocess. For example,numerous studies have synthesized"texture" information from the "spectral"data in the imagery. Severalattempts at "contextual"classification make use of neighboringpixel values enabling the classificationof different urbanfeatures.
2.14 Whilea computercannot differentiate between a type of urbandevelopment such as a squatter settlement and the surrounding rocky area, a planner with experience in aerial photographic interpretationcan. The reflectancefrom housingroofs and rocks surroundingthis area may be the same but the patterns of urban form are different. This was the case for the Karachiimage, where the land use classificationcould not be madeby computeralone. For most of the urban area, the reflectancefrom roofs, rocks, and dry land was the same, thus the land classificationwas done visually based upon a photographicprint of the image.
D. Cost ComparisonBetween Different Satellite Image Processing Technigues
2.1 Essentially,there are three generallevels of processingremote sensingdata. These are summarizedin Figures2.8 and 2.9, and described below: 21
1. High Levelof Data Processing 2.16 A high levelof data processinginvolves the use of digital tapes and image processing software.This computer assisted method for enhancingsatellite images allows one to better interpreturban features such as land use classifications.Most image processingsystems offer the possibilityto draw vectors and polygonsover the image display on screen.This feature can be used to trace road networksand to produce thematic maps such as land use maps. Another useful application in image processingis merging existing computerizeddigital maps with an image.This techniqueis used in cartographyto create or update maps and for changedetection when using time series data.
2.17 Processingimages at this level requires a good knowledge of image processing techniquesand requiresinvestment in computerhardware and software and training.The average cost of an image processingsystem running on microcomputeris in the range of US$40,000to US$60,000.Companies specialized in the processingof remote sensing data do offer image enhancementand mappingservices. Depending on the application,the cost per image or scene may vary betweenUS$10,000 to US$24,000.
2.18 High level data processingtechniques can be incorporatedwith medium level techniquesas well, particularlyfor adding additionaldata for analysis.
2. Medium Levelof Data Processing
2.19 A mediumlevel of data processinginvolves the use of negativefilms or transparen- cies that are reproducedas photographicprints at variousscales. Satellite image data are visually interpretedfrom the panchromaticor color prints. Land use and road maps are digitizeddirectly from these prints at their scale. Using secondary sources other infrastructurenetworks (water supply, sewerage,drainage, etc.) can be superimposedon the road base map. This can form the basis for an urban informationsystem. 2.20 Mediumlevel data processingrequires some knowledgeof CAD (ComputerAided Design) and GIS (GeographicInformation System) to digitize maps and to develop database programsfor exportingdata from maps into informationsystems for use in urbaninfrastructure and managementmodels. This levelof processingrequires a microcomputerand peripherals(digitizer, plotter and printer) costing approximatelyUS$10,000 to US$25,000and the use of softwaresuch as CADor GIS costing betweenUS$3,000 and US$12,000.
3. Low Level of Data Processing 2.21 A low level of data processinginvolves using photographic prints madefrom films or transparencies.Satellite image data are visuallyinterpreted and maps are manuallytraced from these prints.
2.22 This level of processingdoes not requirespecialized computer equipmentnor skills since all spatialinformation would be producedmanually, including the measurementof areas by planimeter.However, inexpensive micro-computers (costing betweenUS$2,000 and US$3,000) can be used to processthe data derivedfrom the interpretationof the photographicprint. 22
FIGURE2.8 REMOTESENSING SATELLITE IMAGE DATA PROCESSING AND THEIR APPLICATIONS IN REGIONALAND URBAN PLANNING
HIGH LEVEL
LANDWUSE CLSIICTO ROADNEVNORK - ~~~~~~~~~~~~~~CHAkNGE DETECnON ReceivingStation IMAGESAREA PROCESSED FROM DIGITAL TAPES
MEDIUM LEVEL
IL n BffE U~~~MPcmDIhZ FOlt - RE1ON.L PAWNNIN Ut PLANNING =lP z l t i -~~~~~L4lD DEVEUOIEWT *i2. - LAND MAKAODAD4T - UPCRADIN DIGMZERTABLET AND CADOR GIS SYSTEM
LOW LEVEL D USEE - -FRRMUCTUREASEMN - HOUSINGSTOCK ASMENT -POPULATlON E5TMATES LAND MACXEr EVAWATION INFRASICTURE- PLANNNG
- tANSPORT OONOMICS AND PLANNING - URBM FCAL MANAGEMOCT - PROGRSAMEUA8NAflO &
DATAPROCESSING -URBAN INFORMATIONSYSTEM 23 FIGURE2.9 "SPOT" SATELLITEIMAGE PRODUCTS - PROCESSINGCOST AND RESULTINGOUTPUT
"SPOTS SATELLITE IMAGE PRODUCTS AND PROCESSINGCOST (1989 Cost in US Dollars)
PROCESSING SATELLITE PRODUCT TYPE OF PROCESS- RESULTING O.JTPUT LEVELS PRODUCTS COST/UNIT PROCESSING ING COST
