Estimation of Statewide Origin-Destination Truck Flows Using Large Streams of GPS Data: An Application for the Florida Statewide Model

Transcription

1 Estimation of Statewide Origin-Destination Truck Flows Using Large Streams of GPS Data: An Application for the Florida Statewide Model Akbar Bakhshi Zanjani Graduate Research Assistant Department of Civil & Environmental Engineering University of South Florida Tel: () -; Abdul R. Pinjari (Corresponding Author) Associate Professor Department of Civil & Environmental Engineering University of South Florida, ENB 0 E. Fowler Ave., Tampa, FL 0 Tel: () -; apinjari@usf.edu Mohammadreza Kamali Graduate Research Assistant Department of Civil & Environmental Engineering University of South Florida Tel: () -; mkamali@mail.usf.edu Aayush Thakur Senior Travel Demand Forecaster Cambridge Systematics Tel: (0) -; Fax: (0) -, athakur@camsys.com Jeffrey Short Senior Research Associate American Transportation Research Institute Tel: (0) -0; jshort@trucking.org Vidya Mysore Freight Analysis and Modeling Specialist Federal Highway Administration, Resource Center Tel: (0) -; vidya.mysore@dot.gov S. Frank Tabatabaee Systems Transportation Modeler Florida Department of Transportation, Systems Planning Office Tel: (0) -; Frank.Tabatabaee@dot.state.fl.us Word count:, words + table 0 + figures 0 =, equivalent words Submission date: Aug, 0 Submission for presentation and publication consideration at the th TRB Annual Meeting. Statewide Travel Demand Forecasting Joint Subcommittee of ADA0 and ADB0

2 0 ABSTRACT This paper investigates the use of large streams of truck GPS data from the American Transportation Research Institute (ATRI) for the estimation of statewide freight truck flows in Florida. To this end, first, the raw GPS data streams comprising over million GPS records were used to derive a database of more than. million truck trips starting and/or ending in Florida. The paper sheds light on the extent to which these trips derived from the GPS data capture observed truck traffic flows in Florida. This includes insights on (a) the truck type composition, (b) the proportion of the truck traffic flows covered by the data, and (c) geographical differences in the coverage. The paper applies origin-destination matrix estimation (ODME) methodology to use the GPS data in combination with observed truck traffic volumes at different locations within and outside Florida to derive an origin-destination (OD) table of truck flows within, into, and out of the state. The procedures, implementation details, and experiences discussed in the paper are expected to be useful to a number of transportation planning agencies who are considering the use of GPS data for freight travel demand modeling.

3 INTRODUCTION Accelerated growth in the volume of freight shipped on American highways has led to a significant increase in truck traffic, influencing traffic operations, safety, and the condition of highway infrastructure. Traffic congestion in turn has impeded the speed and reliability of freight movements. As freight movement continues to grow nationwide, appropriate planning and decision making processes are necessary to mitigate these impacts. However, a main challenge in establishing these processes is the lack of adequate data on freight movements such as detailed origin-destination (OD) demand data. As traditional data sources on freight movement are either inadequate or no longer available, new data sources must be investigated. A recently available source of data on nationwide freight flows is based on a joint venture by the American Transportation Research Institute (ATRI) and the Federal Highway Administration to develop and test a national system for monitoring freight performance measures on freight-significant corridors in the nation (). This data, obtained from trucking companies who use GPS technologies to remotely monitor their trucks, provides unprecedented amount of data on freight truck movements in North America. Such truck GPS data potentially can be used to support planning, operation, and management processes associated with freight movements. ATRI s truck GPS data have been used for a variety of freight performance measurement and planning applications in the U.S., including the measurement of truck speeds on major freight corridors in the nation, truck speed reliability measurements, identification of truck flow bottlenecks, and analysis of truck parking issues. In addition, the data provides an opportunity to develop OD demand data for large geographical regions such as truck flows between urban areas, megaregional flows, statewide truck flows, and even nationwide truck flows. Since a majority of freight being shipped across the U.S. is via the truck mode, OD data and analysis of truck flows across multiple jurisdictions at a large geographical scope can significantly improve freight planning at all levels of the government. It is important to note, however, that while ATRI s truck GPS data comes from a large sample of trucks in North America, it is not necessarily the census of all trucks from any region. Before applying the data to estimate OD flows for any region, it is important to understand the extent to which the data covers truck flows in the region and the nature of trucks in the data (e.g., the truck type composition and the types of businesses served). As such, additional information and procedures must be employed to weight the sample of OD trip flows derived for a study area from the ATRI data to represent the population of heavy truck flows within, to, and from the study area. The weighting process is required not only for inflating the sample to the population but also for ensuring that the spatial distribution of the resulting truck flows is representative of the truck flows in the study area. One approach to do this is Origin-Destination Matrix Estimation (ODME), which involves combining the sample OD trip flows derived from the ATRI data with other sources of information on truck flows observed at various links of the highway network in the study area to estimate a full OD flow matrix representing the population of truck flows in the study area. This paper demonstrates the use of ATRI s truck GPS data in combination with other observed data on truck traffic flows to estimate a statewide OD table of truck flows within, into, and out of Florida. In doing so, the paper sheds light on the extent to which ATRI data captures the observed truck traffic flows in Florida. This includes insights on (a) the types of trucks (e.g., heavy trucks and medium trucks) present in the data, (b) the geographical coverage of the data in Florida, and (c) the proportion of the truck traffic flows in the state covered by the data. Similar

