Implementing the Freight Transportation Data Architecture: Data Element Dictionary (2015)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

43 6.1 Introduction Differences in data element definitions hinder the process of (1) combining elements from individual sources into a single dataset; (2) combining elements within individual data sources from different time periods for analysis; and (3) inferring statistics from joined data elements. Identifying differences in data element definitions is critical to performing freight transporta- tion analysis. This chapter presents the study’s findings on the inherent differences among data sources for commonly utilized freight data elements such as origin and destination, commodity, mode of transport, industry, imports and exports, safety, and units of measure. Differences in element definitions were assessed based on a variety of characteristics, including the type of data, level of measurement, attribute definitions, and the spatial and temporal characteristics of each element. The discussions are limited to publicly available data sources. Privately held freight data sources have been excluded because of confidentiality concerns and the unavailability of certain data sources. 6.2 Methodology The RBCS (role-based classification schema) was first used in identifying similar or related data elements within and across multiple data sources. Several sources from the literature also were referenced. Particularly helpful were user guides, data dictionaries, and the metadata asso- ciated with each data source. These documents provided detailed attribute descriptions and caveats for using the data, allowing the research team to compare similar elements across sources to determine whether relevant differences existed. Differences in element definitions were categorized into three main groups: (1) taxonomic differences, (2) temporal differences, and (3) methodological or analytical differences. Complex topics within each of the three main categories were broken down further into “sub-differences.” For example, under “Differences in Origin and Destination Data Elements” (Section 6.3 of this chapter), items categorized as having temporal differences were placed into one of three catego- ries: (a) infrequent data collection, (b) changes in methodology over time, or (c) data elements accounting for temporal differences. This extra level of classification enables data users to clearly identify the interactions and interrelationships of the data elements for each topic. 6.2.1 Taxonomic Differences Data element definitions may vary by taxonomy in terms of how the individual elements are classified. These differences can be as basic as the definition of a truck (e.g., by size, weight, axles, Differences in Data Element Definitions C H A P T E R 6

44 Implementing the Freight Transportation Data Architecture: Data Element Dictionary and so forth) or a traffic volume count (e.g., AADT [annual average daily traffic] or AAWDT [average annual weekday traffic], seasonal factors), or they may be slightly more complex, such as geographic differences for which location references have been reported differently (e.g., by point or polygon). Differences in geographic scale also can exist with regard to data from various sources, and data may be statistically valid only at certain levels of geographic detail (e.g., state, county, district). 6.2.2 Temporal Differences Data elements across data sources or within a single data source may include inherent internal inconsistencies that make data comparisons difficult. The data may have been collected in dif- ferent reporting years, or the source may have changed the definition of the data element over time. For example, changes may occur in the way industries are classified within a particular industry classification system. If an entity such as Home Depot is reclassified from a wholesale to a retail establishment within the structure of the North American Industry Classification System (NAICS)/Standard Industrial Classification (SIC), failing to account for the change can affect an analysis. 6.2.3 Methodological or Analytic Differences Often, highly compatible data sources with commonalities in a substantial number of areas will diverge in their reporting or analytical properties. For example, two data sources that both analyze commodity movements may report the results in divergent terms, such as dollar value of cargo, tons of cargo, or units of cargo (e.g., 40-ft. container equivalent units [CEUs] or 20-ft. equivalent units [TEUs]). Note: In the Freight Data Dictionary web application, in this chapter, and in Chapter 7, data element names given in all capital letters appear as they are represented in the actual data sources (e.g., ORIGINID, DESTINATIONID, and COMMODITYID). 6.3 Differences in Origin and Destination Data Elements Keywords: origin, destination, place, terminal, terminus Origin and destination are critical inputs for conducting a wide range of analyses related to freight movement. When working with origin and destination data elements, data users should be aware of taxonomic, temporal, and methodological/analytical differences among and within data sources. 6.3.1 Taxonomic Differences The taxonomic differences among and within the origin and destination data elements discussed in this report relate to differences in geographical scale and definition. 6.3.1.1 Geography Scale and Definition Origin and destination data elements often are represented at different geographic scales, making it difficult to perform certain types of analysis between data sources and to disaggregate the data within a single data source to smaller geographic scales. Some examples of differences in scale or definitions among origin and destination data sources are described in the balance of this section.

Differences in Data Element Definitions 45 Commodity Flow Survey (CFS) • CFS Areas are drawn from a subset of combined statistical areas and metropolitan statistical areas as defined by the Office of Management and Budget. When they include more than one state, however, CFS Areas are divided into their state parts.1 For example, the Kansas City– Overland Park–Kansas City, MO–KS Metropolitan Area has both a Kansas area and a Missouri area. Given that not all origin-destination data elements disaggregate metropolitan areas into their separate state areas, caution should be used when comparing CFS Area data to other data of similar geography. • Remainder of State is a unique geographical category used in the CFS to represent those areas of a state not contained within the CFS-defined metropolitan areas. Remainder of State can encompass large geographic zones and may introduce challenges for data users attempting localized analysis or bridging with other data sources, as other origin-destination data elements may not have the ability to identify the Remainder of State category. Freight Analysis Framework (FAF3) • FAF3 uses geographic definitions that may not be present within other data sources.2 – FAF REGION—FAF3 contains 123 domestic regions3 and includes data on eight foreign regions.4 Although statistical methods exist that allow analysts to disaggregate FAF data from FAF regions to counties or smaller areas, FHWA has not measured any of these methods to establish estimates of reliability or accuracy. FAF estimates of truck tonnage and number of trucks on the network, particularly in regions with multiple routes or significant local traffic between major centers of freight activity, should be supplemented with local data to support local applications. – ZONE OF ENTRY—For import shipments, this data element represents the origin of flow (the FAF REGION or state of entry). – ZONE OF EXIT—For export shipments, this data element represents the destination of the flow (the FAF REGION or state of exit). Carload Waybill Sample • Standard Point Location Code (SPLC)—SPLC is used to identify the origin and destination stations (ORIGIN SPLC and DESTINATION SPLC). Other freight databases do not use this code. • FREIGHT AREA, FREIGHT RATE AREA, FREIGHT RATE TERRITORY, and other related data elements are imputed from the SPLC. • FREIGHT STATION ACCOUNTING CODE (FSAC)—FSAC is used to identify origin and destination stations (ORIGIN FSAC and TERMINATIONS FSAC). Other databases do not use this code. • Business Economic Area (BEA) Codes—BEA codes are used to identify the reported waybill movement’s origin and termination location. Other databases included in the review do not use this code.5 • Despite revisions made in November 2004 to the U.S. Department of Commerce’s BEA codes and their regional boundaries to reflect changes in economic and population growth, as of the 2013 release, the Carload Waybill Sample has continued to use the February 1995 desig- nations. The November 2004 definitions contain 179 economic areas, and the February 1995 definitions contained 172 economic areas. • In addition to the 172 BEA codes, the Carload Waybill Sample includes 13 codes representing Puerto Rico, Mexico, and provinces in Canada. Data users are advised that the following codes are not recognized by the Department of Commerce: – 173: Newfoundland – 174: Nova Scotia

46 Implementing the Freight Transportation Data Architecture: Data Element Dictionary – 175: Prince Edward Island – 176: New Brunswick – 177: Quebec – 178: Ontario – 179: Manitoba – 180: Saskatchewan – 181: Alberta – 182: British Columbia – 183: Yukon/Northwest Territories – 184: Puerto Rico – 185: Mexico. • Princeton Transportation Network Model number—This number is used to identify the node to which the waybill movement’s origin location is assigned. The number incorporates the data elements ORIGIN NET3 NUMBER and TERMINATION NET3 NUMBER. Other data sources do not use this code. Air Carrier Statistics • ORIGIN and DESTINATION data elements in the Air Carrier Statistics signify airport codes. • _CITYMARKETID is used as a data element in the Air Carrier Statistics to identify and con- solidate airports serving the same city market. Other data sources do not use this code. • Other data elements unique to the Air Carrier Statistics database include: – _AIRPORTID, which is an identification number assigned by U.S. DOT to identify a unique airport. This field is recommended for use when performing airport analysis across a range of years, because airports can change their airport codes and airport codes can be reused. – _AIRPORTSEQID, which is an identification number assigned by U.S. DOT to identify a unique airport at a given point of time. – _WAC, which reports the world area code where an airport is located. • Data elements within the Air Carrier Statistics database that are similar to those in other data- bases include _CITYNAME, _STATE, _STATEFIPS, _STATENAME, and _COUNTRY. These data elements represent the location of the originating or destination airport. 6.3.2 Temporal Differences Temporal differences among and within origin and destination data elements fall under the following categories: • Infrequent data collection • Changes in methodology over time • Data elements accounting for temporal differences The balance of this section presents examples of temporal differences and how these can pose challenges in data analysis. Sources that collect data infrequently may make trend analysis or data interpolation difficult within a single data source. Similarly, making comparisons across data sources can be difficult if the sources use different collection periods. 6.3.2.1 Infrequent Data Collection Commodity Flow Survey (CFS) • Since 1993, the CFS has been conducted every 5 years (during years ending in 2 or 7). Tempo- ral gaps in data collection years may create difficulty when filling gaps and making compari- sons with other datasets that have more complete temporal coverage.