HIGH LEVEL ------> HIGH LEVEL: Using digital tapes, MULTI-SPECTRAL $1,700 Digital tapes, multi- This type - Enhanced image satellite images DIGITAL TAPE spectral and panchroma- of pro- are processed and enhanced (Color image) tic are merged to im- cessing - Land use classification with Image Processing software 20m. resolution prove resolution. Images cost are enhanced in order to approxi- - Road network map Comouter system PANCHROMATIC S 2,200 enlight urban feature. matly & Image processing software DIGITAL TAPE - -Photographic printout of S12,000 - Change detection cost: S60,000 (8&W image) the enhanced images 10m. resolution are comparable to high - Three dimentional image altitude aerial photos. for terrain modeling
MEDIUM LEVEL ------> MEDIUMLEVEL: Using prints made PANCHROMATIC S1,500 Photographic Print - Land use and road from panchromatic FILM (B&W) enlargement: (B&W) computerized maps and multi-spectral Negative transpa- scale: 1:100,000 S65 films, maps are rency made from scale: 1:50,000 $100 - Computerized overlay digitized with digital tape scale: 1:24,000 $350 maps for water, sewerage GIS or CAD 10m. resolution drainage & electricity. software scale= 1:400,000 - Urban Management System Computer system MULTI-SPECTRAL S1,300 Photographic Print using GIS System. & software FILM (Color) enlargement: (Color) cost: $10,000 Negative transpa- scale: 1:100,000 S160 rency made from scale: 1:50,O00 $275 digital tape - - 20m. resolution scaLe= 1:400,000
LOW LEVEL L------O>LOW LEVEL: Using B&W and/or PANCHROMATIC S1,500 Photographic Print - Land use, road and color prints, FILM (B&W) enlargement: (B&W) utility maps drawn maps are manually drawn Negative transpa- scale: 1:100,000 S65 manually. rency made from scale: 1:50,000 S100 digital tape scale: 1:24,000 S350 - Urban Management System Computer system 10m. resolution using LOTUS 1-2-3 and & software scale= 1:400,000 database management cost: S3,500 systeme .~~~~~~~~~~~~~~~~~~~ytm 24
E. Cost DifferenceBetween SPOT SatelliteImages and Aerial Photos 2.23 UsingSPOT satellite images for mappingat a scaleof 1:50,000is less expensivethan conventionalmapping with aerial photos.Although it is too earlyto haveprecise figures from large surveys, IGN France(Institut Geographique National) reported a ratio of 3:1 in favor of SPOTdata. The cost comparison presentedin Figure2.10 suggeststhis ratio is generallycorrect.
FIGURE2.10 COST COMPARISONFOR MAPPINGWITH AERIAL PHOTOS AND SPOT SATELLITEIMAGES
Photos or Images Total Acquisition Cost * Price per Km2 Staff US DoLLars
Conventional 170 USS Surveying& 142,500 mapping using Mapping aeriaL photos 24 Person/Months
Mapping with 45 US$ Surveying& 80,000 SPOT satel!ite Mapping images 3 Person/Months
*Cost for mappinga coastalzone of 500 km2 in Polynesia,Loyalty Islands at a scale of 1:50,000.
Studyconducted by the FrenchHydrographic and OceangraphicSurvey (Service Hydrographique et Oceanographiquede la Marine),FOURGASSIE and LE GOUIC,1987. NOTE: Cost differencesbetween aerial photosand satelliteimages are due not only to product cost but diverse processingtechniques and significantlydifferent timetables.
2.24 From this finding it can be concluded that SPOT satellite images, particularly panchromaticimages with ten-meterresolution, offer an alternativeto medium-scaleaerial photos for urban planning purposes. It should be noted again that for tropical countries, image interpretabilitymay be limitedwhen vegetation obscures housing development. Another area where photo interpretationmay not be possibleis in desertic regionswhere houses are made of mud blocksand villagesare dispersedwithout any visibleroads to link settlements.Scale limits are also a constraint and Konecky in a 1987 UN publication "World Cartography' pointed out that 'interpretability' is a factor which limits satelliteimages scale to 1:50,000for cartographicuses. Such scale limitsdo not permitthe use of satelliteremote sensing products for cadastralpurposes in urban areas. 25
PART 3
USINGSATELLITE IMAGES FOR URBAN PLANNING
KARACHI CASE STUDY
To illustratethe usefulnessof remotesensing in urbanplanning and urban management,a case study Is presentedIn this section startingwith a brief overviewof the City of Karachi,Pakistan. 26 FIGURE3.1 KARACHIOLD CITY MAP - 1885
f'23t As-f 3,'2B]3 -S,Z C j - --'. ;.. i; ,_';'G'j5JA.
1, -1, I T -9 D. -? R~-~:~
- 7,
s , ;,*S>~~ o < 11 < ! < ; , X __. -.C4J
_ <~~~~~~~~~~~~~~~~~~~~~4
I N 1) E~~~~~~~~~~~~~~~-
twsa s5 'Pu2 I!-: ', ' :- P2I--
i' I _ . ,
I jJ~~ I'IN1}t \X ZX OLIYZ |, TAMWAf YS CO//rS7,90CTtO J ._X>
KEY MAR. *...... :Af;t#. ssoA'sAPtfL/ fOf _ __. …_ Phoco: Library oJfCongress, Wahington, D.C. 27
A. Karachi Urban Development from 1943 to 1988
3.01 Karachihas a long recordedhistory of settlement,but its rapidgrowth has taken place in the past severaldecades.
3.02 By 1843,Karachi was still a walledtownship of 14,000people on a site of 35 acres that had been developedas a port by the British for exportingthe products of the Punjab.See Figure3.1, an old map of Karachidated 1885.As recentlyas 1941,the population of Karachi Districtwas less than half a million.A metropolisburst into being with the birth of the new nation.
3.03 Between1947 and 1951,approximately 600,000 displaced persons who migratedinto Karachifrom India,as a result of the creationof Pakistanin 1947,found shelterin the city, swelling the populationto 1,130,000.
3.04 Karachi'spopulation increasedat a rate of 4.8 percent per year between1971 and 1981.If this trend continues,metropolitan Karachi could have closeto 13 millioninhabitants by the end of the century,or more than twice the estimatedpopulation of six millionin 1987.
