Conflating a Traffic Model Network With a Road Inventory. David Knudsen

Size: px

Start display at page:

Download "Conflating a Traffic Model Network With a Road Inventory. David Knudsen"

Shawn Banks
5 years ago
Views:

1 Conflating a Traffic Model Network With a Road Inventory David Knudsen

2 Traffic modelers need to derive attributes of their abstracted networks from road layers maintained by highway departments, and planners want to present model predictions using these road layers, which the public recognizes and understands. GIS and data analysts may be responsible for developing and maintaining a codified relationship between the two data stores, neither of which they own or control. In the worst possible case, both stores may be completely redesigned simultaneously, breaking previously created relationships. Both out-of-the-box and ArcObjects-customized GIS tools making use of path finding on network datasets, topologies, versioned editing, SDE views, and carefully-designed symbolization ease the task of conflating two related but dissimilar data layers and errorchecking the resulting many-to-many relationship. With both planning and hindsight, the conflation can become more responsive to continuing edits to the member data layers and more resilient to future wholesale changes. Conflating a Traffic Model Network with a Road Inventory 2

3 Traffic Modeling Primer Traffic analysis zones Network defined by nodes and links Classic four-step modeling process Trip generation Trip distribution Mode choice Route assignment Predicts traffic volumes, speeds, travel times Zone 840 Zone Zone Zone Zone 836 Centroid Zone 830 Conflating a Traffic Model Network with a Road Inventory 3

4 GIS and Models Complement Each Other GIS is a source of model inputs demographic data network attributes, i.e. length lanes capacity GIS maps model outputs with greatest flexibility and sophistication Relationships must exist to exchange data Zone Zone Zone Zone 836 Zone 837 Centroid Zone Conflating a Traffic Model Network with a Road Inventory 4

5 CTPS First Road/Network Relationship Established in 2001 on a coverage using AMLs Coded road features with model node IDs Many-to-many cardinality ROADS MODEL NETWORK Conflating a Traffic Model Network with a Road Inventory 5

6 Road/Network Relationship Ends: Irreconcilable differences PK 16 SK 25 PK SK 1 A1 PK SK 4 A2 PK 14 SK 23 PK 10 SK 20 PK 2 SK A1 PK 15 SK 24 PK SK PK SK PK SK A1 PK SK Conflation of roads to new linework Orthophoto-derived centerlines more detailed, accurate than old DLGs New primary key No published conversion from old to new primary key Secondary keys not comparable Conflating a Traffic Model Network with a Road Inventory 6

7 Road/Network Relationship Ends: Irreconcilable differences (2) Expansion of model X X From 987 to 2727 zones New nodes and links for additional roads New centroid connectors New shape nodes Node renumbering Conflating a Traffic Model Network with a Road Inventory 7

8 CTPS New Road/Network Relationship Creating from scratch estimated to be no more costly than fixing and error-checking existing one ArcGIS desktop and server technology used to manage workflow perform semi-automated conflation aid quality control create plots with good cartographic quality Conflating a Traffic Model Network with a Road Inventory 8

Workflow Management ArcSDE versions Versions for each editor Parent, reconciliation version Status and progress Town-level table Map symbolization Verbal coordination

9 Workflow Management ArcSDE versions Versions for each editor Parent, reconciliation version Status and progress Town-level table Map symbolization Verbal coordination still needed to avoid timeconsuming conflict reconciliation SDE.DEFAULT Public Editing Editor 1 Editor 2 Editor 3 Conflating a Traffic Model Network with a Road Inventory 9

10 Semi-automated Conflation Manually snap network nodes to appropriate road intersections 2. Automatically select roads along shortest path between intersections 3. Automatically transfer corresponding link IDs 4. Manually transfer link IDs when path finding fails Conflating a Traffic Model Network with a Road Inventory 10

11 Snapping Nodes Topology ensured links moved with nodes, introduced a problem (geometric network better choice) ArcObjects script set node symbol field when node dropped Topology errors do not provide same degree of progress-checking topology adds vertex at intersection link now has bend Conflating a Traffic Model Network with a Road Inventory 11

12 Shortest Path Selection INPUT ROAD FEATURE CLASS BUILD NETWORK DATASET No connection: OK overpass RESULTING NETWORK DATASET No connection: network error Shortest path selection is only provided in ArcGIS in network datasets under the Network Analyst A network dataset was built using roads as an input feature class GIS road data had some connection errors, which could cause the true shortest path not to be found Conflating a Traffic Model Network with a Road Inventory 12

Shortest Path FROM Four road features, Three network links Selection (2) 1 TO FROM 1. Shortest path solver creates path as new feature 2. ArcObjects code selects roads used in path feature 3.