Differences in Data Element Definitions 47 6.3.2.2 Changes in Methodology Over Time Commodity Flow Survey (CFS) • Before the 2007 CFS, a survey was conducted to obtain information on shipping status and value of shipments for the auxiliaries. The U.S. Census Bureau concluded that the advance survey enabled more accurately assigned shipper status for both the warehouse and managing office auxiliaries on the 2007 CFS sampling frame as compared with the 2002 sampling frame; however, the accuracy of shipper status for managing offices on the frame was less than for non-auxiliaries. • The 2012 survey included the addition of 11 additional metropolitan statistical area geographies.6 Carload Waybill Sample • Despite revisions made in November 2004 to the U.S. Department of Commerce’s Business Economic Area (BEA) codes and their regional boundaries to reflect changes in economic and population growth, as of the 2013 release, the Carload Waybill Sample has continued to use the February 1995 designations. The November 2004 definitions contain 179 economic areas, and the February 1995 definitions contained 172 economic areas.7 Foreign Trade Statistics (FTS) • The FTS’s Origin of Movement identifier was added in 1985. This identifier indicates the state where the export journey began. It allows the compilation of the Origin of Movement—Based on Origin State series. Available since 1987, this series provides export statistics based on the state from which the merchandise starts its journey to the port of export; that is, the data reflects the transportation origin of exports. 6.3.2.3 Data Elements Accounting for Temporal Differences Air Carrier Statistics • ORIGINAIRPORTID and ORIGINAIRPORTSEQID are data elements that address tempo- ral differences within the data source, accounting for the fact that cities can change their air- port codes over time. ORIGINAIRPORTSEQID identifies a unique airport at a given point in time.8 6.3.3 Methodological/Reporting Differences Commodity Flow Survey (CFS) • The methodology for verifying origin changed slightly in 2012 and differs from previous reporting years. Additional information is available in the Freight Data Dictionary web appli- cation or in the CFS documentation available at http://www.rita.dot.gov/bts/help_with_data/ commodity_flow_survey.html#naics_table. Freight Analysis Framework (FAF3) • The FAF3 documentation states that the methods and data sources used have changed com- pared to those used in developing previous FAF versions, and that versions should not be compared to each other. For example, FAF version 2 (FAF2) has 114 domestic origins and destinations but FAF3 has 123 domestic regions.

48 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.4 Differences in Commodity Data Elements Keywords: commodity, HS (Harmonized System) code, STCC (Standard Transportation Commodity Codes), SCTG (Standard Classification of Transported Goods), SITC (Standard International Trade Classification), bulk, break-bulk, hazardous materials, Schedule B, HTS (Harmonized Tariff Schedule) 6.4.1 Taxonomic Differences Taxonomic differences among and within commodity data elements fall under the following categories: • Data elements that include or exclude commodities or commodity groups • Differing classification systems 6.4.1.1 Data Elements that Include or Exclude Commodities or Commodity Groups Data sources that report commodity information often include certain industries or modes of transport but exclude others, which makes it difficult to directly compare commodities across multiple data sources. The balance of this section discusses which industries or commodities are included or excluded in the definition of a commodity across multiple data sources. Carload Waybill Sample • This data source uses the Standard Transportation Commodity Codes (STCC) to identify the product designation for the commodities transported. In the Carload Waybill Sample, this field includes the first five digits of the seven-digit STCC; however, the codes for some com- modities, like STCC 19 series commodities (ordnance [guns and artillery] or accessories) are reported only at the two-digit level. Commodities in the STCC 49 series (hazardous materials) and STCC 50 series (bulk materials in boxcars) also have been translated to actual product commodity codes. • The Public Use Waybill Sample, which is a sub-parent of the Carload Waybill Sample, does not include hazardous materials (STCC series 49xxx) or bulk materials in boxcars (STCC series 50xxx).9 Center for Transportation Analysis Intermodal Terminals Database • Only grain elevators, cement terminals, petroleum tank farms, and liquid bulk storage and transfer terminals with waterway connections are included in the data source. All other data on intermodal terminals will need to be obtained from additional sources.10 • This data source excludes many public warehouses served by rail or truck-rail reload centers for lumber, steel, paper, or other break-bulk freight. Additional information on these public warehouses will need to be obtained from other sources. Commodity Flow Survey (CFS)11 • Within the CFS, a commodity is defined as a product that an establishment produces, sells, or distributes. This definition does not include items that are considered excess or operational waste. Survey respondents report the description and the five-digit SCTG (Standard Classifi- cation of Transported Goods) code for the commodity contained in the shipment. Shipments having multiple commodities are grouped together, and the commodity with the greatest

Differences in Data Element Definitions 49 weight is selected to represent the total shipment. Commodities that are part of the shipment but are not the majority weight are not classified. • Shipments originating from business establishments located in Puerto Rico and other U.S. possessions and territories are excluded from the data file. Thus, commodities originating from these locations also are excluded. • Data for government-operated establishments are excluded from the CFS. These establish- ments include public utilities, publicly operated bus and subway systems, public libraries, and government-owned hospitals. • The CFS also excludes establishments or firms with no paid employees. Data users should be aware that any commodities imported or exported from or for these establishments are excluded. • Commodities shipped via containerized cargo are labeled Intermodal. Commodities that move by more than one mode are labeled “multiple modes and mail.” These classifications differ from those used in other data sources. Fatal Analysis Reporting System (FARS)12 • To be included in FARS, a crash must involve a motor vehicle traveling on a trafficway customarily open to the public within the 50 states, the District of Columbia, or Puerto Rico that resulted in the death of a person (occupant of a vehicle or a non-motorist) within 30 days of the crash. Notably, any hazardous commodities not meeting the criteria for inclusion in the FARS are excluded from the report. Therefore, FARS data on hazardous materials can only be used with other data on hazardous materials in circumstances that involved a fatal crash. • NHTSA updates the FARS Analytical User’s Manual every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. Foreign Trade Statistics (FTS)13 • Imports and exports of commodities on vessels moving under their own power or afloat, and on aircraft flown into or out of the United States, are included in the “All Methods” data table but excluded from the “Vessel and Air” statistics. Thus, commodities that are shipped on ves- sels moving under their own power and via aircraft are included in the former, but excluded from the latter. • Mail and parcel post shipments (including those transported by vessel or air) are included in the “All Methods” data table but excluded from the “Vessel and Air” statistics. • Low-value shipments, which are defined as exports valued under $2,501 or imports valued under $1,251, are included in the “All Methods” data table but excluded from the “Vessel and Air” statistics. Commodities that qualify under low-value shipments are included in the “All Methods” data. Low-value shipments are estimated, and may not directly correspond to other data source estimates.14 Freight Analysis Framework (FAF3)15 • The FAF3 is built primarily on the Commodity Flow Survey (CFS), but the FAF3 does not have the level of commodity detail found in the CFS, nor does it identify hazardous cargo. • Commodities shipped via containerized cargo are labeled “intermodal.” Commodities that move by more than one mode are labeled “multiple modes and mail.” These classifications differ from those used in other data sources.

50 Implementing the Freight Transportation Data Architecture: Data Element Dictionary Motor Carrier Management Information System (MCMIS) • Hazardous materials information is not available to the general public, so authorization is needed to obtain data on these commodities. Without proper authorization, commodity data on hazardous materials cannot be used with other sources. North American Transborder Freight Database (Transborder)16 • Commodities are identified using the two-digit commodity code indicated by Schedule B for U.S. export shipments and the HTS (Harmonized Tariff Schedule) for U.S. import shipments. • Because of customer requests, the U.S. Bureau of Transportation Statistics discontinued the inclusion of trans-shipment activity in Transborder freight data beginning with the January 1997 data month. • Air and vessel data by month or year are not available before 2004. • Import values from Mexican states are not available. U.S. Waterway Data17 • The U.S. Waterway database is the only data source focused exclusively on waterways, includ- ing inland waterways, offshore waters, the Great Lakes, and the Saint Lawrence Seaway. Data on commerce, facilities, locks, dredging, imports and exports, and accidents are included, along with the geographic waterway network. • Commodities reported in this data source include coal, petroleum products, chemicals, crude materials, manufactured goods, farm products, machinery, and waste, and commodities labeled “unknown.”18 All other commodities are excluded from the data source. Vehicle Inventory and Use Survey (VIUS)19 • The VIUS excludes federal, state, or local government vehicles, as well as ambulances, buses, motor homes, farm tractors, unpowered trailer units, and trucks that have been reported to have been sold, junked, or wrecked before January 1 of the survey year. Any commodities, including hazardous materials, carried via the excluded vehicles also are excluded from the survey, which may create difficulties when bridging with other data sources. 6.4.1.2 Differing Classification Systems Data sources that report commodity information often utilize commodity classification systems that differ from one data source to another. The most commonly used commodity classification systems are the Standard Classification of Transported Goods (SCTG), Standard Transportation Commodity Codes (STCC), Harmonized System (HS), and Standard Inter- national Trade Classification (SITC). Data users are advised to be aware of the different resolu- tions of these systems and account for them when performing data analysis. A detailed discussion of each classification system appears in Appendix B. Carload Waybill Sample20 • The Carload Waybill Sample contains various resolutions of freight movements reported at the Business Economic Area (BEA)-to-BEA level (or across multi-county BEA areas) and the seven-digit STCC level. Data users are advised to note the following differences in resolution between tables:

Differences in Data Element Definitions 51 – UNIQUE SERIAL NUMBER—To allow for unique identification of waybills, the Associa- tion of American Railroads/Railinc assigns a unique, six-digit serial number to all waybills processed. ▪ Hardcopy waybills are assigned serial numbers in the 100,000 to 199,999 range. ▪ Machine Readable Input (MRI) waybills are assigned serial numbers in the 200,000 to 999,999 range and 000,000 to 099,999. This unique serial number does not correspond with commonly used system codes.21 – WAYBILL NUMBER—This number is the number an originating railroad document assigns to each waybill. The waybill number gives detailed instructions relating to a ship- ment, and the codes vary depending on the consignor or consignee, the point of origin, its destination, and route.22 – CONFIDENTIAL CARLOAD WAYBILL SAMPLE COMMODITY CODE (STCC)—This data element uses the STCC coding to identify the product designation for the commodity being transported at the seven-digit STCC level.23 ▪ The STCC 48 series (hazardous waste) is part of the regular STCC. ▪ The STCC 49 series (hazardous materials) is used only for hazardous materials, in lieu of the regular STCC. ▪ The STCC 50 series is used for bulk commodities transported in box cars. – STCC W/O HAZARDOUS (49) CODES—This data element on the Confidential Carload Waybill Sample takes the hazardous codes (STCC series 49xxxxx) and bulk codes (STCC series 50xxxxx), and translates them to the actual product commodity codes. – Public Use Waybill Sample Commodity Code (STCC)—The STCC identifies the product designation for the transported commodity. This data field includes the first five digits of the seven-digit STCC; however, STCC 19 series commodities are reported only at the two- digit level. Center for Transportation Analysis Intermodal Terminals Database24 • The intermodal terminals data source contains a list of 3,100 transload facilities in the United States where commodities may be transferred between surface modes. Data users are advised to note the following difference in resolution between tables: – CARGO—A three-digit code for the type of cargo or commodity group involved in the intermodal connection. This code does not correspond to other data sources.25 Commodity Flow Survey (CFS)26 • The CFS contains shipment data using varying resolutions of the Standard Classification of Transported Goods (SCTG) system. Data users are advised to note the following difference in resolution between tables: – COMMODITY—A product that an establishment produces, sells, or distributes. Respon- dents report the description and the five-digit SCTG code for the commodity contained in the shipment. Shipments involving multiple commodities are classified as the commodity with the greatest weight in the total shipment. Fatal Analysis Reporting System (FARS)27 • FARS is a nationwide census providing NHTSA (National Highway Traffic Safety Admin- istration), Congress, and the American public yearly data regarding fatal injuries suffered in motor vehicle traffic crashes. Data users are advised to note the following differences in resolution between tables:

52 Implementing the Freight Transportation Data Architecture: Data Element Dictionary – HAZ_ID—This data element identifies the four-digit hazardous material identification number for a vehicle in transit. These numbers are developed by the United Nations (UN) and used worldwide in international commerce and transportation to identify hazardous materials. Materials without a UN number may be assigned a four-digit North American (NA) number, which usually starts with the number 8 or the number 9.28 – HAZ_CNO—This data element identifies the single-digit hazardous material class number for a vehicle in transit. The U.S. DOT has identified nine hazard classes based on the dan- gers posed in transportation.29 – PHAZ_ID—This data element applies to parked and working vehicles and uses the same four-digit hazardous material identification number for a vehicle as HAZ_ID. – PHAZ_CNO—This data element applies to parked and working vehicles and uses the same single-digit hazardous material class number for a vehicle as HAZ_CNO. • NHTSA updates the FARS Analytical User’s Manual30 every year to summarize the evolu- tion of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. Foreign Trade Statistics (FTS)31 • The FTS database contains varying resolutions of HS (Harmonized System) industry clas- sifications for different tables. Data users are advised to note the following differences in resolution between tables: – U.S. Exports of Merchandise—Monthly—This data table contains commodity details using a variety of codes at varying levels. These include the 10-digit Schedule B code, the 5-digit SITC (Standard International Trade Classification) code, the 10-digit HTS (Harmonized Tariff Schedule) code, the 5-digit End-Use code, and the 6-digit HS code.32 – U.S. Imports of Merchandise—Monthly—This data table contains commodity detail at the 2-, 4-, 6-, and 10-digit HS levels.33 – U.S. Exports and Imports by Port—This data element contains various data fields for HS commodities at the six-digit HS level. Freight Analysis Framework (FAF3)34 • The FAF3 contains varying resolutions of Standard Classification of Transported Goods (SCTG) classifications for different tables. Data users are advised to note the following differ- ence in resolution: – SCTG2—This data element contains commodity codes that are based off the SCTG code at the two-digit level.35 Motor Carrier Management Information System (MCMIS)36 • The MCMIS contains information on the safety fitness of commercial motor carriers and hazardous material shippers subject to the Federal Motor Carrier Safety Regulations and the Hazardous Materials Regulations. Data users are advised to note the following differences in resolution between tables: – HAZMAT MATERIAL ID—This identifying code is associated with hazardous materials cargo. The codes correspond with four-digit United Nations/North American (UN/NA) identification numbers.37 – HAZARDOUS MATERIALS CARRIED/SHIPPED—This code identifies the type of haz- ardous material transported or shipped by the entity and whether bulk (B), non-bulk (N), or all (A). It is important to note that the conversion of the Hazardous Materials Data elements

Differences in Data Element Definitions 53 of the new Census File to the old is as follows: Bulk (B) = Tank (T), Non-Bulk (N) = Package (P), and All (A) = Both (B). These codes do not correspond to other data sources. – CARGO—Describes the cargo hauled by a particular carrier. A maximum of three cargo types are printed. These codes do not correspond to other data sources.38 – HAZMAT C—This code identifies the type of hazardous material carried by interstate and intrastate motor carriers. Up to three hazardous material types may be printed. The letter B indicates that the cargo is carried in bulk quantities. N indicates that the cargo is carried in non-bulk quantities. A indicates cargo that is carried both in bulk and non-bulk quantities. These codes do not correspond to other data sources. – HAZMAT S—This code identifies the type of hazardous material shipped by interstate and intrastate shippers, with cargo coded the same as for HAZMAT C. Up to three hazardous materials types may be printed. The letter B indicates that the cargo is shipped in bulk quan- tities. N indicates that the cargo is shipped in non-bulk quantities. A indicates cargo that is shipped both in bulk and non-bulk quantities. These codes do not correspond to other data sources. National Agricultural Statistics Service (NASS)39 • The NASS database contains varying resolutions of agricultural and demographic statistics for different tables within the data source. Data users are advised to note the following differ- ences between tables: – SECTOR—In this data source, sectors constitute five high-level, broad categories that are useful in narrowing down choices: Crops, Animals & Products, Economics, Demographics, and Environmental. These codes do not correspond to other data sources. – GROUP—These data elements are subsets within a sector (e.g., under the sector Crops, the groups are Field Crops, Fruit & Tree Nuts, Horticulture, and Vegetables). These codes do not correspond to other data sources. – COMMODITY—This data element records the primary subject of interest (e.g., Corn, Cattle, Labor, Tractors, Operators). These codes do not correspond to other data sources. North American Transborder Freight Database (Transborder)40 • Transborder contains varying resolutions of freight flow data by commodity type and by mode of transport (rail, truck, pipeline, air, vessel, and other) for U.S. exports to and imports from Canada and Mexico at the North American Free Trade Agreement (NAFTA), national, and state/province level. Data users are advised to note the following differences in resolution between tables: – COMMODITY—This data element contains export or import commodity detail and is available at the two-digit HS (Harmonized System) level. – U.S. Trade with Canada and Mexico Import Commodity Detail—This data element contains commodity codes indicated by a two-digit Schedule B number for U.S. export shipments and two-digit Harmonized Tariff Schedule (HTS) for U.S. import shipments. Schedule B and HTS codes correspond to HS codes up to the six-digit level. U.S. Waterway Data41 • The U.S. Waterway database contains varying resolutions of data on commerce, facilities, locks, dredging, imports and exports. Accidents are included along with the geographic

54 Implementing the Freight Transportation Data Architecture: Data Element Dictionary waterway network. Data users are advised to note the following differences in resolution between tables: – CONTAINER—This data element indicates whether the vessel carries containers (signified by the letter C) or not (left blank).42 These codes do not correspond to other data sources. – PMS_COMM—This data element contains the two-digit Lock Performance Monitoring System (LPMS) commodity code. The LPMS is based off the Standard International Trade Classification (SITC) Revision 3 commodity code.43 – PRINC_COMM—This data element describes the principal commodities carried by a Transportation Lines vessel company.44 Vehicle Inventory and Use Survey (VIUS)45 • The VIUS contains varying resolutions of data on the physical and operational characteristics of the nation’s truck population at the national level and state level. Data users are advised to note the following differences in resolution between tables: – PRODUCT_PRINCPL—This data element indicates the principle product carried, at the two-digit SCTG (Standard Classification of Transported Goods) level, by this vehicle con- figuration.46 Products are recoded to the highest percent; if the highest percent occurs for more than one category, the record is assigned to “multiple categories.”47 6.5 Import and Export Data Elements The narrative given in this section is derived from the following documents: • Border Crossing/Entry Data: FAQ (frequently asked questions). Retrieved from http://Transborder.bts.gov/programs/international/Transborder/TBDR_ BC_FAQs.html • A Description of the FAF3 Regional Database and How It Is Constructed. The Freight Analysis Framework Version 3 (FAF3). FHWA, June 16, 2011. Retrieved from http://faf.ornl.gov/fafweb/Data/FAF3ODDoc611.pdf • Principal Ports of the United States. U.S. Waterway Data. Navigation Data Center. Retrieved from http://www.navigationdatacenter.us/data/ datappor.htm • Guide to Foreign Trade Statistics. U.S. Census Bureau. Retrieved from http:// www.census.gov/foreign-trade/guide/ • Transborder Freight Data Program (Transborder Documentation). U.S. Department of Transportation, Research and Innovative Technology Administration Bureau of Transportation Statistics (RITA/BTS), September 2009. Retrieved from http://Transborder.bts.gov/programs/international/ Transborder/PDF/TransborderFreightDataProgram.pdf The authors recommend reviewing these sources for additional information concerning each data source. This narrative serves only as a summary of information gleaned from these sources. Keywords: import, export, port of entry, trade, foreign trade The import and export data elements discussion relates to databases that report on U.S. trade with other countries or geographical regions. The Bureau of Transportation Statistics (BTS)