3.05 The DevelopmentPlan Report of 1974 defined metropolitanKarachi as an area coveringsome 349 squarekilometers. Today, it is ten times larger(3520 km2) and stretchesfrom the DhabejiWater Works in the east to Baldiaand the BanderQasim port in the west when the urbanarea calculatedfrom satelliteimage dated from January1987 is 1,012km2 (Figure3.2). 3.06 Despiteofficial attemptsto contain Karachi's growth and dispersedevelopment to secondarycities, economic activities in the metropolishave been expandingsteadily. Incentives and tax reliefdesigned to stimulategrowth, in fact, promptedindustries to set up operationsat the edge of the Karachiadministrative district in areas such as Hub Chowki in Baluchistan,just 24 kilometersfrom Karachi'scenter, and Dader in Sind,80 kilometersalong the Karachi-Hyderabad Highway.
3.07 Closeto 30 percent of the populationof Sind province lives in Karachimetropolitan areaand 33,000households are added to the area annually.
3.08 Although close to 45 percent of all employmentis concentratedin four areas -- the Sindand Landhiindustrial trading estates,the port, and the city center -- the adjoiningareas are not highly populated.In fact, morethan one quarter of the built-uparea has been pre-emptedfor defensepurposes and developed under cantonnementboards at very low densities.About five square kilometers of fully or partly serviced land lie unused in the heart of the city, and infrastructurehas been extended into outlying areas at great expense in responseto private speculation.Labor outflowsto the Middle East and remittancesfrom the 1.25 million emigrant workersthere (saidto averageRs3O,000 a year per worker) havebeen largelyresponsible for the speculation on land and housing. Although approximately 80 percent of the land in the metropolitanarea is publicly owned,and a planningstudy was undertakenbetween 1968 and 1973 to deal with land use, Karachihas been unableto controlthe developmentof its urban land. 28 FIGURE3.2 KARACHIREGION
t~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ /44~~~~~~~~~~~H
I A
S I N
DA _/oga *rumian ' "^' < + t <3Hab H b / Sari Sing 5 I I u a LASSEL Dom
> CIrAo t Jh pi o a ' Moz a % Khade ai GodanCii
ARABIAN Bundac3 SEA| a r XMirpur Sakr l
0 ~~~25KM. 29
3.09 A large part of Karachi'shousing is now situated on the periphery of the city, in sprawlinginformal settlements that exhibitlittle planningand nominalinfrastructure. To add to its problems,Karachi has by far the largestsquatter settlements in the province.Referred to as Katchi Abadis, these house about one third of Karachi'spopulation in generallyunsanitary conditions (Figure3.3). 3.10 Recently,national policy has becomecognizant of the needs of low- income groups. For example,land leases have been grantedto squatters,and in situ ameliorationof squatter settlements(Kachi Abadis) is a current priority.Upgrading of servicesin squattersettlements has resultedin visible self- improvementin the quality of shelter.
3.11 In the past ten years, Karachihas endeavoredto reorganizeits institutionsin order to improve urban management.In some fields, such as water supply and sewerage,sectoral objectiveshave been accepted and the institutionalmechanisms needed to achievethem have been established.
3.12 The Governmentof Pakistanasked UNCHS (Habitat, Nairobi) to assist KDA(Karachi DevelopmentAuthority) on "Strengtheningof the KarachiMaster PlanningProcess (1986-2000)."
3.13 The project's developmentobjective is to contributeto the Government'sefforts to improvethe overallconditions of humansettlements in Karachiand further morein the country by: o Improvingthe efficiencyof the deliveryof urban servicesand shelterto inhabitantsof Karachi; o Developingnew approachesand methodologiesthat could be replicatedin other parts of the country; * Developingand implementingwithin KDA a new urban growth managementtool; o Creating and establishingmechanisms for the continuousupdating of DevelopmentPlans preparedby KDA,and o Trainingthe operativestaff and senior officialsof KDAin the use of these methodologiesand approaches.
3.14 The UNCHSrecommended the useof satelliteimages to evaluatethe actualextension of Karachiurban area and its land use.SPOT satellite image products (films and digital tapes, both panchromaticand multispectral)were acquiredto conduct this study. Filmswere reproducedat differentscales from 1:400,000,1:100,000, and 1:24,000and digitizedfor the making of land use and road network maps. Imageson digital tapes were enhancedto analyzeurban development patternssuch as populationdensity.
3.15 The following casestudy illustrateshow data derivedfrom remotesensing and other sources were utilized to develop computerized maps and the beginning of an urban or geographicalinformation system. 30
FIGURE3.3 KATCHI ABADI OR SQUATTERSETTLEMENT - KARACHI
w_- ;;47
- ,g,/'/, 'S'' ' '; l.E ' . ,~~~~~~~~~~~~~~~~4'
._,.~~ _ ,,:.,v.- U,,.._, ...... -...... 4is -v 4i0/0r<0t* 31
B. ComputerizedMapping and GIS System
3.16 Thematicland use mappingfrom photographicprints of a satelliteimage can be done manuallyusing mylar or tracing paper.A base map is made over the original print including all visible roadnetworks. A secondland use map mayalso be derivedfrom the same print aftervisual interpretationand field checksHf possible. Using available secondary data, various overlay maps for water supply, sewerage,electricity networksand density maps, etc., may also be prepared.