4 procedures can be used to evaluate ATRI data in terms of its coverage of truck flows in other states as well as the entire nation. Further, the paper applies an ODME methodology to use ATRI s GPS data for estimation of OD truck flows over large geographical regions (and describes the practical aspects in doing so). The data is being considered for use in regional and statewide freight travel demand models by a number of transportation planning agencies in the U.S. and Canada. To aid these agencies, the paper documents lessons learned from using the data for estimating statewide truck OD flows. Finally, since the truck flow OD tables estimated in this research are primarily for validating and calibrating the freight component of the Florida statewide travel demand model (FLSWM), this paper sheds light on using ATRI s GPS data for statewide freight travel demand modeling purposes. Section briefly describes ATRI s truck GPS data used in this study and the procedure used to convert it into truck trips and a corresponding OD table. Section provides an assessment of the data in terms of its truck type composition, its coverage of truck flows in Florida, and geographical differences in the coverage. Section presents the ODME methodology, the inputs and assumptions for the ODME procedure used in this study, and the results and their validation. Section concludes the paper. DESCRIPTION OF ATRI S RAW GPS DATA AND ITS CONVERSION INTO A TRUCK TRIP OD TABLE. ATRI s GPS Data To derive truck OD flow tables for Florida, the research team worked with Florida specific raw GPS data from ATRI for four months March, April, May, and June in 00 for the state of Florida. Specifically, for each of these four months, all trucks from ATRI s database that were in Florida at any time during the month were extracted. Subsequently, all GPS records of those trucks were extracted for the entire month, as they traveled within Florida as well as in other parts of North America. This allows the examination of truck movements within Florida as well as truck flows into (and out of) Florida from (to) other locations. The number of GPS records for each month was over million, summing up to over million records for the four months. Each GPS record contained information on its spatial (latitude/longitude) and temporal (date/time) location along with a unique truck ID that did not change across all the GPS records of the truck for a certain time period varying from a day to over a month (at least two weeks for most trucks in the data). In addition to this information, a portion of the GPS data contained spot speeds (i.e., the instantaneous speeds) of the truck and the remaining portion of the database did not contain spot speeds. These two types of data were separately delivered, presumably because they come from different truck fleets with different GPS technologies. The frequency (i.e., ping rate) of the GPS data streams varied considerably, ranging from a few seconds to over an hour of interval between consecutive records. Information about individual trucks such as the commodity, weight or volume carried or the type of truck was not available. Since the data was collected originally for measuring truck travel speeds on freight-significant corridors, the data comprises predominantly tractorsemitrailer combinations or larger trucks (or heavy trucks) that tend to travel on such corridors. Such trucks can be categorized as class to class of FHWA s vehicle classification scheme.

5 Conversion of GPS Data into a Truck Trip OD Table The raw GPS data must be converted into a truck trip format before utilizing it form most transportation modeling and planning uses, including OD table estimation. The algorithm for converting the GPS data into truck trips is briefly described here. Pinjari et al. () provide full details on the procedure and the rationale behind it. () Sort GPS data for each truck ID into time series, in the order of date & time of GPS records. () Identify potential trip-ends (origins/destinations) based on travel speed between consecutive records (calculated using spatial movement and time gap between consecutive records). a. If travel speed between consecutive GPS records was less than mph, the truck was assumed to be at rest (i.e., at a stop). mph speed cut-off was verified using data with spot speeds and based on the literature (-). b. Not all truck stops are valid trip-ends. A truck stop was considered to be a trip-end if the stop dwell-time (i.e., stop duration) was greater than a minimum dwell-time buffer value. Stops of smaller dwell-time than dwell-time buffer were considered to be insignificant stops, including traffic signal stops, congestion stops, fueling stops, etc. At the beginning of the algorithm, a 0-minute minimum dwell-time buffer was used to identify truck tripends. Stops of less than 0-minute duration were considered to be intermediate stops not intended for pickup/delivery. c. Combine very small trips (< mile trip length) with preceding trips or eliminate them, because most such small movements were within in large establishments. d. Eliminate poor quality trips based on data quality issues such as consecutive GPS records with large time gaps or with unrealistically high travel speeds. () Eliminate trip-ends in rest areas and other locations that are unlikely be pickup/delivery stops a. by overlaying trip ends on a geographic file of rest areas, wayside parking stops, and similar locations, b. by eliminating stops within close proximity (00 feet) of interstate highways, most of which are most likely to be rest areas or wayside parking stops, and c. by joining consecutive trips ending and beginning at such stops. () Find circular (i.e., circuitous) trips and break each of them into multiple valid trips. a. Trips with a ratio between air-distance to network-distance less than 0. were considered circular trips, with a high likelihood of valid a intermediate trip-end that was missed due to counting only stops with at least 0-min dwell-time as valid trip-ends. b. Use raw GPS data between the origin and destination of circular trips to split them into appropriate number of shorter, non-circular trips by allowing smaller dwell-time buffers at the destinations. For this, implement step with a smaller dwell-time buffer ( minutes) and go through steps and to find any remaining circular trips. Repeat the process with a dwell-time buffer of minutes to split remaining circular trips. () Conduct additional quality checks and eliminate trips that do not satisfy quality criteria. Using the above-described procedure, a total of over. million truck trips were derived using ATRI s Florida-specific raw GPS data of over million records from four months in 00. Over. million of these trips were either within Florida or had one end in Florida. The