Differences in Data Element Definitions 55 classifies U.S. international trade and transportation data into three primary categories: admin- istrative trade data, carrier-based data, and shipper-based data. These categories are based on how the data is collected and the scope of each data source. The taxonomic, temporal, and methodological/analytical differences that still exist in these data sources are discussed in this section of NCFRP Report 35. • Administrative Trade Statistics—These international trade statistics are captured from administrative documents required by the Department of Homeland Security (DHS). U.S. Customs and Border Protection (CBP) is responsible for collecting this information either in paper form or electronic form at U.S. ports of entry, exit, or clearance. Currently, electronic information is captured through the Automated Broker Interface for imports and through the Automated Export System for exports. Together, the Automated Broker Interface and Automated Export System are known as the Automated Commercial System. The Auto- mated Commercial System is being replaced by the Automated Commercial Environment (ACE), which will serve as the primary system through which the trade community will report imports and exports. CBP has established a schedule for completing development of all trade processing capabilities in ACE by the end of 2016.48 • The U.S. Census Bureau’s Foreign Trade Division is responsible for verifying, processing, and distributing the data after collection by the CBP. Other federal agencies receive special tabulations from the Census Bureau, based on the official U.S. international trade statistics. These agencies then perform additional quality assurance reviews and analyses for their own purposes and to meet the needs of their customers. Following are the types of administrative trade statistics gathered: – Foreign trade statistics – North American land trade, disseminated as the North American Transborder Freight Data (Transborder) – U.S. international maritime trade, released to the Maritime Administration and the U.S. Army Corps of Engineers (USACE) – U.S. transportation-related goods and overall trade data, released to the U.S. Bureau of Economic Analysis • Carrier-Based Sources—The main sources of carrier-based international trade data are: – International air freight data from the Research and Innovative Technology Administration (RITA)/BTS, disseminated as the Air Carrier Statistics – Maritime data from the Journal of Commerce’s Port Import/Export Reporting Service (PIERS) – Special periodic surveys, such as Canada’s National Roadside Survey • Shipper-Based Sources—The Commodity Flow Survey (CFS, conducted in 1993, 1997, 2002, and 2007) is the only publicly available shipper-based survey that provides some information on U.S. international trade and transportation. The export data is limited, however, and not directly comparable to merchandise trade exports released by other sources, including the Census-based Foreign Trade Statistics (FTS). 6.5.1 Taxonomic Differences Taxonomic differences among various data sources relating to their scope and definitions of import/export data fall under the following categories: • Sources that report on foreign trade movement origin and destination, including the port of entry (e.g., Foreign Trade Statistics [FTS], Transborder, Freight Analysis Framework) • Sources that report on foreign trade movement only at the port-of-entry level (e.g., Border Crossing Entry Data, U.S. Waterway Data)

56 Implementing the Freight Transportation Data Architecture: Data Element Dictionary • Sources that report on foreign trade movement origin and destination but exclude the port of entry (e.g., Carload Waybill Sample) 6.5.1.1 Sources that Report on Foreign Trade Movement’s Origin and Destination, Including the Port of Entry Foreign Trade Statistics (FTS) • The Foreign Trade Statistics (FTS) data compiled by the Census Bureau are the official U.S. import and export statistics and reflect both government and nongovernment shipments of merchandise between foreign countries. The data is made available for subscription in four different formats: (1) Merchandise Trade, (2) State Data, (3) Port Data, and (4) Special Prod- ucts.49 Following are the taxonomic differences in the four types of data: – Merchandise Trade—These data files provide commodity information for different com- modity classification codes, as follows: ▪ 10-digit Schedule B ▪ 10-digit Harmonized System (HS) ▪ Standard International Trade Classification (SITC) ▪ End-Use, North American Industry Classification System (NAICS) ▪ USDA ▪ Advanced Technology Products Merchandise Trade data files also include only port district information. Individual ports of entry/exit are not reported in these data files. – State Data—These data files report summarized trade statistics by U.S. state. In comparison to the merchandise trade data files, commodity data is available in just two formats: the six-digit HS code and four-digit NAICS code. State Data files do not include information on the port of entry/exit. The State Exports/Port database reports trade data by U.S. state, district, and port of exit. Time data, however, is reported in periods rather than in statistical year and month as in the other databases. For example, the period coded using the num- ber 1 covers January, February, and March; period 2 is April, May, and June; period 3 is July, August, and September; and period 4 is October, November, and December. State Imports/ Port data is not published. – Port Data—These files report trade information through the individual port of entry/exit but exclude the U.S. state of origin or destination. Commodity data is reported at the six- digit HS commodity code level. – Special Products—These data files report on the following: ▪ U.S. general imports assembled abroad from components produced in the United States (textile summary) ▪ U.S. imports for consumption and general imports for all imports entered under second- ary or Census special program indicators ▪ Shipments of merchandise from the United States to Puerto Rico and the U.S. Virgin Islands, and shipments from Puerto Rico, the U.S. Virgin Islands, Guam, American Samoa, the Northern Mariana Islands, and U.S. minor outlying islands to the United States. North American Transborder Freight Database (Transborder) • Transborder is a subset of the Foreign Trade Statistics (FTS) database. It is the first attempt by the U.S. Census Bureau to disaggregate U.S. foreign trade statistics into the various surface modes of transportation. Transborder contains freight flow data by commodity type and by mode of transport for U.S. exports to and imports from Canada and Mexico. Taxonomic

Differences in Data Element Definitions 57 characteristics of data elements in this database that may differ from those in the other data sources include the following: – Geographic scope: ▪ USASTATE—U.S. states (introduced January 2007)  This data element is based on the two-digit U.S. Postal code.  It identifies the U.S. state of origin for exports to or state of destination for imports from Canada and/or Mexico. The state may not always represent the physical origin or destination of the import or export goods, because the exporter’s or importer’s address may not be in the same state as the origin or destination of the goods. ▪ TRDTYPE—This data element identifies the direction in which the commodity is moved. The USASTATE with TRDTYPE will identify the origin and destination information based on whether TRDTYPE is an import or an export. ▪ DEPE—The U.S. Census Bureau is responsible for maintaining the classification of U.S. Customs districts/ports of entry, codes, and descriptions. This classification is known as “Schedule D.”  For imports, this data element represents where the entry documentation was filed with Customs and the duties paid, and it may not always be where the goods physically entered the United States.  For U.S. exports, this data element represents the last port where the shipment is cleared for export.  State totals for trade can be based on the state of destination for imports and the state of origin for exports, not on the state Customs port of entry or exit. This is because many border ports serve as national gateways, and not all the goods that enter or exit through a port either originated in or are destined for that particular state. ▪ CANPROV—Canadian provinces  For U.S. imports from Canada, the Canadian province represents where the goods were grown, manufactured, or otherwise produced. However, the province information may also reflect the province used as the mailing address of the Canadian exporter or the address of an intermediary; therefore, in some instances, the mailing address may not be the actual province of physical origin.  For U.S. exports to Canada, the Canadian province represents the Canadian province of clearance. The province of clearance is the province in which Canadian Customs cleared the shipment, and is not necessarily the province of final destination. ▪ MEXSTATE—Mexican states  The Mexican state of destination is the state in which the ultimate consignee is located in Mexico, and is not necessarily the state of final destination. The Census Bureau cap- tures the data field for MEXSTATE from the ultimate consignee’s address. If a Mexican state of destination cannot be identified for a particular shipment, it is considered unknown and coded as OT in the data field.  Data for the Mexican state of origin for U.S. imports from Mexico is not captured as part of current trade filing requirements. – STATMO—Data reported for each calendar month – COMMODITY—Commodity data reported using the two-digit HTS (Harmonized Tariff Schedule) – DISAGMOT—Disaggregated Mode of Transport ▪ This data element represents only the mode by which shipments enter or exit the United States and does not reflect all the modes of transportation used throughout the entire journey of the shipment, from foreign point of origin to final destination. ▪ DISAGMOT uses numerical codes to signify the following modes:  1 = Vessel  3 = Air