3.17 In a traditionalmapping process, modifications usually require the preparationof new maps. Changesor additionsare not made without redrawingthe entire maps. 3.18 Usingmicro-computer applications such as a GeographicInformation System (GIS) computerizedbase maps can be created, modified and completed by overlays of information. Facilitiesmanagement can be introducedby transformingall geographicaldata into digital or numericaldata. In other words, a map or a chart can be altered, expanded,sorted and output printed accordingto the particularspecifications of a project.
3.19 GeographicInformation Systems (GIS) is a term whichfirst appearedin generaluse in the late 1960s.It is normallyused to describecomputer facilities, which are usedto handledata referencedto the spatial domain, with the capability to inter-relatedatasets to assist in their analysisand the presentationof the results.This system allowsthe user (a planner,an engineer, a site assessor, a data processor, etc.) to efficiently manage and manipulate data about geographiclocations. An importantconcept to understandis that geographicdata include both spatialand descriptivedata. Spatial(or geographic)data are simple data that have a geographic location such as a zone, an institution,a building, etc. Descriptive-data, or attribute data, are informationassociated with a particularphysical object. Thesedata describe an entitythat has a uniquegeographic location, for examplean industrialestate covering 12 hectareswith 40 percent of land area allocated to circulationand open space.These items are descriptiveelements or attributesof this zone. 3.20 A GIS offers substantiallymore featuresand flexibilitythan traditional mappingand data managementtechniques. A key featureis the capabilityto store and managegraphical and alpha-numericaldata simultaneously within the samesystem, allowing cross-referencing, merging and sortingof all relevantinformation stored in the databaseof the GIS system (Figure3.4). 3.21 For this casestudy all maps presentedbelow were computerized.The land use map and road network were digitized from the photographic print at a scale of 1:100,000using AutoCAD,a ComputerAided Design(CAD) software package running on micro-computersystem. Auxiliarydata suchas water supply,sewerage, and drainagenetworks were also preparedas base map overlays.Computerized maps and associateddata may form the basis for an embryonic urban informationsystem or GIS system. 32
FIGURE3.4 GEOGRAPHICINFORMATION SYSTEM
Contours TABLE: WATER
'Roadways / < / /7 Street: Malik Rd Roadways Pipedia: 200 Rd_Surf: Asphalt Dip: .75 Land Use Const: 1956 Break: 0 TABLE: PARCEL Infrastructure Street: Malik Rd Network Street_No: 155 Word: 12 Parcel_No: 125 Cadastre Type_PI: RESID =BldgFloor: 1 SPATIAL ALPHA-NUMERIC OR INFORMATION TEXTURAL INFORMATION 33
C. MappingLand Use and Road Networksfrom SatelliteImages: Starting an Urban Database
3.22 The steps followed in preparingthe Karachicase study are described below and graphicallyrepresented in Figure3.5.
Use of SatelliteImages for MappingPurposes
3.23 At the presenttime, SPOTsatellite panchromatic and multispectralimages are most appropriate for urban applications. Transparenciesor film negatives can be enlarged on photographicpaper at a desirablescale and usedto trace roads and land use base maps. Use of digital tapes requiresaccess to an image processingsystem. Images on tapes can be enhanced for better interpretationand/or classificationof areas of interestsuch as squatter and dense settlementsand other criticalareas of concern.Digital images can also be reproducedon films or printed using an inkjet or thermal printer. ConvenientScales of PhotographicPrints made from SatelliteImages
3.24 Imagesfrom SPOTsatellite negatives may be printed at differentscales. For a large metropolitanarea, broad analysis using a panchromaticphoto at a scale of 1:100,000may be sufficient.For generalplanning purposes panchromatic photos at a scaleof 1:25,000are suitable to identify most urban features.
3.25 NOTE:Multispectral films at a scale of 1:50,000- 1:100,000are also very usefulfor metropolitananalysis. Visual Interpretationof Urban Features(roads and land use)
3.26 "SPOT"photographic prints at a scale of 1:100,000scale permit major road networks and various land usesto be identifiedand a thematicland use map to be prepared.
3.27 The major road network on panchromaticprints appears in dark grey and are continuouslines. Industrial areas can be recognized by their large roofs which have a higher reflectancethan any other type of construction.These appear on black and white print as large white squareswith a more important road networkthan in residentialareas (Fig. 3.6).
3.28 Through visual photo interpretation,dense and loose grid patternswere identified. Thesewere used to determineurban densityand to classifyresidential areas. Other urban areas where no street patternscould be identifiedwere classifiedas squattersettlements locally known as "Katchi Abadies." Various residentialurban fabrics, or housing types, are schematically presentedin Figure 3.7.
3.29 High densityhousing types (A and B) includesquatter settlements located next to the industrialarea (S.l.T.E.)on the east side of Karachi,along the Malir Riverand in the center of the old city where the road network is more visible than in slum areas.The old center of Karachi appears in a homogenousdark grey tone (Figure2.2 and 2.6). 3.30 Mediumdensity types of development(type C) were partly identified by the road network,which was more visible than in high density areas. A mixture of light and dark grey shadesindicates that the land is occupied by small buildingswith open space and vegetationin between. 34
3.31 Lowdensity types of development(type D) are locatedin northernand southernparts of Karachiand in its peripheralareas. Roofs in this type are noted as a smallwhite dot surrounded by a mixedtone of dark and light grey background(see Figure3.6).