6 trip end locations of the trips derived from the above procedure were overlaid on a traffic analysis zone (TAZ) layer of the FLSWM to aggregate the trips into TAZ-to-TAZ OD flows. ASSESSMENT OF ATRI S TRUCK GPS DATA AND ITS COVERAGE OF TRUCK TRAFFIC IN FLORIDA. Truck Type Composition in ATRI Data The major sources of ATRI s data are large trucking fleets, which typically comprise tractorsemitrailer combinations that predominantly serve the purpose of long distance freight hauling. However, a close observation of the data, through following several trucks on Google Earth and examining travel characteristics of individual trucks, suggested that a small proportion of trucks in the data were more likely to be medium trucks (e.g., single unit trucks and straight/box trucks) that predominantly serve the purpose of local delivery and distribution in urban areas. Although small in number, these trucks were observed to make a large number of short trips, most likely for local delivery and distribution, which are not of primary interest for FLSWM. Since the data does not provide information on the vehicle classification of each individual truck, heuristics were developed to classify the trucks into heavy trucks and medium trucks utilizing the travel characteristics of individual trucks over extended time periods (i.e., at least two weeks). The. million truck trips derived from four months of ATRI s GPS data corresponded to, unique truck IDs. From these trucks, those that did not make at least one trip of 00 miles in a two week period and trucks that made more than trips per day were assumed to be medium trucks that are used predominantly for local delivery and distribution and removed from further consideration. These comprise.% of the trucks ( trucks) in the database that made.% of all trips extracted from the database. After removing these trucks, over. million trips extracted from GPS data of over, unique truck IDs were considered as trips made by heavy trucks that predominantly carry freight. These trips were further used for OD matrix estimation. Note: It is not necessary that only heavy trucks carry freight over long distances while only medium trucks serve the purpose of local delivery and distribution. Further research is needed to identify the composition of trucking fleet in the ATRI data and the purposes served by those trucks.. What Proportion of Heavy Truck Traffic Flows in Florida Is Captured in ATRI Data? To address this question, truck traffic flows in one week of ATRI s truck GPS data was compared with observed truck traffic volumes from Telemetered Traffic Monitoring (TTM) sites in Florida for that week. This section describes the procedure and results from this analysis. One week of ATRI data (May -, 00) was used to derive weekly ATRI truck traffic volumes at the TTM sites. Generating data on weekly ATRI truck traffic volumes at each TTM location required counting the number of times the trucks in ATRI data crossed the location in the week. To do so, the truck trips generated from the procedure discussed earlier were isolated for the week of May -, 00. For each of these trips, given the origin and destination, a sample of en-route GPS records between the trip-ends (sampled at a -minute interval) were map-matched to the FLSWM highway network using the network analyst tool in ArcGIS. The map-matching algorithm snaps the GPS points to the nearest roadway links and also determines the shortest path between consecutive GPS points. Since intermediate GPS points between the trip-ends were sampled at only a -minute interval, this procedure results in a sufficiently