13 Shortest Path FROM Four road features, Three network links Selection (2) 1 TO FROM 1. Shortest path solver creates path as new feature 2. ArcObjects code selects roads used in path feature 3. ArcObjects code transfers corresponding link IDs to selected roads if more than half of road used in path 4. Attribute Transfer Tool used when automatic path finding failed 2 FROM 3 4 TO TO Less than half 2 nd road used in path do not transfer FROM-TO link ID Conflating a Traffic Model Network with a Road Inventory 13

14 Conflation Logic Problems Model intersection simplification Snap node to which road intersection? Double barrels and ramp fans C B D B Bi-directional link represents both ramps. A Should extra, non-path roads be coded with link IDs? A Fields needed to store forward and reverse link IDs explicitly (B-A link inference from A-B link not always correct) Conflating a Traffic Model Network with a Road Inventory 14

been conflated to each road feature Road and link are conflated but have not had check lines generated for them

15 Quality Control Illustration Snapping node to intersection of Main and Albion Unedited node Roads with no link IDs Node snapped to road intersection Link labeled from conflated street Pink check lines show which link has been conflated to each road feature Road and link are conflated but have not had check lines generated for them Green check lines show which road has been conflated to link Conflating a Traffic Model Network with a Road Inventory 15

16 Quality Control Illustration Extra Credit Challenge Conflating a Traffic Model Network with a Road Inventory 16

17 Quality Control Careful symbolization key to controlling quality Visual checks/feedback originally envisioned for review stages integrated with initial editing as well Completeness addressed by showing nodes, links, and roads not yet edited with special symbols Correctness addressed by node symbols, check lines, and links labeled from conflated roads Thoroughness of review addressed by highlighting missing check lines using ArcSDE views Conflating a Traffic Model Network with a Road Inventory 17

greater information density (labels and other data

18 Checkplots Modelers liked to work on paper Network shown both prior to and after editing Paper allowed much greater information density (labels and other data layers) Conflating a Traffic Model Network with a Road Inventory 18

19 Mapping Model Output Volume-tocapacity ratios or congested speed mapped using fixed width and different colors Conflating a Traffic Model Network with a Road Inventory 19

20 Mapping Model Output (2) Volumes or volume changes mapped using varying widths and single color Conflating a Traffic Model Network with a Road Inventory 20

21 Preparing Data for Mapping MODEL OUTPUT ROADS WITH LINK IDs CONFLATED SHAPE POLYLINE POLYLINE ROAD ID 1 2 LINK ID A B B C IMPLIED/EXPLICIT REVERSE LINK ID B A C B LINK ID A B B A B C C B VOLUME SPEED SHAPE ROAD ID LINK ID IMPLIED/EXPLICIT REVERSE LINK ID LINK VOLUME LINK SPEED REV. LINK VOLUME REV. LINK SPEED POLYLINE POLYLINE 1 2 A B B C B A C B Model output keyed by link ID May need to be converted to table readable by GIS Joined twice to roads using both link ID and opposing direction link ID fields in road attributes Conflating a Traffic Model Network with a Road Inventory 21

22 Preparing Data for Mapping (2) To show directional data on correct side of line, must determine if direction of road feature matches direction of conflated links ArcObjects calculates direction of roads and their related links and sets a direction flag field accordingly ROAD ID 1 2 A NODE LINK ID A B B C 1 SAME DIR. YES NO 2 C NODE B A C B B NODE IMPLIED/EXPLICIT REVERSE LINK ID Conflating a Traffic Model Network with a Road Inventory 22

23 Preparing Data for Mapping (3) width: 48 pt offset: -24 pt width: 24 pt offset: -12 pt width: 36 pt offset: 18 pt ROAD AND LINK NOT SAME DIR. * MULTIPLY OFFSET BY 1 WHEN ROAD AND LINK NOT SAME DIRECTION width: 24 pt offset: 12 pt ROAD AND LINK SAME DIR. width: 48 pt offset: 24 pt ROAD/LINK SAME DIR. width: 36 pt offset: -18 pt LINK VALUE OPPOSITE DIRECTION LINK VALUE When symbol width varies for directional data, symbol offset should vary With classification, offset is hard-coded for symbol for each numeric range With proportional width, offset is infinitely variable and must depend on a field value, only supported in a route-event system Conflating a Traffic Model Network with a Road Inventory 23

24 Cartographic Considerations SIMPLE LINE SYMBOL CARTOGRAPHIC LINE SYMBOL Cartographic line symbol slower but better defined than simple line symbol Small symbols may be lost in, or hidden under, large symbols Cartographic line symbol has fill limitations No beveling is available at intersections Conflating a Traffic Model Network with a Road Inventory 24

is slow Through scripting and building geoprocessing models we

25 Work Continues! The QC tools we developed will help us keep up with new model scenarios as modelers code them Even with road-network relationship re-established, preparing model output for mapping is slow Through scripting and building geoprocessing models we want to streamline the process and improve cartographic quality Conflating a Traffic Model Network with a Road Inventory 25

26 Contact David Knudsen Ten Park Plaza, Suite 2150 Boston, MA Conflating a Traffic Model Network with a Road Inventory 26

Vision Zero High Injury Network Methodology

Vision Zero High Injury Network Methodology DATA SETS USED: 1. Reportable crashes in Philadelphia from 2012-2016, available as open data through PennDOT 2. Street Centerline geographic layer, maintained