3.32 From a multispectralphotographic print (false color photo) at scale 1:100,000,the urbanareas can be clearlydefined. They appear in a grey-bluishshade while non-developedland appearsin the grey-yellowishtones. Squattersettlements are definitelymore easyto locate than in the panchromaticimage where their grey shade may be interpretedas vegetation.In the multi- spectralmode, slumsappear in a bluish shadewhile vegetation stands out in red or brown-redin the color print. For visualinterpretation and generalland use classification(residential, industrial, transportnetwork, agriculture), multispectral prints are certainlyof greateruse than a panchromatic image printed at the same scale; however,both serve usefulpurposes. 35
FIGURE3.5 USE OF SATELLITEIMAGES IN URBANPLANNING AND URBANMANAGEMENT
MAPSAND DATA MAPSAND DATA DERIVEDFROM DERIVEDFROM SATELLITEIMAGE SATELLITEIMAGE MADEAT SCALE 1:100,000 MADEAT SCALE 1:25,000
MAJORROAD LAND USE ROAD NETWORK LAND USE MAP NETWORK CLASSIFICATION PRIMARY,SECOND. AND HOUSING PER TYPE & TERTIARYROADS TYPOLOGY
FWATER,SEWER FHOUSING -WATER,SEWE7R IOPULATiON ELECTRICtTY TYPOLOGY ELECTRICITY j ESTIMATE I MAIN NETWORKSI I POPULATION I I NETWORKS i I EST IMATE ,
NOTE: Items enclosedin broken line are not directly obtainedfrom satelliteimages. these data are derivedfrom existing engineeringmaps and field surveys. 36
FIGURE3.6 IDENTIFICATIONOF DIFFERENTURBAN FEATURES FROM "SPOT' PANCHROMATICPHOTOGRAPHIC PRINT - SCALE 1:100,000 KARACHIPAKISTAN 37
FIGURE 3.7 DIFFERENT TYPES OF RESIDENTIAL DEVELOPMENT KARACHI, PAKISTAN
TYPE A (Katchi Abadi) TYPE 2 (City Center) SQUATTERSETTLEMENT HIGH DENSIlY Average plot size (m2) 75 Average plot size (m2) 90 No. of persons/plot 6.5 No. of persons/plot 8.5 7. circul. & open space 25 % circul. & open space 25 Density persons/ha 650 Density persons/ha 625
TYPE C (North-East Karachi) TYPE 0 (North Karachi) MEDIUM DENSITY LOW DENSITY Average plot size (m2) 150 Average plot size (m2) 350 No. of persons/plot 6.5 No. of persons/plot 6.5 X circul. & open space 40 % circul. & open space 45 Density persons/ha 260 Density persons/ha 102 38
Maps Derivedfrom SatelliteImaaes 3.33 The land use and road networkmaps for Karachiwere deriveddirectly from satellite images.The thematicland use map and road networkpresented in Figure3.8 was digitizedfrom a panchromaticphotographic print at a scaleof 1:100,000.The road networkmap, shownin Figure 3.9 was digitizedfrom a panchromaticphotographic prints made at a scale of 1:24,000. 3.34 At a scale of 1:24,000,depending on the quality of the image,different residential types of developmentfrom high densityto mediumand low densitysettlements can be identified. Theirareas can be measuredand after field surveystheir densities can be evaluatedfor population estimationby sector or for the entire city. To conduct a housing typology used to evaluatethe populationof Karachifor the year 1988,both panchromaticand multispectralimages at a scale of 1:24,000were usedto identifydifferent urban fabrics before selection of the housingtypes. The Karachihousing typology is to be presentedin Section D. Other Map Overlays 3.35 The roadnetwork map derived from the "SPOT"photo can be usedas computerbase map to overlay other network systems (water supply, sewerage,drainage, etc.) by digitizing existing network maps and matchingthese to the base map createdfrom satelliteimages. (See Fig. 3.10,3.11, 3.12.) 3.36 The land use map, developed at a scale of 1:25,000,can also be used as a computerizedbase map for other land informationsuch as zoning map, land value, etc. Urban MapDingand Urban Data Integration 3.37 Datadirectly derivedfrom satelliteimages: land types and their correspondingareas, roadtypes and lengths,plus the auxiliarydata concerningwater supply, sewerage,electricity and other networkscan be listed on maps as shown in Figure3.13. These data can be extractedfrom the computerizedmap done in AutoCADand imported to databaseand/or to spreadsheetfor further analysis(Figure 3.14). This is the principleof a (GIS)or GeographicalInformation System.
3.38 Dataextraction from map to databaseis possibleonly if the data listed on the map has beenwritten as attributesand linked to a point, a line,or a specific area.As shown in Figure 3.13, the notation or attribute "INF - 2375.22"corresponds to informal or squatter settlement developmentof 2375.22hectares and is registeredin the drawingfile with the same "tag"name as in the databasefield name.