7 0 0 0 accurate route for the trip. The output from this process was an ArcGIS layer containing the travel routes for all trips generated from ATRI s one-week truck GPS data. This ArcGIS layer was intersected with another layer of FLSWM network containing the TTM stations. This helped estimate the number of ATRI truck trips crossing each TTM station (i.e., the volume of ATRI trucks crossing TTM stations). The truck traffic volumes derived from ATRI data were compared with observed volumes of heavy trucks (of class to ) extracted from FDOT s TTM data for the same week. Figure shows the average daily heavy truck volumes at over 00 TTM locations in Florida (top map in the figure). Only 0 of these locations had traffic count data for all days in the specific week under consideration. Therefore only these locations were selected for comparing the ATRI truck traffic volumes with observed truck traffic volumes. The bottom map in Figure shows these results of this comparison for TTM locations with observed heavy truck traffic volumes greater than,000 per day (location with smaller volumes are not shown for clarify in presentation). Clearly, at no single location does the ATRI data provide 00% coverage of the observed heavy truck volume. However, the data does provide some coverage of the heavy truck traffic at all locations. At most of these locations, at least % of the observed heavy truck volumes are captured in ATRI data. Table shows these results aggregated by highway facility type. Note from the third column that a bulk of heavy truck traffic counts (.%) are observed on freeways and expressways that represent only.% of the 0 TTM sites considered in this analysis. The last row in the fourth column shows that a total of, ATRI truck crossings were counted at the 0 TTM locations. Note from the same column that the distribution of these ATRI truck traffic counts across different facility types is similar to the distribution of observed truck counts across facility types in the third column. This result suggests that the ATRI data provides a representative coverage of heavy truck flows through different facility types in the state. The last column expresses truck traffic counts from ATRI data as a percentage of observed heavy truck traffic counts at the TTM locations. Overall, it can be concluded that the truck trip OD table derived from ATRI data (in 00) provides 0% coverage of heavy truck flows observed in Florida. This result is useful in many ways. For example, the OD table derived from ATRI data can be weighted (0-fold) to create a seed matrix for use as an input into the ODME process.

8 (a) (b) FIGURE (a) Observed heavy truck traffic flows at different TTM sites in Florida; (b) Percentage of observed heavy truck (classes -) volumes represented by ATRI data at TTM sites in Florida during May, 00.

9 TABLE Coverage of Heavy Truck Traffic Volumes in Florida in ATRI Data (for One Week from May to, 00) Facility Type No. of TTM Traffic Counting stations Observed Truck Traffic Volumes (Class -) during May -, 00 Truck Traffic Volumes in ATRI data during May -, 00 % Coverage Freeways & Expressways (.%),0, (.%),0 (.%) 0.% Divided Arterials (0.0%), (0.%) 0, (.%).% Undivided Arterials (.%) 0,0 (.%), (.%).% Collectors (.0%), (.%), (.%).% Toll Facilities (.%) 0, (.0%), (.%).% Total 0,,, 0.%. Geographical Coverage of ATRI Data in Florida The bottom map in Figure sheds light on the geographical differences in the extent to which ATRI data captures observed heavy truck volumes in the state. Specifically, it can be observed that the coverage in the southern part of Florida (within Miami) and the southern stretch of I- (i.e., in and below Tampa) is relatively lower compared to the coverage in the northern and central Florida regions. To further assess the statewide geographical coverage of ATRI data in detail, the OD table derived from ATRI data was aggregated into trip productions (i.e., # trips beginning from) and trip attractions (i.e., # trips ending at) for TAZs in the FLSWM. The trip attractions and productions were then plotted on a GIS layer of FLSWM TAZs within Florida. These maps are not presented in the paper to conserve space (see () for the maps), but the findings are briefly discussed here. It was observed that the Everglades region in the south Florida and some TAZs in northwest Florida had zero trip productions and/or attractions in the OD table derived from ATRI data. This could be due to two reasons: () low penetration of ATRI data in those TAZs, or () those TAZs did not have heavy truck trip generation in reality. It is reasonable to expect the Everglades region in Florida to have little to no truck trip generation. To investigate zero truck trip generation among the northwestern TAZs in the state, we examined Figure (map in the top) for observed heavy truck flows in the TTM data. Except along the I-0 corridor, the northwest region of the state does not have high truck traffic volumes. This suggests that the zero trip generations in ATRI data for several TAZs in northwest part of Florida is a reasonable representation of truck flows in that region. The TAZ-level trip productions and attractions were further aggregated to a county-level to identify any potential spatial biases in ATRI data. It was observed that, Duval (Jacksonville), Polk, Orange (Orlando), Miami-Dade (Miami), and Hillsborough (Tampa) counties, in that order, had the highest truck trip generation. It is reasonable that counties within major metropolitan areas in the state have the highest heavy truck trip generation. Whereas Polk County is expected to have a high truck trip generation due to the presence of several freight distribution centers, it is interesting that the county had higher truck trip generation than that in

