Object-Oriented Image Object-Oriented Image Classification of Brownfields in Classification of Brownfields in
Syracuse, NYSyracuse, NYGreg BaconGreg Bacon
Master of Science Degree CandidateMaster of Science Degree CandidateEnvironmental Resources and Forest EngineeringEnvironmental Resources and Forest Engineering
SUNY College of Environmental Science and ForestrySUNY College of Environmental Science and Forestry
April 5, 2006April 5, 2006
Discussion TopicsDiscussion Topics
Introduction to Brownfields and Introduction to Brownfields and RedevelopmentRedevelopment
Site IdentificationSite Identification
Research Objectives and ProcessResearch Objectives and Process
Additional Considerations Additional Considerations
SummarySummary
IntroductionIntroduction
Brownfield DefinitionBrownfield Definition
“…“…real property, the expansion, redevelopment, or real property, the expansion, redevelopment, or reuse of which may be complicated by the reuse of which may be complicated by the presence or potential presence of a hazardous presence or potential presence of a hazardous substance, pollutant, or contaminant.” substance, pollutant, or contaminant.”
Section 211(a) of the Small Business Liability Section 211(a) of the Small Business Liability Relief and Brownfields Revitalization Act of 2002 Relief and Brownfields Revitalization Act of 2002 (Pub.L. 107-118) (Pub.L. 107-118)
Current StatusCurrent StatusEPA estimates there are 500K – 1M U.S. EPA estimates there are 500K – 1M U.S. brownfield sites brownfield sites
85-90% of these not evaluated or cleaned 85-90% of these not evaluated or cleaned upup
Brownfields Revitalization Act expected to Brownfields Revitalization Act expected to expand number of sites assessed for expand number of sites assessed for cleanup/redevelopmentcleanup/redevelopment– Liability protectionLiability protection– Grant fundingGrant funding
Source: U.S. EPA, 2004b
Brownfield RedevelopmentBrownfield RedevelopmentBenefitsBenefits
Grants available to “eligible entities” forGrants available to “eligible entities” for– Site inventorySite inventory– CharacterizationCharacterization– Assessment Assessment – PlanningPlanning
– Increase tax base
– Job growth
– Conserve open land
– Use existing infrastructure
– Improve environment
How do you find them?
Source: U.S. EPA, 2004a
Brownfield Site IdentificationBrownfield Site Identification
Traditional Site IdentificationTraditional Site Identification
Government derived information: tax/ Government derived information: tax/ ownership records, state environmental ownership records, state environmental datadata– Currency, completeness, costCurrency, completeness, cost
Site visitsSite visits– Site access, practicality, costSite access, practicality, cost
City of Syracuse site inventory used EPA City of Syracuse site inventory used EPA grant grant – Reference data for accuracy assessment Reference data for accuracy assessment
Research ObjectivesResearch Objectives
Apply a brownfield site identification Apply a brownfield site identification method to produce a GIS-ready productmethod to produce a GIS-ready product– More efficient resource useMore efficient resource use– Visual supplement to other site inventory Visual supplement to other site inventory
methods methods
Evaluate accuracy of classificationEvaluate accuracy of classification– Could this be a useful tool in other places?Could this be a useful tool in other places?
Source: Myeong et al., 2001.
City of Syracuse Land Cover Thematic Land Cover Map
ModelingModeling
AnalysisAnalysis
Suitability StudiesSuitability Studies
No Indication of Land Use
Need more informationNeed more information
New classification New classification procedure can help to procedure can help to address thisaddress this
Classify “image objects,” not pixelsClassify “image objects,” not pixels
Classification based on spatial context Classification based on spatial context rulesrules
Classify complex ground features Classify complex ground features
Object-Oriented Image ClassificationObject-Oriented Image Classification
Example ApplicationsExample Applications
Built-Up LandBuilt-Up Land– Johnsson, 1994Johnsson, 1994
Undeclared Nuclear FacilitiesUndeclared Nuclear Facilities– Niemeyer and Canty, 2001Niemeyer and Canty, 2001
Forest Cut BlocksForest Cut Blocks– Flanders Flanders et. alet. al., 2003., 2003
BrownfieldsBrownfields– Banzhaf and Netzband, 2004Banzhaf and Netzband, 2004
ProcessProcess
Land Cover Classification
Structure GroupAssignment
Classification
Export Output
Rule Refinement
Data
Knowledge
Image Segmentation
Rule Development
Project Data Needs Project Data Needs
Syracuse streets (vector shapefile) Syracuse streets (vector shapefile)
Tax parcels (vector shapefile)Tax parcels (vector shapefile)
Brownfield addresses (Excel spreadsheet)Brownfield addresses (Excel spreadsheet)
Emerge Imagery Emerge Imagery – NIR, red, green bandsNIR, red, green bands– 0.61 m (2 ft) ground sample distance0.61 m (2 ft) ground sample distance– 8-bit radiometry 8-bit radiometry – Collected 13 July 1999Collected 13 July 1999
What Does a Brownfield Look Like?What Does a Brownfield Look Like?
