Skip to content

3.Study site and Data

Jump to bottom
smart-fm edited this page Nov 9, 2018 · 7 revisions

3.1 Study site

SimMobilityST is calibrated by running the traffic for the extended Central Business District (CBD) in Singapore. This area contains more than 1200 intersections, which covered by more than 2000 loop detectors. A smaller sub-network called Bugis, located inside CBD (10 intersections), was also tested for assessing the impact of daily variability in the calibration process. The aggregated demand generated by SimMobility MT has 1497 observed ODs pairs and a total of 48,988 trips. These trips (demand), 11 route-choice parameters for demand and 112 driving behavior parameters (supply) are the set of parameters to calibrate.


Extended CBD area of Singapore (Shaded area)

3.2 Data

For the calibration, two types of data were available: loop count and GPS travel-time data from probe vehicles, collected in August 2013 by the Land Transport Authority of Singapore and Taxi data location respectively. Counts had a resolution of 5min while the GPS data was structured in OD a travel-time tables for each 30min interval of the day. Counts were also preprocessed for outlier detection. A total of 360 sensors were used in the calibration. The following data was also used in network settings: SCATS signal phases, GIS network configuration, Google transit network data for the buses routes and schedules, as well as freight (background) traffic data.

Preprocessing data: Count

  • Data is available from multiple days in Aug./2013. After preprocessing with PreProSenData.R for each week days, SensorMapping_for2013_For Aggregation (5minAndLaneLv).R will aggregate the data from 00:00 to 24:00 with the capacity constraints (HighVal / LowVal in the code). The data is located in data/SensorMapping/RealData_sec_LaneLv5min_Var.RData:

<Time>, <SensorID>,<SegID>,<LaneID>,<Count>,<Variance>,<Observation>

Sensor mapping: Travel-time

  • Firstly, TravelTimeDataPreprocess.R maps the longitude and latitude to node ID, then generate the raw travel-time data. Then, it take average node-to-node travel-time from multiple occurrences over the data set over the observation period (Aug./2013). The data is located in data/TravelTimeData/New_RealData_TravelTime_CBD.RData:

<Time>, <OriginID>,<DestinationID>,<TravelTime>,<Variance>,<Observation>

Clone this wiki locally