10 Tampa, Orlando and Miami regions. Further, the truck trip generation in Southeast Florida (Miami, Broward and Palm Beach counties) was smaller than that in Polk County. Recall from the first paragraph in this section that the ATRI data coverage of heavy truck flows in the southern part of Florida was relatively lower compared to other locations in Florida. These trends are likely to be a manifestation of spatial biases in the data. To address such spatial biases, ODME process combines the truck trip flows derived from the ATRI data with observed heavy truck traffic volumes at different locations in the state. ORIGIN DESTINATION MATRIX ESTIMATION OF STATEWIDE TRUCK FLOWS ODME is a procedure used to update an existing matrix of OD flows using information on traffic volumes observed at various locations in the transportation network (). The method has been used widely for passenger travel demand estimation and to a relatively small extent for freight demand estimation (). However, estimation of reliable OD matrices from traffic count data is a challenging exercise, since the observed OD data (from typical approaches such as establishment surveys or roadside interviews) is often limited and the ODME procedures can lead to multiple non-unique OD matrices that may provide equally good fit to observed traffic volumes. ATRI s GPS data provides unprecedented amounts of observed truck OD flow data that offers an opportunity to estimate truck OD flow matrices in a potentially more reliable manner. ODME can be used to factor the seed matrix derived from ATRI data in such a way that the resulting estimated OD matrix, when assigned to the highway network, closely matches with observed heavy truck counts at various locations on the network. Recent attempts at estimating truck OD flows using this data include studies by Bernardin et al. () and Bernardin and Short ().. The ODME Methodology The specific ODME procedure used in this research, embedded in the Cube Analyst Drive software, is an optimization problem that tries to minimize a function of the difference between observed traffic counts and estimated traffic counts (from the estimated OD matrix) and the difference between the seed matrix and the estimated OD matrix, as below: arg min J X F AX b G X X0 X () subject to X 0 and Xlower X Xupper In this optimization problem, X is the OD matrix to be estimated, X 0 is the seed OD matrix, G is a function measuring the distance between the estimated matrix and the seed matrix, b is a vector of observed counts at different locations in the study area, A is the route choice probability matrix obtained from assignment of OD flows in X on the network using user equilibrium method, AX is a vector of estimated traffic counts, and F is a function measuring the difference between estimated and observed traffic counts. The procedure attempts to arrive at an OD flow matrix X in such a way that the resulting traffic volumes (AX) match closely with observed traffic flows (b ). At the sametime, the procedure avoids overfitting to observed traffic flows by including the term G X X 0 so that the estimated matrix has a similar structure as the seed matrix. X lower and X upper are boundaries (lower and upper bounds) within which the estimated matrix should fall. The analyst can use these boundary constratins to set lower and upper bounds on the estimated matrix, relative to the seed matrix. The estimated OD matrix may be evaluated by comparing of estimated heavy truck traffic volumes and observed heavy truck traffic volumes at different locations within and

11 outside Florida (for a set of validation data that was not used for ODME). Specifically, a root mean square error (RMSE) can be evaluated as: i i RMSE C avg N N V C where, V i is the estimated truck volume on link i, C i is the observed truck volume on link i, C avg is the average heavy truck traffic counts of the entire set of observations, and N is the total number of truck counting locations. In addition to comparing observed and estimated traffic volumes, it is important to assess the reasonableness of the estimated OD matrix in different ways. Aggregating the OD matrix to a coarser spatial resolution and examining the spatial distribution of flows, examining the trip productions and trip attractions for each aggregate spatial zone, and examining the trip length distribution of the estimated OD matrix in comparison to the seed OD matrix are different ways of assessing the estimated OD matrix.. Inputs for ODME The primary inputs to ODME procedure are the seed OD matrix (derived from ATRI data), a highway network for the study area along with information on the travel times and capacity of each link in the network (extracted from FLWSM), and observed truck traffic volumes on different links in the network. In addition, OD flow matrices corresponding to travel other than freight truck flows non-freight truck travel and passenger travel (both extracted from FLSWM) were provided as inputs to generate realistic travel conditions in the network. Seed matrix: This is a matrix of FLSWM TAZ-to-TAZ truck trip flows derived from ATRI s truck GPS data. Specifically, the OD matrix of heavy truck flows obtained from months ( days) of ATRI data was divided by to obtain the seed matrix for an average day and then multiplied by 0 to account for the fact that the data represents 0% of observed truck traffic flows in the state. In FLSWM, Florida, other states in the U.S., and Canada are divided into, TAZs, with,0 of these zones in Florida. Therefore, the seed matrix has a total of million OD pairs. The. million trips extracted from four months of ATRI data for this OD matrix were between only 0. million of the million OD pairs. The remaining. million OD pairs in the seed matrix were zero-cells (i.e., they had no trips). This is an important issue to address because most ODME methods used in practice result in zero trips for OD pairs that began with zero-cells in the seed matrix. A common approach to address this issue is to introduce a small positive number (say, 0.0) for zero-cells in the seed matrix that the analyst believes should have trip flows. To assess which zero-cells in the seed matrix were expected to have trip flows, the structure of the seed matrix was examined by aggregating it into county-level in Florida and state-level outside Florida. Out of a total of (,) county-to-county OD pairs in Florida, the seed OD matrix derived from ATRI data had trips for.% (,) OD pairs. The remaining 0.% OD pairs did not have trips. A closer examination suggested that some rural counties in northwest Florida and few rural counties in the southwest (such as in Everglades) have higher occurrence of zero trip flows to/from other counties in Florida and other states outside Florida. Combining this information with earlier discussion (section.) on geographical coverage of ATRI data, it can be concluded that the zero truck flows in the seed matrix to/from counties in northwestern and southwestern parts of Florida is likely because these counties may not actually have truck flows i ()