3.39 GIS Systemswhich were first developedfor miniand mainframecomputers are now being used for urban mappingand data managementpurposes. ARC- Info is an exampleof such software.Micro-computer GIS softwaresuch as MunMAPmay be used to extract and/or insert attributesto a computerizedmap drawn in AutoCADto databasemanagement systems such as RBASE5000. MunMAPcan be used as a GIS systemfor data querieswithin a map with enough flexibilityfor the user to developa proprietarysystem adaptedto countriesand cities that have different needs.The other GIS softwarefor micro-computersystems is pcARC/INFO,which is describedin AnnexI, Section D "Softwareneeded for Mappingand Data ManagementSystems." FIGURE3.8 THEMATICLAND USE MAP WITH MAJOR ROAD NETWORK KARACHI,PAKISTAN KARACHI AREA X AREA X AREA X LAND USE (HLA) (HA) (HA) KATCHIABADI 5722 5.65 IIIIIIII INDUSTRIES 5370 5.30 AGRICULTURE4619 4.56
HIGH DENSITY 2740 2.71 DEVELOPEDLAND 9965 9.84 OTHER 50515 49.89
MEDIUMDENSITY 8607 8.50 SOVERNMCT/LNT.1009 1.00 Map derived -from SPOTr TOTAL satellite image of Jan. 1987 LOWDENSITY 3584 3.54 OEFENSE 9125 9.01 URBANAREA 101256100.00
D ~ ~ SN North :. Northe t / Karoqi I Karai/ H A B V A L L E Y (Schem (Schee33)
WESTKARACHI MAURANG mr
WSEAS
Sondapit~~ ~~ ~ ~ ~ ~ ~ ~ ~~~~~~~NtoalHgw
Korongi Creek SCALE Monara N 1OKm A R A B I A N S E A I
I 40
FIGURE 3.9 KARACHI ROAD NETWORK
KARACHI ROAD NETWORK Highway Primary Road Secondary Road I.~~~~~~~~~~~~~~ t ) \ S \ 0 $ ~~~~~~~~~~~~~SUP£IH
at Scaie 1:25,000. 41
FIGURE 3.10 KARACHI WATER SUPPLY NETWORK
KARACHI WATER SUPPLY NETWORK
Existing Water Main Proposed Water Main/Zone 1 Filter Plant
All Zones Zone 0' - 50' * Reservoir
-- _ Zone 50' - 120' A Pumping Station
Conduit - Zone 120'- 200' e Pressure Valves
BALUC ISTAN SIND
i \ ~~~~~SUPERHIGHWA'
Aurnangro Swam
say Creek MaKorang
SCALE 10 K)"l _ ARABIA SEA N
Source: Karachi Development Plan 1976 - 1985 Karachi Development Authority - Karachi, Pakistan 42
FIGURE 3.11 KARACHISEWERAGE NETWORK
KARACHI SEWERAGENETWORK
Existing Trunk Sewer Proposed Trunk Sewer - - - Existing Rising Main 3 Treatment Plant - - - Proposed Rising MaTn ^ Pumping Chambers
BALUC HISTAN S JHW SIND
AurngiKorgi Creek
SCALE0 KM1 ~~~~ARAB!ASEAN
Source:Karachi Development Plan 1976-1985 Karachi Development Authority - Karachi, Pakistan 43
FIGURE 3.12 KARACHI DRAINAGE NETWORK
KARACHI SURFACE DRAINAGE
Trunk Drain // Sea Zone Nalas-Natural Drain r Catchment Area Foot Hill Zone L.J Boundaries
BALUC iSTAN SiND
d \ i SU~~~~~~~~~~PER HIGH'NA) * C~Manshapir
Bay ~~~~~~~~~~~~~~~~~MangroveSwamp
> Korangi Creek
SCALE,_ 0 KM ARABIAIARABIASEA E
Source: Karachi DevelopmentPlan 1976 - 1985 Karachi DevelopmentAuthority - Karachi,Pakistan 44
FIGURE3.13 LANDUSE TYPE AND AREA ATFr iUTES AREUSTED ON MAP
RECORD# 00001 URGIURAN -CI . ID# :1 LrYPE :INF LAREA : 2375.32: skiand /
RECORD# 00002
LAREAI / 427 55: X*
DEY :._4ALD1IA ) ,D
Ail data attributeslisted on mapsare directiyimported to databasemanagiement systems such as dBASEIII,R:BASE andi LOTUS1-2-3 spreadsheet. Mapderived from "SPOT"satellite image dated January 1987. I 45
FIGURE3.14(A)
ATTRIBUTESOR DATA LISTED ON "AUTOCAD"MAP ARE DIRECTLY IMPORTEDTO DATABASEAND LOTUS SPREADSHEET
AUTOCADDATA
1 I NIF 2775. 7 ) L r" zt 2 I NF 4,2 7.5 5 :4 LCAJ :7.:: ,,6 f3 Si. E WI- _ : _ L_27.c 63 5 63i
I,w. 7 O T.^ E F* , 4 :, .L_ v , , .4 1 77 I'I .Fa -- t F-rr)ftlt_w_I_jSc, ..
/, T 127 , _,,~ _~ f _: L2EE -s. 7 6 ., I 4FF A-,42 4 'E ,-Vf 5, -75 - ,N,-. v4n 5Cf, Lf,_, t- sX^ A 1 7 fl 1 ,-E7 -_-. =f
7 iN= 612 77I4 , 77, LE.;= 6 :Q, I I.r4 -I,. 64s s; I ;'l --il, '. 7. .. -- ;EW
1tL16 ;3IINf N-@F x{F 125.0927ff,A'.7C r 4:I INDIN:r. ^) ,f'-;- r7.) _' . - ~-Lx: :t:,.-- - 5- 17 IFx I:.5 L- * _ r^ 2:.i' ,t f2. f .-
l. .ih-,IF 256. 05 'f^. . 7 "-- i ->St;-tOm-t)s1INF -5, " - '7* + I-^^4F 2-- .rl_, _;:, ' th'-. _.dlLw _* . *--- r -- , s* , r- 171:.f_- ,, -, T ' 26{.4 v17,_J c - ,-,f_ ' .