58 Implementing the Freight Transportation Data Architecture: Data Element Dictionary  4 = Mail (U.S. Postal Service)  5 = Truck  6 = Rail  7 = Pipeline  9 = Foreign trade zones  8 = Other (including unknown) (Foreign trade zones was added as a mode of transport in 1995.) “Mail” is used as the mode used for U.S. Postal Service shipments and cannot be further divided into either rail or truck shipments. The category “Other” includes “flyaway aircraft, or aircraft moving under their own power (i.e., aircraft moving from the aircraft manufacturer to a customer and not carrying any freight), powerhouse (electricity), vessels moving under their own power, pedestrians carrying freight, unknown, and miscellaneous other.”50 Users should note that the actual mode of transport for a specific shipment into or out of a foreign trade zone is unknown because U.S. Customs and Border Protection (CBP) does not collect this information. – VALUE and SHIPWT—These data fields report data by value and weight. ▪ For imports, the data field VALUE refers to the Customs value or the value of merchandise for duty purposes. It is usually the selling price in the foreign country of origin. VALUE excludes freight costs, insurance, and other charges incurred in bringing the merchandise from the foreign port of export to the United States. ▪ For exports, the data field VALUE refers to the value of the merchandise, usually the selling price, plus insurance and freight at the U.S. port of export. The value, as defined, excludes the cost of loading the merchandise aboard the exporting carrier at the port of export and also excludes freight, insurance, and any charges or transportation costs beyond the U.S. port of export. ▪ For exports, the weight of U.S. exports by land modes of transportation is not available because this data is not required to be reported on the paper Shipper’s Export Declara- tions documents required by the U.S. Census Bureau. The new electronic filing system for exports, the Automated Export System, does require that export weight be filed for all modes of transport. RITA/BTS uses the value-to-weight ratio of U.S. imports at the two-digit commodity code level to calculate the export weights. Although the export weights are not published as tables, RITA/BTS uses these numbers for U.S. Transborder publications. Freight Analysis Framework (FAF3) • Taxonomic characteristics of data elements in this database that may differ from those in the other data sources include the following: – Geographic scope for imports and exports are either reported at the state level or based on FAF3 zones and regions (e.g., FR_ORG, DMS_ORG, FR_DEST DMS_DESTST), which include FAF zone-specific ports of entry and exit. – Historical data is available for a limited number of years (1997, 2002, 2007, and 2012). Projected data is reported in 5-year increments from 2015 to 2040. – Two-digit commodity codes (SCTG2 is the data field) are based on the SCTG (Standard Classification of Transported Goods) classification system. – Mode of transport is classified as inbound, outbound, or domestic, as follows: ▪ FR_INMODE—This data element represents mode of transport from foreign origin to zone of entry. ▪ FR_OUTMODE—This data element represents mode of transport from zone of exit to the foreign destination.

Differences in Data Element Definitions 59 ▪ DMS_MODE—This data element represents the domestic mode of transport from zone of entry to destination zone for imports, and from origin zone to zone of exit port for exports. – Data is reported by value (VALUE), weight (TONS), and ton-miles (TMILES), and is deter- mined based on mode-specific data modeling procedures. (See Section 6.5.3 Methodological Differences in this chapter for a discussion of import/export data elements.) Air Carrier Statistics • Air Carrier Statistics—This data source reports data differently from other data sources. The T-100 Market data tables, which report only on trips from origin to destination, exclude port-of-entry/exit information if the port of entry/exit is an intermediate stop for the ship- ment. The T-100 Segment data tables, on the other hand, include the port of entry/exit for international shipments but exclude the original origin-destination when a shipment has mul- tiple stops. The Market and Segment data tables report only on the weight of cargo shipped and exclude both value and commodity type data. Unlike other data sources, however, the Air Carrier Statistics reports data at a more disaggregate level (i.e., origin and destination city and airport). 6.5.1.2 Sources that Report on Foreign Trade Movement Only at the Port-of-Entry Level Border Crossing/Entry Data • Border Crossing/Entry Data provides summary statistics for incoming crossings at the U.S.– Canadian and the U.S.–Mexican border at the port level. The data element MEASURE provides information on the number of personal vehicles, trucks, buses, containers, trains, passengers, or pedestrians entering the United States. The data, which is reported on a monthly basis (YEAR and MONTH), does not include information on the place of origin or final destina- tion of commodities transported. It provides information only on the port of entry or exit (PORT LOCATION). U.S. Waterway Data—Principal Ports of the United States • The U.S. Waterway Principal Port file contains USACE commodity tonnage summaries (total tons, domestic, foreign, imports, and exports), port codes (PORT), geographic loca- tions (LONGITUDE and LATITUDE), and names (PORT_NAME) for the top 150 ports for a particular year. Commodity names and descriptions are not available; neither is there data on place of origin or final destination. 6.5.1.3 Sources that Report on a Shipment’s Origin and Destination and Exclude the Port of Entry Confidential Carload Waybill Sample The Confidential Carload Waybill Sample contains the data element field TYPE MOVE, whose options include details from which a user can infer whether that particular data record involves an import or export commodity or none. Available field options include the following: • Neither import nor export • Imported commodity • Exported commodity • Commodity imported and exported (e.g., land bridge traffic) • Unknown

60 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.5.2 Temporal Differences Data users should be aware of temporal differences among and within databases as a result of varying frequency of data collection and changes in the definition of a data element over time. Foreign Trade Statistics (FTS) Temporal differences in the FTS database, as documented in the Guide to Foreign Trade Statis- tics, include the following: • The United States is substituting Canadian import statistics for U.S. exports to Canada in accor- dance with a 1987 Memorandum of Understanding signed by the Census Bureau, U.S. Customs and Border Protection (CBP), Canadian Customs, and Statistics Canada. This data exchange includes only U.S. exports destined for Canada and does not include shipments destined for third countries by routes passing through Canada or shipments of certain grains and oilseeds to Canada for storage before exportation to a third country, which are reported on and compiled from Electronic Export Information documents. • The statistical month of importation is the month in which U.S. Customs and Border Protec- tion (CBP) releases the merchandise to the importer. • The statistical month of exportation is based on the date when the merchandise leaves the United States. (For vessel or air shipments, it is the date when the carrier departs or is cleared from the port of export.) • The Census Bureau seasonally adjusts the merchandise trade data at the five-digit end-use commodity category level, the most detailed end-use level possible. These detailed data are then summed to the one-digit level for release with the monthly merchandise trade totals. • Effective with the release of January 2014 statistics on March 7, 2014, the Census Bureau pub- lishes seasonally adjusted selected countries and world areas in FT-900 Exhibit 19. Unlike the commodity-based adjustments, these adjustments are developed and applied directly at the country and world area level. North American Transborder Freight Database (Transborder) • For temporal differences in data element definitions for the Transborder database and several significant reporting changes since the first release of data, see the section on “Major Report- ing Changes” in the Transborder Freight Data Program Documentation.51 For example, start- ing in January 2007, the Bureau of Transportation Statistics (BTS) used a new data structure to release the Transborder data. Twenty previously separate data tables were consolidated into the current form of three data tables: – U.S. Imports and Exports with State and Port Detail – U.S. Imports and Exports with State and Commodity Detail – U.S. Imports and Exports with Port and Commodity Detail • In addition, trans-shipments data (covering shipments from a third country through Canada or Mexico to the United States or from the United States to a third country through Canada or Mexico) were excluded from the public database beginning with the January 1997 data. (Note: Before January 1997, documentation for this dataset referred to this type of activity as “in-transit shipments.”) Freight Analysis Framework (FAF3) • For temporal differences, see the discussion on data collection methods and limitations of the FAF database available on the Freight Data Dictionary web application.

Differences in Data Element Definitions 61 Border Crossing/Entry Data • Temporal differences within the Border Crossing/Entry Data include the following: – Data on passenger vehicles and passengers in personal vehicles for the Cape Vincent, NY, ferry are available beginning in 2007. The ferry between Wolfe Island (Canada) and Cape Vincent does not operate in the winter. – Until May 2011, truck and rail data for the port of entry of Otay Mesa, CA, was reported by the Bureau of Transportation Statistics (BTS) as Otay Mesa/San Ysidro, CA, which is the same as the CBP’s reporting of the data to the BTS. However, San Ysidro has been a pas- senger crossing for many years and no freight is allowed through this port of entry by truck or rail. Hence, BTS changed the name to Otay Mesa for truck and rail crossings to avoid any confusion to the data user. Thus, Otay Mesa, CA, and San Ysidro, CA, are now reported as separate ports of entry for all data elements. 6.5.3 Methodological Differences Methodological differences arise among import and export data elements of different data- bases as a result of the processes by which the data is collected and disseminated. Foreign Trade Statistics (FTS) • Electronic Export Information—These mandatory documents are filed by the U.S. Principal Party in Interest or its agents through the Automated Export System and record U.S. exports data for merchandise from all countries except Canada. • Automated Commercial System—This automated U.S. Customs database compiles U.S. imports data on merchandise. U.S. imports data on merchandise also is compiled from import entry summary forms, warehouse withdrawal forms, and foreign trade zone docu- ments as required by law to be filed with U.S. Customs and Border Protection (CBP). Data on imports of electricity and natural gas (NG) from Canada is obtained from Canadian sources. North American Transborder Freight Database (Transborder) and Border Crossing/Entry Data • Transborder Surface Freight Data—This data is extracted from the foreign trade statistics collected by the Census Bureau. • Border Crossing/Entry Data—This data is based on transportation mode count data collected by U.S. Customs and Border Protection (CBP). Freight Analysis Framework (FAF3) • Import and export flows are constructed using mode-specific data sources, each of which is converted from agency specific commodity codes to FAF3’s two-digit SCTG (Standard Clas- sification of Transported Goods) codes. In addition, commodity flows from the respective databases are either spatially aggregated or disaggregated to match FAF3 regions. • This list summarizes the mode-specific data modeling procedures used in developing the FAF3: – Waterborne Imports and Exports ▪ Main sources of data include the following:  FAF3-specific extraction of data from the Port Import/Export Reporting Service (PIERS) maritime database  USACE’s International Waterborne Commerce database  Foreign Trade Statistics (FTS) database