12 to/from a large number of locations. Further, considering that the seed matrix was derived from four months of GPS data (which is a large amount of data), if some OD pairs at a county-level resolution did not have any trip exchanges, it was considered reasonable to assume that those OD pairs may not have truck flows in reality. On the other hand, for OD pairs with both ends outside Florida, 0 out of the 00 (0 0) state-to-state OD pairs did not have any trips in the OD table. Since the data is Florida centric, it is likely that the seed OD matrix is not necessarily a good representation of truck flows between OD pairs outside Florida. Observed truck traffic volumes: This data was gathered for several locations within and outside Florida for the same four-month duration for which the seed matrix was available (and used in the form of average daily truck traffic). Since the OD matrix to be estimated includes truck flows between Florida and other states as well, it was considered important to include truck traffic counts outside Florida as well. Data on truck traffic counts in Florida was obtained from Florida Department of Transportation (FDOT) s TTM sites at over 00 locations on Florida s highway network (counts were obtained separately for each direction; a total of traffic counts). Of all the different heavy truck counts at different locations in Florida, data from locations (i.e., input stations) were used the ODME while data from the remaining locations (i.e., validation stations) were kept aside for validation. For Georgia, truck traffic counts from Georgia Automated Traffic Recorder (ATR) locations were obtained from Georgia Department of Transportation (GDOT). For all other states, FHWA s vehicle travel information system (VTRIS) database was utilized to obtain truck traffic counts on highway network locations. Since the FLSWM network is not very detailed outside Florida, only of the VTRIS and ATR locations fell on FLSWM highway network links. Therefore, except Florida and Georgia, other states in the southeast such as Alabama, Mississippi, Louisiana, and South Carolina had very few locations from which observed truck traffic count data was used in ODME. Tennessee, Kentucky, and North Carolina did not have any traffic counting locations; this will likely have a bearing on the results. Out of heavy truck counts at different locations outside Florida, data from locations were used in ODME while data from the remaining locations were kept aside for validation.. Evaluation of Different Assumptions for ODME The ODME procedure was run several times to evaluate different assumptions on the OD matrix, including assumptions of upper/lower bounds on the number of trips estimated between OD pairs (i.e., X lower and X ) and assumptions on zero-cells in the seed matrix. upper Among the assumptions on upper/lower bounds on the OD matrix the extent of upper bound did not influence the results as long as the bound was large enough. Imposing small upper bounds was leading to poor fit of the estimated OD matrix (estimated traffic volumes, to be precise) to observed traffic counts. Therefore, no upper bound was imposed on the OD matrix. The lower bounds, however, had considerable influence on the estimated matrix. When lower bounds were removed, the estimated truck traffic volumes matched very well with observed truck traffic volumes for locations from which traffic count data was used for ODME. Specifically, the RMSE value between estimated and observed truck traffic volumes was very small (<0%). However, in this scenario, trip length distribution of the estimated OD matrix was skewed toward a much greater share of shorter trips than those in the seed matrix. There are two possible reasons for this: () the seed matrix was biased toward long-distance trips and that combining the seed matrix with the observed traffic counts reduced the bias by increasing the proportion of short trips, or () the estimated OD matrix is over-fitting to observed traffic counts.

13 When we closely examined the estimated OD matrix, no trips were estimated between Florida and some southeastern states that had no observed traffic counts from the VTRIS data. For example, no OD pair between North Carolina and Florida and between Tennessee and Florida had any trips in the estimated OD matrix (recall that we could not use any observed truck traffic counts from Tennessee and North Carolina), whereas the seed matrix did have trip flows between Florida and those states. This suggests that the estimated OD matrix is likely an artifact of over-fitting to the observed traffic volumes. This was reflected in a deteriorated fit of the estimated truck traffic volumes to observed truck traffic volumes from the locations kept aside for validation. When the lower bound was set to be equal to the seed matrix, the estimated OD matrix was very close in its trip length distribution to the seed matrix. However, the RMSE between the estimated and observed traffic volumes was high (0%). This was because the seed matrix was not being modified meaningfully by the ODME procedure. As a middle ground between the above two scenarios, we explored the different lower bounds ranging from 0. to 0. times the seed matrix. Of all these, lower bounds set to 0. times the seed matrix provided the most reasonable results. Note that the seed matrix derived from the ATRI-data was inflated 0-fold to recognize that ATRI data represented 0% of the observed heavy truck flows in the state (at an aggregate level). However, it is not necessary that the data represents 0% of heavy truck flows at every location. In some locations, the data might represent more or less than 0%. Therefore, setting a lower bound of 0. allows for the possibility that the actual heavy truck trip flows might be less than the 0-fold inflated number of heavy truck derived from ATRI data. This scenario provided reasonable results, with RMSE value of 0% for input stations and % validation stations while also allowing trips from (and to) all states to (and from) Florida. Among the assumptions on zero-cells, keeping the zero-cells as is provided better results both in terms of validation measures against observed heavy truck counts as well as reasonableness of the spatial distribution of truck flows. For instance, altering all zero-cells to 0.0 provided high RMSE values results unless the lower bounds were removed on all cells. However, removing the lower bounds on all cells, as discussed earlier, was leading to overfitting of the estimated heavy truck traffic volumes to observed truck traffic volumes.. ODME Results for One Set of Assumptions This section presents results from the following set of assumptions in ODME: () no upper bounds but a lower bound of 0. times the seed matrix on the estimated OD matrix, and () zerocells in seed matrix assumed to truly represent zero truck flows. The seed matrix had trips between nearly 0. Million OD pairs, of which 0. Million OD pairs had both ends in Florida. The same OD pairs have trips in the estimated OD matrix (due to assumption on zero-cells). The seed matrix contained a total of,0 daily heavy truck trips that started and/or ended in Florida while the estimated OD matrix resulted in a total of 0, trips. The daily mileage of estimated trips with at least one end in Florida was over Million miles..% of these miles (i.e., over million miles) were due to trips within Florida. Figure shows a comparison of estimated truck traffic volumes (from user equilibrium assignment of the estimated OD matrix) and observed heavy truck traffic volumes in Florida s TTM data. Blue dots in the figure are for locations for which TTM data was used in the ODME process, while the red dots are for locations for which TTM data was kept aside for validation. All dots on the degree line indicate perfect fit between estimated and observed truck volumes, while the dots that fall between the two dotted lines correspond to locations with less than %

