1. 1 21 st Century MAF/TIGER Enhancements An Integral Component of a Re-engineered 2010 Census By the U. S Census Bureau CLEM2001 Specialist Meeting August 6-7, 2001 Santa Barbara, CA

2. 2 American Community Survey (ACS) 2010 Census Redesign for Short Form Census MAF/TIGER Modernization... American Community Survey (ACS) 2010 Census Redesign for Short Form Census MAF/TIGER Modernization... Planning for 2010

3. 3 governmental unit, census tract, census block In order to be useful, U.S. Census Bureau statistics must be: data accuracy and geographic location data accuracy and geographic location correct reliable

4. 4 What is TIGER? Topologically Integrated Geographic Encoding and Referencing TIGER Content Streets and their names Lakes, streams, and their names Railroads Geographic entity boundaries, names, and codes (for governmental units, census tracts, census blocks, etc.) Housing unit locations Key geographic locations (for airports, schools, etc.) ZIP Codes and address ranges (for streets with city-style addresses). A “digital map” (geographic data base) of the entire United States, Puerto Rico, and the associated Island Areas

5. 5 An accurate and up-to-date inventory of all known living quarters in the United States, Puerto Rico, and associated Island Areas Master Address File The content of the MAF is: -Mailing address, if one exists -Descriptive address, if no mailing address is known - Census geographic location - Source and history data All addresses and locations MUST be kept confidential What is the MAF?

6. 6 Major MAF/TIGER Functions Maps (and address lists) for field operations, such as Block Canvassing Maps for data user reference Mapping Determines geographic location of every structure Identifies areas that need MAF/TIGER update Geocoding Data Products Provides names and codes of entities for data tabulation TIGER/Line fuels the commercial GIS industry

7. 7 Mason Starbridge Major MAF/TIGER Issues Location Information of Mixed/Variable Accuracy Large, non-uniform differences in location accuracy exist in a relatively small area, and no detailed quality measures document the extent of street and address errors 10m 100m

8. 8 “Truth” Contradicts Existing Feature Topology Kirkwall Alcorn Bayuo Stream mistakenly crosses several streets to form “census blocks” that do not exist The problem is how to show the true situation correctly in the future, while maintaining an historical link to Census 2000 block numbers Major MAF/TIGER Issues

9. 9 Good GPS positions over incorrect MAF/TIGER locations would put many houses on the wrong side of the street and, therefore, in the wrong census block Inaccurate locations preclude adopting GPS locational technology for the American Community Survey and the 2010 Census until MAF/TIGER has locations corrected

10. 10 Major MAF/TIGER Issues Original TIGER Data and State/Local/Tribal Roads TIGER GIS Roads in TIGER/housing units in MAF not in “true” geographic locations Inaccurate MAF/TIGER locations constrain efforts to exchange highly accurate location information with willing geographic partners that have GIS files In addition, the existing “home grown” MAF/TIGER processing environment makes development of Web-based review and update processes more cumbersome

11. 11 No process currently exists to update MAF/TIGER with new addresses (and the new streets along which many are located) in areas that do not have (or use) city-style addresses for U.S. Postal Service mail delivery ? Major MAF/TIGER Issues In areas that will not benefit from the twice-yearly “refreshes” of MAF/TIGER with address and street updates from the USPS’s Delivery Sequence File, because the existing housing units do not have (or use) addresses that are “city- style,” the Census Bureau has no automated update source

12. 12 Five Objectives for MAF/TIGER Modernization Correct locations of streets, other map features, and housing units; identify new streets/housing units using automated change detection methods Develop/deploy a new MAF/TIGER processing environment based on COTS and GIS tools Expand and encourage geographic partnership programs Launch the Community Address Updating System (CAUS) Implement periodic evaluation activities

13. 13 Correcting MAF/TIGER Locations Primary Strategy: Use Highly Accurate State/Local/Tribal GIS Files, Where Available TIGER GIS Highly accurate GIS files are available for hundreds of local/tribal governments These files provide the most effective information to correct MAF/TIGER locations, and often are a good source for new streets and addresses

14. 14 Correcting MAF/TIGER Locations Secondary Strategy: Use Highly Accurate Private-Sector GIS Files, Where Available TIGER GIS GIS files also exist in the private sector for some areas When they are available for MAF/TIGER use without restrictions, they provide an effective source to correct MAF/TIGER locations, and sometimes for new streets and addresses

15. 15 Space-based technologies present new opportunities for improving MAF/TIGER locations and detecting changes when GIS files are not available One Meter Resolution Imagery Six Inch Resolution Imagery GPS Correcting MAF/TIGER Locations

16. 16.............................................................................................................................................................................................................................................................. Existing MAF/TIGER superimposed on accurate location information MAF/TIGER data are moved to accurate locations A matching and re-alignment process is required to use all GIS files and all locations extracted from imagery. The goal is the same –- to correct existing MAF/TIGER locations while preserving all other information MAF/TIGER contains about each street and address

17. 17 In the future, correct locations in MAF/TIGER, and devices equipped with GPS receivers, will provide the tools needed by field staff to find the correct housing unit and validate the accuracy of each address Finding the Correct House