62 Implementing the Freight Transportation Data Architecture: Data Element Dictionary ▪ PIERS forms the basis of foreign waterborne flows in FAF3 with several adjustments, including the following:  Ensuring PIERS total commodity tonnages is consistent with USACE Waterborne tonnages  Ensuring PIERS total commodity dollar valued trades is consistent with FTS totals  Inferring missing data for zip-code-level reporting of shipment originations and desti- nations within the continental United States and inland mode of transport within the continental United States  Addressing known issues, such as the reporting of origin and destination data in the FTS dataset (i.e., exporting/importing company addresses are reported rather than the actual physical location of the point of departure or arrival of the shipment) ▪ Missing or questionable data were allocated across domestic FAF3 zones in proportion to the distribution of shipment volumes in the 2007 U.S. Commodity Flow Survey (CFS). – Air Freight Imports and Exports ▪ Main sources of data include the following:  The Air Carrier Statistics T-100 International Market data table provides estimates of total tons shipped annually between an originating airport (where the cargo is first loaded onto an aircraft) and a destination airport (where the cargo is unloaded for final land-based delivery, usually by truck).  The FTS database provides information on value, commodity class, quantity, method of transportation, and shipping weights. ▪ The Air Carrier Statistics T-100 data tables and FTS database are combined into a single FAF3 air freight dataset by reconciling differences in the level of spatial and commodity detail. If differences exist between the T-100 and FTS flow totals, the T-100 data tables are taken to be definitive for total tons shipped, and the FTS database is used to control the allocation of freight shipments across commodity classes and to assign value-to-weight ratios to these flows. – Transborder U.S.–Canada and U.S.–Mexico Imports and Exports ▪ The main source of data is the North American Transborder Freight Database (Trans- border). Shipments were allocated to the most likely counties of origination or destina- tion in each state using the 2007 U.S. County Business Pattern data. ▪ Origin-destination (O-D) estimation is done by:  Removing vessel, air, and pipeline mode movements from the dataset (leaving truck and rail “land” border shipments)  Spatially allocating flows reported at the state level to their most likely FAF3 regions within the United States  Converting Harmonized System (HS) commodity classes to FAF3 SCTG classes ▪ Shipment weight data for exports to Canada and Mexico is estimated on the basis of average dollar/ton statistics generated from export shipments by specific HS commodity class, mode, and country. – Imports of Crude Petroleum ▪ Monthly reported Energy Information Administration (EIA) data containing company, U.S. seaport of entry/exit, and foreign country information is used to estimate crude petroleum imports in FAF3. ▪ O-D flow is represented as movement from foreign country (i.e., source of commodity import) to a U.S. port (domestic FAF3 origin region), then to a U.S. refinery (FAF3 domes- tic destination region). Allocation of these flows to specific modes of transportation are based on EIA data on crude oil refinery receipts, broken down by mode of transport (ship, pipeline, rail, barge, truck), and further broken down by domestic versus foreign sources of production.

Differences in Data Element Definitions 63 – Imports and Exports of Natural Gas (NG) ▪ EIA reports annual movement of liquefied natural gas (LNG), which is carried by larger tanker ships to and from a U.S. seaport of entry/exit. EIA also reports on natural gas (NG) trade by pipeline between the United States and Canada or Mexico. Supporting data used in allocating flows to specific FAF3 O-D pairs came from the U.S. Census Bureau’s County Business Patterns dataset. NG flows were allocated to respective FAF3 domestic regions based on U.S. port of entry or exit, and exporting countries also were allocated to their respective FAF3 foreign regions. • For additional information on the data sources, estimation methods, and data quality issues, please refer to “Estimation of Import and Export Flows” in The Freight Analysis Framework Version 3—A Description of the FAF3 Regional Database and How It Is Constructed.52 Air Carrier Statistics • The Federal Aviation Act of 1958 requires each large certificated air carrier to file Form 41 (reports of financial and operating statistics) monthly, quarterly, semiannually, and annually with the Bureau of Transportation Statistics (BTS). BTS publishes the data submitted on the forms as the Air Carrier Statistics.53 U.S. Waterway Data • In this database, the Principal Ports of the United States data is derived from the Waterborne Commerce Statistics Center. 6.6 Industry Data Elements Keywords: industry, NAICS, SIC, HS (Harmonized System) code When working with data elements related to industry classification, data users should be aware of taxonomic and methodological/analytical differences among and within data sources. This section discusses differences between data elements that classify industries, including the North American Industry Classification System (NAICS), HS (Harmonized System), and SIC (Standard Industrial Classification) industry codes. 6.6.1 Taxonomic Differences Taxonomic differences among and within data elements related to industry classification can be categorized as follows: • Inclusion/exclusion of industry groups • Industry definition resolution 6.6.1.1 Inclusion/Exclusion of Industry Groups Data sources may sometimes include or exclude certain industry groups, making it difficult to directly compare industries across multiple data sources. Annual Survey of Manufacturers • Annual Survey of Manufacturers—This database covers only manufacturing establishments with one or more paid employees, along with non-employers that use leased employees for manufacturing (which are classified in NAICS Sector 31-33).54 All other industries are

64 Implementing the Freight Transportation Data Architecture: Data Element Dictionary excluded from this data source, which creates difficulty in making comparisons with other data sources’ industry classifications. County Business Patterns • County Business Patterns—This database covers most North American Industry Classifica- tion System (NAICS) industries, with the following exceptions: – NAICS 111 and NAICS 112—Crop and animal production – NAICS 482—Rail transportation – NAICS 491—Postal Service – NAICS 525110, 525120, and 525190—Pension, health, welfare, and vacation funds – NAICS 525920—Trusts, estates, and agency accounts – NAICS 814—Private households – NAICS 92—Public administration55 These exclusions should be noted when using this data with other data sources related to industry classification. Foreign Trade Statistics (FTS) • The export statistics contained in the FTS data source consist of goods valued at more than $2,500 per commodity shipped by individuals and organizations (including exporters, freight forwarders, and carriers) from the United States to other countries. Data users are advised to note the exclusion of goods valued at under $2,500 per commodity shipped. Commodity Flow Survey (CFS) and Freight Analysis Framework (FAF3) • The CFS does not contain data from the following freight-generating and freight-receiving industries, or it contains insufficient data to cover the industries in a comprehensive manner. – Multimodal truck, rail and pipeline flows of crude petroleum, petroleum products, and natural gas (NG) – Truck freight shipments associated with farm-based, fishery, logging, construction, retail, services, municipal solid waste, and household and business moves – Imported and exported goods transportation by ship, air, and Transborder land (truck, rail) modes • In FAF3 these industries produce what are called Out-Of-Scope to the CFS freight flows (OOS flows). Their estimation required a great deal of data collection and integration into the larger flow matrix generation process. For the most part, the data sources for these OOS flows are derived from freight-carrier-reported data sources. In some cases they require the use of secondary or indirect data sources, such as location-specific measures of industrial activity, employment, or population, to allocate flows to specific geographic regions. Developing OOS flow estimates represents a good deal of effort, with different commodity classes requiring very different, typically multi-step, treatments, including the use of both spatial and com- modity class “crosswalks” that convert mode-specific and industry-class-specific estimates from their native coding categories into FAF3 regional and commodity class breakdowns.56 6.6.1.2 Industry Definition Resolution Data sources containing industry classification data elements frequently use differing industry coding resolutions. These industry classifications include North American Industry Classifica- tion System (NAICS) or Harmonized System (HS) resolution. Analysts should be aware of the different resolutions and account for them during analysis.

Differences in Data Element Definitions 65 Annual Survey of Manufacturers • The Annual Survey of Manufacturers contains varying resolutions of NAICS (North Ameri- can Industry Classification System) industry classifications for different tables within the data source. Data users are advised to note the following differences in resolution between tables: – Statistics for Industry Groups and Industries—Total manufacturing establishments’ statis- tics are presented at the three-, four-, five-, and six-digit NAICS levels at the national level. – Value of Product Shipments—This file represents shipments statistics for the 471 six-digit NAICS product groups and approximately 1,384 seven-digit NAICS product classes at the national level. – Geographic Area Statistics—This file represents manufacturing establishments’ statistics at the three- and four-digit NAICS levels for each state. – Supplemental Statistics for the United States—This file represents supplemental manufac- turing establishments’ statistics at the two-digit NAICS levels for each state. County Business Patterns • The County Business Patterns data source contains varying resolutions of North Amer- ican Industry Classification System (NAICS) industry classifications for different tables within the data source. Data users are advised to note the following differences in resolution between tables: • County File—Provides data at the six-digit North American Industry Classification System (NAICS) industry code at the county level – State File—Provides data at the six-digit NAICS industry code at the state level – U.S. File—Provides data at the six-digit NAICS industry code at the national level – Metropolitan Area File—Provides data at the six-digit NAICS industry code at the metro- politan area level – Zip Code Industry Detail File—Provides data at the six-digit NAICS industry code at the zip code level – Commonwealth of Puerto Rico File—Provides data at the six-digit NAICS industry code for Puerto Rico and the Island Areas of American Samoa, Guam, the Commonwealth of the Northern Mariana Islands, and the Virgin Islands of the United States. Foreign Trade Statistics (FTS) • The Foreign Trade Statistics (FTS) data source contains varying resolutions of North Ameri- can Industry Classification System (NAICS) and Harmonized System (HS) industry classifi- cations for different tables within the data source. Data users are advised to note the following differences in resolution between tables: – USA Trade Online—This online subscription service provides U.S. export and import sta- tistics of industries at a high level of granularity, up to the 10-digit HS and six-digit NAICS classifications, by state, country, and Customs district. Data categories include the following: ▪ District Data (10-digit HS detail)—2003–current ▪ Port Data (six-digit HS detail)—2003–current ▪ State Export Data (six-digit HS detail and four-digit NAICS detail)—2002–current ▪ State Import Data (six-digit HS detail and four-digit NAICS detail)—2002–current ▪ NAICS Data (six-digit NAICS detail)—2002–current – U.S. Exports and Imports of Merchandise—This data source provides export and import statistics for industry-level HS commodities at the two-, four-, six-, and 10-digit levels.