14 0 0 difference. A table embedded in the figure with RMSE values for different ranges of daily observed truck volumes shows that the estimated truck volumes are matching reasonably well with the observed volumes, especially at locations with daily truck volumes higher than,000 trucks. Figure shows the trip length distributions of the trips in the seed and estimated OD matrices, for trips with at least one end in Florida. It can be observed that the distribution of the trips in the estimated OD matrix is closely following those from the seed matrix derived from the ATRI data, albeit the estimated OD matrix has a slightly greater proportion of shorter trips. County-level trip productions and attractions for both the seed and estimated OD matrices were also examined (see Figure for county-level trip productions in seed and estimated matrices). As discussed earlier, the seed matrix shows lower than expected trip generation in the south Florida region (especially in and around Miami) and the southern stretch of I- beginning from the Tampa region (when compared to those in the Polk County). The estimated OD matrix, due to its use of additional information (observed heavy truck traffic volumes in the state), addresses this issue to a certain extent. Counties in the southeast Florida and Hillsborough County have higher trip generation in the estimated matrix than in the seed matrix. The seed matrix had around % of the trips staring or ending in Florida staying within Florida, while the estimated matrix adjusts this distribution to %. At the county level, the seed matrix shows Polk County as one of the major origins/destinations for trips from/to other counties. The estimated matrix makes adjustments to this trend for Miami-Dade, Palm Beach, and Broward Counties. Specifically, the estimated matrix shows greater flows between these three counties. Also, the estimated OD matrix shows smaller proportion of flows between Hillsborough and Miami-Dade Counties than that in the seed OD matrix. While one would expect greater amount of flows between these two counties, the observed truck traffic volumes on major highways between these two counties are not high enough to support this notion.

15 000 All Input Stations Validation Stations 000 Linear ( Degree Line) Linear (% Error Line) Estimated heavy truck volumes per day Observed truck counts per day RMSE for input stations RMSE for validation stations % ( locations) 0% ( locations) % ( locations) % ( locations) % ( locations) % ( locations) % ( locations) % ( locations) Total 0% ( locations) % ( locations) Observed heavy truck volumes per day FIGURE Observed vs. estimated heavy truck counts per day at different locations in Florida.

16 Percentage of trips with at least one end in Florida 0 ATRI Data (Seed Trips) Estimated Trips < Trip Length/Distance between TAZs (Miles) FIGURE Trip length distributions of truck trips in seed and estimated OD matrices.