is I- N les- N, L)3$t __ , IN og E S4f f _-- L 2 6 1 'C,. 7 f-f-- 1'[i '13 27.571I 6. NF_IE,,_ 7' _ ir Nfo_ 51: * i T F _4 N2. 4 E- ''^ 0 xz-D r- _~~~~1c. :o zw i -. L s.r _ 92 ,,. G~ 46_525-,fL,_l7. , _ ,tE 1i ~::T,-3 4.1 7=.,Ci s 3-'ItW} rE-f18c ~ * H IED 12. - . - '- 4-.716
:1 MED _ l _52,X. 8 r-c. z rfG '7 EEI '.C6 35
I - -Polygon Area of ID Number. different land use type Abbreviation for different land use type
The above data filed in AUTOCAD,(polygon ID number, land use types and areas), are imported to dBASEIII,RBASE and LOTUSspreadsheets and can be usedtogether with physicaland socioeconomicdata gatheredfrom field surveysas seen Figure3 14(C) 46
FIGURE 3.14(B)
ATTRIBUTESOR DATAON "AUTOCAD"MAP ARE DIRECTLY IMPORTED TO DATABASE AND LOTUS SPREADSHEET
DATAEXTRACTED FROM AUTOCAD ARE IMPORTEDTO dBASEIII
=, ~=Z _ ~ ~ ='~~ -= ==_== =. =. = :. = :_ = _: _,. =1 =- _: . _ .. _ AREA (Ha) %
KATCHHI ABADI 1(hJSqiuattter) 5722 5. 6,5 HIGH DENSITrY 2 740 2. 71. t1EI:DIUM1 DENSITY 8607 .-, 50 LOW DENSITY 3584 3- 54 INDUSTRY 5370 5 . 30 DEVELOPED LAND 91.:65 I.) 4 SERVJICEf/TJIL.//G0VMT/INST- 1009 1 - 00 DEF ENSE 9Y125 9. 01.C AGR'rJIULTUR:E 4619 4S56 Ti-HER 5051. t5 4 9,9
TOTAL 10(1256 100.0(3
Dataextracted from AutoCADare importedto databasemanagement system. A dBASEIIIprogram classified the land use types and computedthe areas by categoriesas shown Figure3.14(B). 47
FIGURE 3.14(C)
ATTRIBUTES OR DATA ON 'AUTOCAD" MAP ARE DIRECTLY IMPORTED TO DATABASE AND LOTUS SPREADSHEET
DATA EXTRACTED FROM AUTOCAD ARE IMPORTED TO LOTUS 1-2-3
A B C 1 ID# L.TYPE LAREA 2 1 INIF 2375.32 .3 2 INiF 42755 4 3 INF 4z4 4.4) .S, 4 INF l.1 4.: [<: 5 INF 473 .6,2 7 ,6 INF 56 1. 7 INF 1.2. 77; .9 ,.-3 IINF 74.5 10 9 INF 256f.. 05 1.1 10 I NF .147.75 12 11 INF 130 -l1.4 :9 - 12 ? IN4F 1 4z4 .n 1.4 INF D509 5/ 15 14 INF 59: 1 16 15 INF f,,2 t,4- 17 f16 :hlF 27-57 1:8 -, INfF 136.f, .SS:
10 17, I NF 15 - -9 0 20 19 INF 16A-3';9 21 20 HIG 5322. 0=-4 22 21 HIG 17 2. - 23 222 HIG 2244..1 24 23 MIED 142 -.c' 25 S24 MED 329.9 26 25 MED 1713.05 27 26 MED1.004-1_
A B C D E F H I TYPE AREAHA AREAICIR OPENPLOTSIZE PERSONPL POPULAT DENSITY POPJ TINF 5722 5.65 25 75 6.5 3719300 650 45.61 THIG 2740 2.71 25 90 7.5 1712500 625 21.00 TMED 8607 8.50 40 1SO 6.5 2237820 260 27.44 TLOW 3584 3.54 45 350 6.5 366080 102 4.49 TIND 5370 5.30 100 1 6.5 0 0 0.00 TDEV 9965 9.84 45 1 0 0 0 0.00 TGOV 1009 1.00 100 1 0 0 0 0.00 TDEF 9125 9.01 90 500 6.5 118625 13 1.45 TAGR 4619 4.56 100 1 0 0 0 0.00 TOTH 50515 49.89 100 1 0 0 0 0.00 -______--__----_____--___--___--__----_--___-- ______--____--___ TOTAL 101256100.00 8154325 81 100.00 ------…----- __ ------____
Data extracted from AutoCAD, (land type and their corresponding areas), can also be read in LOTUS 1-2- 3 spreadsheets using the LOTUS Commands: / File Import Number. There they can be sorted by categories and other data such as percentage of circulation and open space, average plot size, persons per plots can be added to the table to evaluate population density. 48
D. Establishinga HousingTypology Usina "SPOT"Satellite Images and a Set of Aerial Photographs 3.40 A housingtypology is a classificationof differentresidential patterns of development or urban fabrics that have the same physical and socioeconomiccharacteristics. A housing typology may structureor complementsocioeconomic data gatheredthrough censusand sample surveys.Such study helpsdisaggregate data spatiallyafter different patterns of developmenthave been identifiedfrom a satellite image. (See Figure3.6 for the Identificationof Different Urban Featuresfrom "SPOT"Panchromatic Print.) 3.41 The methodologyfor developingthe typology,selecting housing types, gatheringdata and performingthe appropriateanalyses is describedin this section. Selectingthe HousingTypes 3.42 Housingtypes are selectedfrom satelliteimages or aerial photographsaccording to distinctivedevelopment patterns consisting of plot and dwellingunit sizes,construction types, and density.Areas to be studiedare selectedfrom aerialphotos enlarged at scales1:5,000 to 1:10,000. If aerial photosare not available,a typology may be developedfrom SPOTsatellite images. Ideally, both aerial photos and SPOT satellite images should be used because each has relative advantages.A satellite image covers an area of 60km x 60km, eliminatingthe need for photo mosaicswhen analyzinga large urban area such as Karachi.On the other hand, aerial photos provide excellentresolution for moredetailed analysis. 3.43 Using a panchromaticphotographic print at the scale of 1:24,000,four different patterns of developmentwere identifiedby the study. Some of the selected urban patternswere blown up to a scale of 1:6,000after computer enhancementof the image. While these prints (Figures3.15 to 3.18)do not offerthe same detailas aerial photosat the same scale,they can be used to evaluatedensities based on block dimensions,street widths, and the numberof plots per block. This was possibleon two of the prints (Figures3.17 to 3.18)where streetsand blocks were well defined.