Radja, 1994
Lillesand et. al., 2004
Input LayersInput Layersfor for
SegmentationSegmentation
255*31
31
bb
bbNDVI
222 321
255*11
bbb
bb chrom
Image Object Creation (Segmentation)Image Object Creation (Segmentation)
Scale Parameter = 25
Scale Parameter = 100
Image Objects – Lives of Their OwnImage Objects – Lives of Their Own
Rule DevelopmentRule Development
Rule DevelopmentRule Development
Combinations of Combinations of functions can be functions can be appliedapplied
Working with object Working with object values directlyvalues directly
TransparencyTransparency
Land Cover Land Cover Classification Classification
Level 1Level 1
Land Cover Land Cover Classification Classification
Level 2Level 2
Level 1 Objects
Extracted from Level 2
Structuring of Structuring of Image ObjectsImage Objects
Potential Brownfield Site
Land cover classes Land use indicator
Classification Classification StabilityStability
Low (ambiguous class assignment)
High (good class separation)
Classification StabilityClassification Stability
Classify smaller, Classify smaller, more homogeneous more homogeneous objectsobjects
Refine rulesRefine rules
Create a new classCreate a new class
Live with itLive with it
Tree Grass
0.860.83
Mem
bers
hip
Mem
bers
hip
Tree Grass
0.89
0.62
Accuracy AssessmentAccuracy AssessmentOutput vector layer of potential brownfield parcelsOutput vector layer of potential brownfield parcels
Evaluate classification based on agreement with Evaluate classification based on agreement with reference datareference data
Error Matrix
Additional ConsiderationsAdditional ConsiderationsBrownfield definition Brownfield definition – What qualifies as a brownfield is debatableWhat qualifies as a brownfield is debatable– Characteristics not described by legal definitionCharacteristics not described by legal definition– Remote sensing alone cannot fully examine site Remote sensing alone cannot fully examine site
functionfunction, only , only formform
Accuracy IssuesAccuracy Issues– Quality of land cover classification directly affects land Quality of land cover classification directly affects land
use indicatoruse indicator– Completeness and quality of reference dataCompleteness and quality of reference data– Temporal difference between image and reference Temporal difference between image and reference
data collectiondata collection
SummarySummaryBrownfields represented by group of Brownfields represented by group of collocated cover typescollocated cover types– Accuracy is affected by strength of this Accuracy is affected by strength of this
assumptionassumption
Object-oriented classification Object-oriented classification – Attempt to imitate human pattern recognitionAttempt to imitate human pattern recognition– Membership functions classify objects on a Membership functions classify objects on a
sliding scalesliding scale
Transition from land cover to land use Transition from land cover to land use
AcknowledgementsAcknowledgements
Dr. Lindi Quackenbush – SUNY ESF Faculty of Dr. Lindi Quackenbush – SUNY ESF Faculty of Environmental Resources & Forest EngineeringEnvironmental Resources & Forest Engineering
Dr. Stephen Stehman – SUNY ESF Faculty of Dr. Stephen Stehman – SUNY ESF Faculty of Forest & Natural Resources ManagementForest & Natural Resources Management
Mr. Mike Haggerty – (formerly) City of Syracuse Mr. Mike Haggerty – (formerly) City of Syracuse Department of Economic DevelopmentDepartment of Economic Development
Ms. Amy Santos – Environmental Finance Center, Ms. Amy Santos – Environmental Finance Center, Maxwell School of Citizenship and Public AffairsMaxwell School of Citizenship and Public Affairs
ReferencesReferencesBanzhaf, E. and M. Netzband, 2004. Detecting Urban Brownfields by
Means of High Resolution Satellite Imagery. International Society for Photogrammetry and Remote Sensing (ISPRS) Conference Proceedings, July 2004, Istanbul, Turkey.
Flanders, D., M. Hall-Beyer, and J. Pereverzoff, 2003. Preliminary Evaluation of eCognition Object-Based Software for Cut Block Delineation and Feature Extraction. Canadian Journal of Remote Sensing. 29(4), 441-452.
Johnsson, K., 1994. Segment-Based Land-Use Classification from SPOT Satellite Data. Photogrammetric Engineering and Remote Sensing. 60(1), 47-53.
Lillesand, T.M., R.W. Kiefer, and J.W. Chipman, 2004. Remote Sensing and Image Interpretation, Fifth Edition, John Wiley & Sons, Inc., New York, 763 p.
Myeong, S., D. Nowak, P. Hopkins, and R. Brock, 2001. Urban Cover Mapping Using Digital, High-Spatial Resolution Aerial Imagery. Urban Ecosystems. 5, 243-256.
References (cont’d)References (cont’d)Niemeyer, I. and M.J. Canty, 2001. Knowledge-Based Interpretation of Satellite
Data by Object-Based and Multi-Scale Image Analysis in the Context of Nuclear Verification. Proceedings of the International Geoscience and Remote Sensing Symposium (IGARSS), July 2001, Sydney, Australia,. 7, 2982-2984. URL: http://www.niemeyer.de/publications/igarss01nie.pdf.
Radja, P.G., 1994. Green: Segmentation of an Aerial Video Recording for Tree Counting, M.S. Thesis, University of Illinois at Urbana-Champaign, 104 p.
U.S. Environmental Protection Agency 2004a. Brownfields Assessment Grants: Interested in Applying for Funding? EPA560-F-04-254, URL: http://www.epa.gov/brownfields/facts/fy05assessment_factsheet.pdf.
----- 2004b. Cleaning Up the Nation’s Waste Sites: Markets and Technology Trends, 2004 Edition, EPA542-R-04-015.