Interactive Data Visualization Tools for Travel Demand Model Datasets Shuyao Hong GIS-T Symposium Phoenix, Arizona April 10-13, 2017
MAG Modeling Area Area: 16,080 Square Miles Population: 4.9 Million Employment: 2.1 Million N Transportation system analysis and forecasting are critical components in the regional transportation planning process. Analysis of current travel patterns Forecasts of future travel demand? Planners MAG Metropolitan Planning Area County Boundary Decision Makers Stakeholders General Public 2
We searched through DOTs and large MPOs websites in late 2015: All DOTs and 90%+ of large MPOs have at least one data visualization online for their transportation-related data. This has been the direction we are heading in for our transportation-related data. Online interactive visualization is the trend MAG Transportation Data Management System www.magtrans.org MAG Bottleneck Study BottleneckStudy.azmag.gov 3
Infrastructure Data Land Use Data Socioeconomic Data Traffic Data Translating geospatial data involved to easy-to-digest insights could be challenging Travel Data Transportation Models Regional 4-step trip-based transportation forecasting model Mega-regional Activity- Based Model Regional Microsimulation Model Mega-regional Behavioral Agent-based Freight model. Truck models. Special events model. etc. Transit Trips Vehicle Trips Travel Pattern Modal Spilt Goods Flow Special Events Airport Travel LOS VMT Auto Occupancy Freight Demand Vehicle Trajectory ASU Travel Ex 4
Figure 1.2 INPUTS County Business Pattern Data Macro and Meso zone shapefiles SED data (employment by type) BEA s I/O Make and Use Table Highway and Rail networks Location of IMX, DC, warehouses SED Data Model skims (times, distances) SED Data Multiclass Daily truck flows by TOD Case Study: MAG Next Generation Freight Demand Model Funded by FHWA SHRP2 C20 IAP Grant, a joint proposal by ADOT, MAG, and PAG to build a proof-of-concept, advanced freight demand model. Supply Chain Tour-Based Freight Modeling Framework MODEL MODEL ESTIMATION MODELS OUTPUTS GEOGRAPHY CALIBRATION/ Awarded Best Data Fusion Application by TRB DATA VALIDATION DATA Supply Chain and Logistics- Trucks Macro zones TRANSEARCH Standing Committee on Freight Data (ABJ90) Based Freight Model (carrying external External station FAF3 Establishment surveys (to collect establishment, shipment, cost information) ATRI Truck GPS Data Other vendor(s) GPS Data Commercial vehicle establishment surveys VIUS Data DMV Data Commercial vehicle establishment surveys VIUS Data DMV Data Green Public Data / Borrowed Models Blue Available to MPO / Current MPO Models Orange Developed as part of MAG SHRP2 C20 Grant Source: Cambridge Systematics, Inc. FIRMS, SUPPLIERS, COMMODITY FLOWS, SHIPMENTS, MODES, PATHS Trucks (freight) Trucks (nonfreight) Rail Pipeline Air Tour-Based Truck Model TOUR AND STOP GENERATION, STOP PURPOSE, LOCATION AND TOD CHOICE Service Sector Model FUNCTION OF POPULATION AND EMPLOYMENT DISTRIBUTIONS Behavioral Trip Assignment Model MULTICLASS ASSIGNMENTS ALONG WITH AUTO VEHICLE TRIP TABLES freight) Rail Pipeline Air (Daily Trip tables by Industry, TOD) Trucks (carrying freight) Trucks (nonfreight) Trucks (nonfreight) Daily truck flows by TOD Mesozones internal (Macro to Meso, Meso to Macro, Meso to Meso) Meso zones to TAZ level TAZ to TAZ level TAZ level Mega- Regional Multi- Modal Agent- Based Link level truck counts Compare external flows versus current freight model flows Compare freight mode shares versus current freight model shares O/D survey data (trip rates, TLFDs, TOD) Compare tour-based model results versus trip-based model Estimates from other regions Compare service sector trip rates from other models Plot spatial distribution of service truck Os/Ds versus pop and employment Truck VMT by county Vehicle classification data screenlines Compare new freight model versus existing freight model volumes Purple Collected as part of 2014 On-Call Gold Outputs of Supply Chain Tour based Freight Model Figure 1.4 Individual Components of the Proposed Behavior-Based Freight Model More about the model can be found: MAG SHRP2 C20 Mega-Regional Multi-Modal Agent-Based Behavioral Freight Model Final Report. http://www.azmag.gov/projects/project.asp?cmsid=1137 