18. 18 MAF/TIGER Modernization lCurrent Status Request for Information – February, 2001 l List of respondents on the Internet Visits to Selected Vendors – April-May, 2001 l Additional Market Research l Best Practices l Current Capabilities l Utilization of Tribal/State/Local files l Imagery – satellite, airborne l LIDAR – Light Detecting and Ranging using Laser l IFSAR – InterFerometric Synthetic Aperture Radar

19. 19 Other Activities Underway (ITS088) Purpose: Improving Census 2000 data products Improve the positional accuracy and shape fidelity of the coordinates of selected TIGER/Line 2000 Files Add features that currently are missing to the Files Additionally, collect structure centroid coordinates Flag existing features in the Files that are not on the source data

20. 20 Other Activities Underway ITS 088 was awarded in September 2000 to The Centech Group Visual Intelligence Systems, Inc Acquire LIDAR, Imagery Schafer Wireless Technologies Feature extraction, TIGER realignment RS Information Systems (RSIS) Pixxures Acquire/rectify imagery Pictoform Feature extraction, TIGER realignment Harris Corp, BTG, GDT TIGER realignment from imagery or local files

21. 21 The Centech Experience lExisting TIGER/Line files contain spatial and topological errors as a result of dissimilar data sources and use of digitized aerial photographs for feature generation lCensus is demonstrating the use of co-registered aerial imagery and LIDAR data to improve and correct the TIGER/Line database TIGER/Line Roads (red) Extracted 2D Vector Road Data (green)

22. 22 Centech Project Goals lUse AFE to rapidly create an accurate, visible feature set: Roads, Hydrology, Railroads, Powerlines and Structures/Dwellings lDemonstrate that a feature set can be rapidly created over large areas of different morphologies lDemonstrate feature positional accuracy of 3 - 10 m lClassify features using defined attributes lEnsure topology is maintained for TIGER features lUse human-in-the-loop processes as quality control

23. 23 What is LIDAR? lLight Detection and Ranging using a Laser lReflections from most objects, hard surfaces, and terrain lVery accurate height and location information about features lSmall laser spot size allows identification of feature details Grayscale depicts object heights

24. 24 Advantages of LIDAR – Elevation Data House Roads Accurate Height Information is Critical for Identifying Individual Features with a High Confidence Level Elevation/Height Profile Plot

25. 25 Advantages of LIDAR – Elevation Data LIDAR Allows for Feature Classification using Multiple Attributes

26. 26 Advantages of LIDAR – Obstruction Removal Structures and terrain hidden by trees in aerial imagery can identified with LIDAR Obscured Hydrology Obscured Structures Foliage Removed

27. 27 Advantages of LIDAR - Classification

28. 28 Advantages of LIDAR - Classification

29. 29 Primary Limitation of LIDAR Boundary definition of similar features Green areas represent roads Bright green = high confidence Dark green = low confidence Features with Similar Height and Reflectance Properties are Difficult to Distinguish

30. 30 Major Challenges to AFE using LIDAR lData Requirements and Quality Adequate data density - 1 return per m^2 Absolute position and height accuracy 2 m horizontal and.5 m elevation Biases, errors and noise removed lNear-Infinite Feature Space Model matching does not work Intelligent software is required lAccurately Determining Feature Shape, Boundaries and End-Points Maintaining planar topology lEstablishing In-Process QC Metrics to Ensure that the AFE Results Meet Census Requirements

31. 31 LIDAR AFE Software lSoftware is proprietary to Schafer Corp. Developed to create 3-D city models for telecom and geo-spatial industries l Able to extract buildings, roads, hydrology, foliage, power transmission lines, railroads, and pipe lines. l Feature classification and sub- classification under evaluation l Performs raster to vector data conversion l Data formats compatible with commercial GIS applications Is often used in combination with digital photography to improve performance

32. 32 Centech Summary lCurrent AFE Results Near 100% of buildings/structures extracted (1% false alarms) 95% of primary roads and highways extracted Near 100% of major hydrology features extracted Human-in-the-loop being applied to ~5% of visible features l Goal is less than 2% Current processing time allows hundreds of square miles per month to be completed l Goal is 500 to 1000 square miles per week

33. 33 The RSIS Experience Fully Automated Feature Extraction Based on feature geometry and pixel color values ERDAS Imagine 8.0 with Sub-Pixel Classifer Tool tested (gray-scale and color imagery) Feature isolation using specified pixel coloring Feature filtering using geometric characteristics Manual removal of “false features” Perfection of feature line quality

34. 34 The RSIS Experience Problems: Quantity of “false features” quite high Number of gaps in identified features Vectorization – unreliable “skeletonizing” of features

35. 35 The RSIS Experience Semi-Automated Feature Extraction Hitachi TRACER software Purported to be able to follow a feature once a point on the feature manually identified Product requires a stark contrast between the object of interest and its surroundings Not usable on imagery

36. 36 MAF/TIGER Modernization lNext Steps Determine Requirements Determine Acquisition Process l Use State/Local/Tribal Files FIRST l Use Private Sector Files SECOND l Contract for remaining areas and elements not obtained from State/Local/Tribal/Private Sector files Draft RFP by Labor Day l Draft Statement of Objectives shortly Goal is award around the end of calendar year