3.44 The resolution of a SPOT image is not precise enough to evaluate densities in compact urban areas,such as squattersettlements and the congestedcity center.Aerial photos coupled with field surveysare recommendedto calculatewith accuracy urbandensity. However, recent SPOT images in combinationwith older aerial photos can produce the same level of accuracy.
3.45 In the case of Karachi,both SPOT imagesand aerial photographswere used but it was not possibleto obtainfull aerialcoverage for the entirecity. The availableset of photos,dated from October 1988,covered three areas: a strip along the LayariRiver, Baldia, and Korangi.These photoswhere used to moreaccurately evaluate the four types of developmentpreviously selected from satelliteimages. From the aerial photo (shownin Figure3.19) it was possibleto differentiate a fifth residentialtype, an apartmentcomplex, which was misclassifiedin the first interpretationof the satelliteimages. Most apartmentcomplexes are locatedalong the HyderabadSuper Highway and in the center of Karachi. 49
FIGURE 3.15
Type A - SquatterSettlement "SPOT" Panchromatic Satellite Image PhotographicPrint at Scale 1:6,000 Karachi, Pakistan
DENSITY:650 PERSONS/HA
Copyright 1986 CNES. SPOT Image Corporation, Reston Virginia This photographic print is the result of 3x3 high'pass filter Kernel applied to enhance urban detail. 50
FIGURE 3,13
Type B - !-Hih) desiv 'SPOTi' PainchromE4~,tc'S'a+emte~ imaage PhotographicPrint aL-Scaba 4:6,000 Karachi,Pakistan.
DE NSIT Y: 625 P ERISONS/HA
Copyridhi. ,86 CES. m3, Cor p,oration, Reston Virginia
This pho--tgra,pi>cPr ni .sio--e resui, cy 3x3 high5passfilter Kernelapplied tQ enhanceurban detail. 51
FIGURE3.17
Type C -Medium density "SPOT"Panchromatic Satellite Image PhotographicPrint at Scale 1:6,000 Karachi,Pakistan DENSITY:260 PERSONS/HA
Copyright 1986ONES. SPOT Image Corpo'ation, RestooVirginia
This photographicprint is the result of 3x3 high'pass filter Kerne' apnpie-dt!, enhanceurban detail. 52
FIGURE3.18
Type D - Low Density "SPOT" Pancnromatic Satellite Image PhotographicPrint at Scale 1:6,000 Karachi,Pakistan DENSITY:102 PERSONS/HA
Copyright~~~ ~ PTIaeOrxain ~ 198 etnVrii ENS This~ ~~ phtgahcpitih euto x ifpssfle enlapidt nac ra eal 53
3.46 After the different urban patterns where identifiedfrom satellite images, identical patterns of development were extracted from the aerial photos and average densities were calculated.From aerial photos, block dimensions,street widths, percentagesof circulationand open space, plot sizes,and numberof plots per block could be evaluatedand used to calculate averagedensities (as shown in Figures3.20 to 3.22). Types A and C were reclassifiedinto sub- types since plot size variationsbetween these types may changethese averages. 3.47 Densitywas calculatedon the basis of block dimensions(from road center line), percentageof open space for each type of development (percentageestimated from satellite photographic print at a scale of 1:24,000),the number of plots per block and the number of persons per plot. Householdsize was drawn from secondary data and was verified by selected field surveys. Density (personsper hectare)was calculatedaccording to the following formula: D = (n x p) / ((bi x b2) + (bi x b2)/100 x s) x 10000 where D = Densitypersons/ha n = Number of plots in block p = Numberof personsper plot bi = Block length b2= Block width s = Percentage of open space
3.48 Variousdensities calculated from the sevendifferent types are listed in Table3.1. The densitiesestimated from satelliteimages were updated to reflect the data calculatedfrom aerial photographic anaiysis. On the basis of this analysis it was possible to estimate Karachi's population.A map of Karachi'spopulation distribution by housingtype was derivedfrom this study and is presentedin Figure4.13. TABLE3.1 AVERAGEDENSITIES AMONG DIFFERENTTYPES OF RESIDENTIALDEVELOPMENT
Density Type Description Persons/Hectare
Al Katchi Abadi - Next to waterways & railways 655
A2 Katchi Abadi - Illegal subdivision - 496 B Semi Pucca Houses - Old city center 630 Cl Semi Pucca & Pucca Houses - Small plot size 424 C2 Pucca Houses - Medium plot size 213 D BungaLow - Houses on large plot size 110 E Apartment Complex 1125 54
'_tF 0 F,,FE REN T H0 US IN G TYP ES G-4RAFFIHAT SCALE 1:4,000
0, A. Q. X ArrNMm 5:4 TYPE C "N
"k.
,T S
I-M F. A J
m T A J,gg RW "j-
A K 7, A-2
9 N- -
A
V Q T Y PE Q
J :
TYPE E TYPE C APART
Rai;,,mays TYP E A: Squn er Lo' Ai,- Ir- a,,I s,ard X TYPE B: Srnali Plot :-;,zs - Sulb-&