66 Implementing the Freight Transportation Data Architecture: Data Element Dictionary Country and Customs district data for value and quantity are provided on a monthly, year-to-date, and annual basis. ▪ Merchandise Trade Exports—This data source offers multiple files (12 in all)—1989– current ▪ Merchandise Trade Imports—This data sources offers multiple files (12 in all)—1989– current – U.S. Exports and Imports of Merchandise—This data source provides 5 years of historical annual revised export and import statistics for industry-level HS commodities at the two-, four-, six-, and 10-digit levels. Commodity data for value and quantity are provided on an annual basis. ▪ Merchandise History Exports—This data source offers multiple files (12 in all)—1989– current ▪ Merchandise History Imports—This data source offers multiple files (12 in all)—1989– current – U.S. Exports and Imports by State—This data source provides export and import statistics by State of Origin of Movement (export) and State of Destination (import) for industry- level commodities at the six-digit HS level or the three- or four-digit NAICS level. Data is provided on a monthly, quarterly, or annual basis. ▪ State Exports—1987–current, and ▪ State Imports—This data release was discontinued in 1988 but reinstated in 2010; 2008– current data is now available.57 – U.S. Exports and Imports by Port—This data source provides export and import statistics by State of Origin of Movement for industry-level commodities at the six-digit HS level on a monthly, quarterly, or annual basis. Service Annual Survey • In the Definitions data table, the Service Annual Survey data source classifies businesses into categories using the six-digit North American Industry Classification System (NAICS) code. 6.7 Mode of Transport Data Elements Keywords: mode, transport, air, rail, pipeline, truck, waterway, vessel, vehicle, multimodal, intermodal, unknown In this section, differences in data element definitions are broken down by the various modes of transport. If applicable, taxonomic, temporal, and methodological differences across data sources are presented for each mode. The following modes of transport are addressed: • Air • Highway • Rail • Water • Pipeline • Multimodal/intermodal • Unknown/other 6.7.1 Air 6.7.1.1 Taxonomic Differences Data sources often use unique data elements to identify and differentiate mode of transport. Data users are advised to note these differences when using air mode data elements from multiple data sources.

Differences in Data Element Definitions 67 Air Carrier Statistics • T-100 Market (All Carriers) – This data table differentiates and reports freight class in four categories using letter codes: ▪ F—Scheduled passenger/cargo service ▪ G—Scheduled all cargo service ▪ L—Non-scheduled civilian passenger/cargo service ▪ P—Non-scheduled civilian all cargo service • T-100 Segment (All Carriers) – This data table differentiates mode by aircraft group, aircraft type, and aircraft configuration. ▪ AIRCRAFTGROUP—This data element includes codes such as:58  Piston—Single engine  Piston—2-engine  Helicopter/Stol  Turbo-prop 1- and 2-engine  Turbo-prop 4-engine  Jet ▪ AIRCRAFTTYPE—This data element includes codes such as:59  Cessna 180  Piper PA-32  Convair CV-340/440  McDonnell Douglas DC-6 ▪ AIRCRAFTCONFIG—This data element includes codes such as:60  Passenger configuration  Freight configuration  Combined passenger and freight on a main deck  Seaplane – The Air Carrier Statistics T-100 Segment data table also reports air service type, including service class, as follows: ▪ F—Scheduled passenger/cargo service ▪ G—Scheduled all cargo service ▪ L—Non-scheduled civilian passenger/cargo service ▪ P—Non-scheduled civilian all cargo service The full list of T-100 Segment codes can be found at the url provided in references 57–59 in the endnotes to Chapter 6 and 7. Fatal Analysis Reporting System (FARS) • FARS uses a data element called “Transported to Medical Facility By,” which reports details on travel mode using numerical codes. For example, the number 1 signifies “EMS air,” mean- ing transport by emergency medical services using air mode.61 • NHTSA updates the FARS Analytical User’s Manual62 every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. 6.7.1.2 Temporal Differences Temporal differences among and within air mode data sources fall under the following categories: • Changes in methodology over time • Data elements that account for temporal variation

68 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.7.1.2.1 Changes in Methodology over Time. Data sources may change their data col- lection or reporting methods over time, making it difficult to compare data elements across multiple years within a single data source, or across data sources. North American Transborder Freight Database (Transborder) • Beginning in 1997, the Bureau of Transportation Statistics (BTS) restructured the Transbor- der freight data files to simplify the table structure and improve usability of the data. Under the new reporting methodology, the DISAGMOT data element uses numerical codes to iden- tify mode of transport for shipments entering and exiting the United States. DISAGMOT 3 indicates air mode;63 however, DISAGMOT 8 (“Other and unknown”) includes “flyaway aircraft,” defined as aircraft moving under their own power (i.e., aircraft moving from the aircraft manufacturer to a customer and not carrying any freight). • With the release of the January 2004 data, BTS began incorporating vessel and air data pro- vided by the U.S. Census Bureau into the Transborder data. The vessel and air data provides information on U.S.-North American Transborder trade similar to U.S.-North American Transborder surface freight. Further reporting changes can be found in the Transborder Freight Data Documentation. Commodity Flow Survey (CFS) • The following methodological changes were incorporated into the CFS in 2012: – Shipments with a respondent-provided mode of “parcel” must weigh 150 pounds or less, while shipments with a respondent-provided mode of “air” are not given a weight restriction. – A shipment’s mode of transport is imputed whenever a respondent has provided a mode of “other” or “unknown,” or has failed to provide a modal response (coded as “missing mode”).64 6.7.1.2.2 Data Elements that Account for Temporal Variation. Data sources sometimes use multiple data elements to identify locations whose names or codes change over time. Cau- tion should be exercised to ensure that the correct identifier is used. Air Carrier Statistics • Over time the code or name of an air carrier may change, and the same code or name may be assumed by multiple airlines. To ensure that data users analyze data from the same airline, Air Carrier Statistics provides four airline-specific variables that identify unique carriers (airlines) or their associated entities: – AIRLINEID—Airline ID – UNIQUECARRIER—Unique carrier code – UNIQUECARRIERNAME—Unique carrier name – UNIQCARRIERENTITY—Unique carrier entity A unique carrier is defined as one holding and reporting under the same department of trans- portation certificate regardless of its code, name, or holding company/corporation.65 Notably, the Air Carrier Statistics data includes large certified carriers with annual operating revenues of $20 million or more.66 6.7.1.3 Methodological Differences When working with data elements related to the air mode of transport, data users should be aware that methodological differences exist not only among data sources but also within indi- vidual data sources.

Differences in Data Element Definitions 69 Commodity Flow Survey (CFS) • In the CFS, air mode shipments include shipments carried by truck to or from an airport. For multiple-mode shipments, if the respondent has reported a shipment’s mode of transport as both parcel and air, the CFS treats the shipment as parcel only. • The 2007 CFS classified air shipments as shipments weighing 100 pounds or more. During mileage processing for the 2007 CFS, an “air” shipment was manually converted to “parcel” if the weight of the shipment was less than 100 pounds. However, airlines do not necessar- ily have minimum weight restrictions when transporting cargo. Hence, for the 2012 CFS, the definition of an air shipment was changed. As a result, an air shipment was acceptable as provided by the respondent, regardless of weight. Furthermore, for the 2012 CFS, parcel shipments conformed to the definition used by the parcel industry that a parcel is a shipment of 150 pounds or less. For shipments submitted by the respondent with mode of Parcel and a weight above 150 pounds, GeoMiler changed the mode to For-Hire Truck during mileage processing.67 • In the case of imports and exports by air, domestic moves by ground to and from the port of entry or exit are categorized as shipments by truck.68 Foreign Trade Statistics (FTS) • The FTS data source presents transportation statistics in three categories—vessel, air, and “All Methods”—based on the method of transportation by which the merchandise arrived in or departed from the United States. Some shipments between the United States and other countries will enter or depart the United States through Canada or Mexico. Such shipments are recorded under the method of transportation by which they enter or depart the United States, regardless of the transportation mode between Canada or Mexico and the country of origin or destination.69 • Data reported on vessel, air exports, and general imports represents waterborne and airborne shipments only (i.e., merchandise leaving or arriving in the United States aboard a vessel or an aircraft). • Imports and exports of vessels moving under their own power or afloat, and aircraft flown into or out of the United States, are included in the “All Methods” data table but are excluded from the “Vessel and Air” statistics. • Mail and parcel post shipments (including those transported by vessel or air) are included in the “All Methods” data but excluded from the “Vessel and Air” statistics. Freight Analysis Framework (FAF3) • Because of a modification in the reporting of multimodal and intermodal categories between the 2002 and 2007 Commodity Flow Survey (CFS) on which the FAF is based, there is no direct equivalence in the modal classes implied between these two sets of definitions, with the exception of the truck-only and rail-only modes. • Air data includes any shipment sent via air mode to its destination. Data users are advised to note that air mode shipments include shipments carried by truck to or from an airport. For multiple-mode shipments, if the respondent reported a shipment’s mode of transport as both parcel and air, FAF3 treats the shipment as parcel only. • This data source includes shipments weighing more than 100 pounds that move by air, or a combination of truck and air, in commercial or private aircraft, including air freight and air express. The CFS/FAF3 does not include shipments weighing 100 pounds or less, which are typically classified as “multiple modes and mail.”70