INPUTS MAG / PAG / AZDOT TAZ Shapefiles (MAG 3009 TAZs, 231 RAZs, 45 MPAs) CBP Data for U.S. (2012 data available in June 2014) (County level for AZ, State level for non-az) Correspondence (lookup) NAICS6 SCTG (Industry to Commodity lookup table) BEA s Input-Output Make and Use Tables (Make what commodity is produced by an industry?) (Use who uses certain commodities?) BEA s Input-Output Make and Use Tables (Make-Use Table percent of commodity used by industry) ORNL Networks for Distance Skims FAF3 (or FAF4 if available) (Annual Tons by Commodity at BEA Level) BEA s Input-Output Make and Use Tables (Make-Use Table percent of commodity used by industry) Highway and Rail Networks Distance and Time Skims Location of IMX Facilities (in AZ and outside AZ) Shipping Cost Data MODELS OUTPUTS GEOGRAPHY Study Area/GIS Tasks Macro Outside AZ State Level Meso Inside AZ RAZ or MPA Level Firm Synthesis Model Computes number of firms by NAICS6, SCTG, Size Disaggregates firms from County to Meso Zones using Employer Determines Supplier (Shipper) Firms Determines Buyer (Consumer) Firms Supplier Selection Model Links Buyers with Suppliers Uses Size of Firm, Distance, Industry-Commodity Lookups Apportionment of Commodity Flows Disaggregates Commodity Flows among Supplier-Buyer Pairs Based on Buyer Firm Size and Tons of Goods Consumed Transport and Logistics Path Choice Model Generates Skims for Highway, Rail, Water, and Air Ben-Akiva/De Jong s Transport and Path Choice Model Generates Most Optimal Path (given costs, times, distances, location of IMXs, supplier-buyer combinations) Rail Pipeline Air Macro shape file Meso shape file TAZ shape file List of Supplier (Make) Firms List of Buyer (Use) Firms Supplier (Shipper) Meso Zone Linked with Buyer (Receiver/Buyer) Annual Tons Traded between Supplier and Buyer Selected Path and Shipment Size for Each Supplier-Buyer Combination Modeling and Data Consultant Team: Cambridge Systematics, RS&H, ATRI, CDM SMITH, StreetLight Trucks Meso Zone Meso Zone Meso Zone Meso Zone MODEL CALIBRATION DATA MPO/State SED data (TAZ level data, InfoUSA, etc.) Reasonableness checks TRANSEARCH Data Existing Truck Model FAF3 (or FAF4) paths TRANSEARCH Data 5
Challenges Various data types: OD matrix (trucks, goods) Link-level truck volume Trajectory (Truck GPS data) Various geographic level International National Regional Various modeling components/layers Wish List Customizable/Flexible Quick turnaround We want something cool to present next week. Low costs Can we do it in-house? Easy to deploy Can you do it yourself? Limited resources (i.e. time, money, staff) allocated for visualization 6
Live Demo Live demo is replaced by the following screenshots in this downloadable version of slides. 7
Polygon Feature Extrusion: Number of simulated employment per TAZ 8
Desire Line: Commodity Flow by Type of Goods 9
Interactive Desire Line: Trucks Flow OD Matrix 10
Volume Map Comparison: Trucks Volume 11
Truck GPS Traces Animation 12
Truck GPS Heat Map 13
Simple Workflow 1 Staff 1 Week, From Scratch to Prototypes to Presentable Visualizations Post-Processing Modeling Input/Output Data File csv, json, etc. Spatial Files (ESRI shapefile, TransCAD network, etc.) Format Conversion Spatial Info GeoJSON 14
Open Source Libraries make visualization development much easier and faster Leaflet: http://leafletjs.com/ The leading open-source library for interactive maps. A wide range of plugins for visualizing geospatial data beyond simple mapping. MovingMarker : trajectory animation Heat : heat map SpatialSankey : interactive desire line MarkerCluster: automatically cluster points at different zoom level. Mapbox GL JS: https://www.mapbox.com/mapbox-gl-js/api/ Rendering interactive maps in 3D while allowing users to control the perspective. Data-Driven Styling: Style spatial features (point, polyline or polygon) with different color, width/radius, and extrusion height, based on associated data. 15
Web Development Technology is advancing fast We don t need to be a web developer ourselves, but Model Data Web Dev GIS Low Costs, Big Impacts Visualization Projects 16
Thank you! Questions? Shuyao Hong (602) 254-6300 shong@azmag.gov GIS-T Symposium Phoenix, Arizona April 10-13, 2017