17 FIGURE County-level trip productions in seed and estimated OD matrices.

18 CONCLUSIONS AND FUTURE RESEARCH This paper presents an investigation of the use of large streams of truck GPS data (available from ATRI) in combination with other observed data on truck traffic flows to estimate a statewide OD table of truck flows within, into, and out of Florida. In doing so, the paper sheds light on the extent to which ATRI data captures the observed truck traffic flows in Florida. This includes insights on (a) the types of trucks (e.g., heavy trucks and medium trucks) in the data, (b) the proportion of the truck traffic flows in the state captured in the data, and (c) the geographical differences in the coverage. Similar procedures can be used to evaluate ATRI data in terms of its coverage of truck flows in other states as well as the entire U.S.. Further, the paper applies an ODME methodology to use ATRI s GPS data for estimation of OD truck flows over large geographical regions. The data is being considered for use in regional and statewide freight travel demand modeling by a number of transportation planning agencies in the U.S. and Canada. To aid these agencies, the paper provides a description of the procedures implemented and a detailed discussion of implementation issues (e.g., evaluation of different assumptions) on using the data for estimating statewide truck OD flows. Over million records of ATRI s truck GPS data during four months April-June, 00 were used for this research. The raw GPS data streams were converted into a database of over. million truck trips. It is known that ATRI data predominantly comprises tractorsemitrailer combinations (or heavy trucks). However, a close examination of the data (via tracing the land-uses served by some trucks in Google Earth and examining the truck-level travel characteristics) in this research suggests that the data has a small proportion of trucks that are likely to be single-unit trucks or straight trucks that do not necessarily haul freight over long distances. The paper devised simple rules to divide the data into two categories: () long-haul trucks or heavy trucks, and () short-haul trucks or medium trucks. ATRI s truck GPS data represents a large sample of truck flows within, coming into, and going out of Florida. However, the sample is not a census of all trucks traveling in the state. To evaluate the coverage of ATRI data in Florida, truck traffic flows implied by one-week of ATRI s truck GPS data were compared with truck counts data at over 0 locations in the state. At an aggregate level, the 00 ATRI data was found to provide 0% coverage of heavy truck flows observed in Florida. Further, geographical differences the coverage were examined. it is worth noting here that the ATRI data has grown significantly between 00 and the present date, therefore it can be assumed that current coverage in Florida is greater than 0%. The OD tables derived from the ATRI data were combined with observed truck traffic volumes at different locations within and outside the state to derive an OD table that is representative of the freight truck flows within, into, and out of the state. The ODME method was employed to achieve this. A variety of different assumptions were evaluated before arriving at a set of defensible assumptions for deriving the OD tables in this research. The resulting OD tables provided acceptable validation results when the estimated truck traffic volumes were compared with observed truck traffic volumes. In addition, the estimated OD matrix was subjected to a variety of reasonableness checks. Therefore, the OD tables derived in this research can be used for statewide freight modeling in many ways, including the validation and calibration of highway freight modeling components in FLSWM. The ODME procedure in this study can be improved in different ways. First, utilizing more robust data on observed truck traffic volumes in several southeastern states potentially can help improve the ODME results. For example, there were little to no traffic count information for states such as Tennessee and North Carolina. Filling such data gaps can potentially help in better

19 0 estimating the truck flows into and out of the state. Second, the ODME procedure itself can be improved in different ways: (a) by allowing different constraints (lower/upper bounds) that are specific to different OD pairs (the constraints in this study were uniform to all OD pairs due to software limitations), (b) by exploring the different weighting schemes used to expand the seed matrix, and (c) by improving the traffic assignment procedure based on observed route choice patterns of trucks using GPS data. ACKNOWLEDGEMENTS This research was funded by FDOT. The opinions, findings, and conclusions expressed in this publication are those of the authors and not necessarily those of FDOT or USDOT. Thanks to Vince Bernardin and Arun Kuppam for sharing their experiences working with ATRI data. Vipul Modi provided valuable assistance with the use of Cube software. Ramachandran Balakrishna provided helpful insights on the theoretical and practical aspects of ODME.

20 0 REFERENCES. Jones, C., D. C. Murray, and J. Short. Methods of Travel Time Measurement in Freight Significant Corridors. Presented at th Annual Meeting of the Transportation Research Board, Washington, D.C., 00.. Pinjari, A. R., A. Bakhshi Zanjani., A. Thakur, A. Nur Irmania., M. Kamali, J. Short., D. Pierce, and L. Park. Using Truck Fleet Data in Combination with Other Data Sources for Freight Modeling and Planning. Research Report, prepared for Florida Department of Transportation, Tallahassee, FL, 0.. Bernardin, V. L., J. Avner, J. Short, L. Brown, R. Nunnally, and S. Smith. Using Large Sample GPS Data to Develop an Improved Truck Trip Table for the Indiana Statewide Model. Presented at th Transportation Research Board Conference on Innovations in Travel Modeling, Tampa, FL, 0.. Kuppam, A., J. Lemp, D. Beagan, V. Livshits, L. Vallabhaneni, and S. Nippani. Development of a Tour-Based Truck Travel Demand Model Using Truck GPS Data. Presented at rd Annual Meeting of the Transportation Research Board, Washington, D.C., 0.. Ma, X., E. D. McCormack, and Y. Wang. Processing Commercial Global Positioning System Data to Develop a Web-Based Truck Performance Measures Program. In Transportation Research Record: Journal of the Transportation Research Board, No., Transportation Research Board of the National Academies, Washington, D.C., 0, pp Van Zuylen, H. J., and L. G. Willumsen. The Most Likely Trip Matrix Estimated from Traffic Counts. Transportation Research Part B: Methodological, Vol., No., 0, pp. -.. González-Calderón, C., J. Holguín-Veras, and J. Ban. Tour-Based Freight Origin-Destination Synthesis. Presented at th European Transportation Conference, Association for European Transport, Glasgow, Scotland, 0.. Bernardin, V.L., and J. Short. Expanding Truck GPS-Based Passive Origin-Destination Data in Iowa and Tennessee. Presented at th Transportation Research Board Conference on Innovations in Travel Modeling, Baltimore, MD, 0.