70 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.7.2 Highway When performing data analysis, data users are advised to be aware of several temporal, taxonomical, and methodological differences associated with data elements related to high- way mode. 6.7.2.1 Temporal Differences Temporal differences among and within highway mode data sources occur because of changes in methodology over time. Data sources may change their data collection or reporting methods over time, making it difficult to compare data elements across multiple years within a single data source, or across data sources. Commodity Flow Survey (CFS) • For the 2012 CFS, a change was made relating to mileage processing. Private truck is now considered a short-haul mode (i.e., private trucks not routed more than 500 miles during shipment routing).71 Data users should be aware of this adjustment. Motor Carrier Management Information System (MCMIS) • Beginning in 1994, states participating in the Motor Carrier Safety Assistance Program were required to report through the SAFETYNET system a standard set of data items on all trucks and buses involved in traffic crashes that met a specific severity threshold. Reportable crashes include one or more of the following vehicle types: – A truck (used primarily for the transportation of property) having at least six tires in contact with the road surface – A vehicle displaying a hazardous material placard – A bus with seating for at least nine people (15 people before 2001), including the driver72 North American Transborder Freight Database (Transborder) • Beginning in January 1997, the Bureau of Transportation Statistics (BTS) restructured the Transborder freight data files to simplify the table structure and improve usability of the data. Land mode tables that were previously separate from air and vessel tables have been combined, and now all modes of transport are covered by the data element DISAGMOT. The DISAGMOT data element identifies mode of transport for shipments entering and exiting the United States using numerical codes. DISAGMOT 5 signifies truck mode; DISAGMOT 4, signifying mail mode, represents U.S. Postal Service and courier shipments, and cannot be further subdivided into a mode such as air, rail, or truck.73 • Before 1993, the U.S. Census Bureau provided mode of transport information only for air, water, and “Other.” No detail was available for surface trade. The current version of the Trans- border Freight Database (Transborder) makes freight transportation data available for all modes of transportation.74 Vehicle Inventory and Use Survey (VIUS) • Now discontinued, the VIUS was conducted every 5 years from 1963 until the final release in 2002. Although data releases are available for all of the surveys, public use microdata files are only available for years 1977 and later. Data users should also be aware that before 1997 the survey was known as the Truck Inventory and Use Survey (TIUS).75

Differences in Data Element Definitions 71 6.7.2.2 Taxonomic Differences Commodity Flow Survey (CFS) The CFS includes the following distinctions between private and for-hire trucks for the high- way mode, as follows: • Private Truck—This data element is defined as a truck operated by employees of the establishment or the buyer/receiver of the shipment, and includes trucks providing dedi- cated services to the establishment. Shipments via private truck are generally short-haul in nature.76 • For-Hire Truck—This data element is defined as a truck operated by common or contract carriers made under a negotiated rate. “For-hire truck” also is used if the shipment mileage was equal to or greater than 500 miles, regardless of the commodity being transported.77 Fatal Analysis Reporting System (FARS) • Transported to Medical Facility By—This data element reports ground travel using the num- ber 5, which signifies EMS ground (i.e., transport by emergency medical services using ground mode).78 • NHTSA updates the FARS Analytical User’s Manual79 every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. Freight Analysis Framework (FAF3) • The FAF3, which uses the same definitions as the Commodity Flow Survey (CFS), reports the highway mode using the number 1 (signifying truck mode), which includes both private and for-hire trucks. The truck mode does not include trucks that are part of “multiple modes and mail” (coded using the number 5), or truck moves in conjunction with domestic air cargo.80 Vehicle Inventory and Use Survey (VIUS) • BODYTYPE—This data element describes which body type a vehicle most closely resembles. This field contains 30 options for body type, such as concrete pumper, sport utility, street sweeper, and tow/wrecker. This level of classification is not used in other data sources.81 6.7.2.3 Methodological Differences Commodity Flow Survey (CFS) • For 2012 CFS mileage processing, if the shipment weighed less than 80,000 pounds, it was routed via highway mode as “for-hire truck”; if the shipment weighed 80,000 pounds or more, it was routed via rail mode. • The CFS does not report on highway shipments weighing 150 pounds or less, which are typi- cally classified as “multiple modes and mail.”82 North American Transborder Freight Database (Transborder) • For this data source, the mode of transport is recorded as the mode in use when the shipment enters or exits the United States. Therefore, if a shipment originates from Dallas, Texas, by

72 Implementing the Freight Transportation Data Architecture: Data Element Dictionary rail but transfers to truck in Austin, Texas, and arrives in the Port of Laredo to cross the U.S.– Mexico border by truck, mode of transport for that shipment is truck.83 • Before 2007, data by port and commodity detail were not available for download or analysis for the land modes. 6.7.3 Rail Several temporal, taxonomical, and methodological differences are associated with data ele- ments related to rail that data users should be aware of when performing data analysis. These differences are discussed in this section. 6.7.3.1 Taxonomic Differences Carload Waybill Sample • The Confidential Carload Waybill Sample contains the data element TYPE MOVE (Move- ment Type), which indicates whether the rail freight is imported, exported, imported and exported (e.g., land bridge traffic), neither, or unknown. • Both the Confidential Carload Waybill Sample and the Public Use Waybill Sample contain the data element TRANSIT CODE, which indicates whether goods were moved using “all rail,” “intermodal” (a continuous movement involving at least one railroad and another mode), or “unknown” mode. • In accordance with Accounting Rule 260, the Confidential Carload Waybill Sample uses mul- tiple data elements for interline transactions, including “origin railroad,” up to eight “bridge” railroads, and a “termination” railroad. This taxonomy is different from the Public Use Way- bill Sample, which does not contain this information. • For additional information, users can refer to Railway Accounting Rules, Association of American Railroads (September 2, 2012).84 Federal Railroad Administration (FRA) Safety Database • For accidents involving rail, the data element VEHICLE indicates whether automobiles, buses, trucks, motorcycles, bicycles, farm vehicles, and all other modes of surface transporta- tion were involved in an incident.85 6.7.3.2 Temporal Differences Temporal differences among and within data sources related to rail are a result of changes in methodology over time. Data sources may change their data collection methodology over time, making it difficult to compare data elements across multiple years within a single data source, or with other data sources. North American Transborder Freight Database (Transborder) • Before 1993, the U.S. Census Bureau provided mode of transport information only for air, water, and “Other.” No detail was available for surface trade. Currently, North American freight transportation data are available for all modes of transport, including rail. • Beginning in January 1997, land mode tables that were previously separate from air and ves- sel tables were combined so that all modes of transport were covered by the data element DISAGMOT. DISAGMOT identifies mode of transport for shipments entering and exiting the United States, using numerical codes. DISAGMOT 6 signifies rail mode. DISAGMOT 4,

Differences in Data Element Definitions 73 signifying mail mode, represents U.S. Postal Service and courier shipments, and cannot be further subdivided into individual modes such as air, rail, or truck.86 6.7.3.3 Methodological Differences Several methodological differences exist within individual data sources, as well as among data sources, that data users should be aware of when working with data elements related to the rail mode. Commodity Flow Survey (CFS) • For 2012 CFS mileage processing, if the shipment weighed 80,000 pounds or more, it was routed via rail mode; if the shipment weighed less than 80,000 pounds, it was routed via high- way mode as a for-hire truck. • Rail includes any common carrier or private railroad, regardless of the class. The CFS does not report on shipments weighing 150 pounds or less, which are typically classified with “multiple modes and mail.”87 Foreign Trade Statistics (FTS) • Transportation statistics are presented in terms of three categories—vessel, air, and “All Methods”—based on the method of transportation by which shipments arrived in or departed from the United States. Some shipments between the United States and other countries enter or depart the United States through Canada or Mexico. Such shipments are recorded under the method of transportation by which they enter or depart the United States, regardless of the transportation mode between Canada or Mexico and the country of origin or destination. Freight Analysis Framework (FAF3) • Because of the redefinition of multimodal and intermodal categories between the 2002 and 2007 Commodity Flow Survey (on which the FAF is based), there is no direct equivalence in the modal classes implied between these two sets of definitions, with the exception of the truck-only and rail-only modes. Appendix A88 of the FAF3 shows the modal class changes between 2002 and 2007 as well as definitions for the modes. • In the FAF3 data source, rail mode (coded using the number 2) includes any common carrier or private railroad. Shipments with multiple modes, including those with rail, are identified as “multiple modes and mail” (using the number 5).89 North American Transborder Freight Database (Transborder) • Before 2007, data by port and commodity detail were not available for download or analysis for the land modes, which includes rail. The “Mode of Transport Bridges” page provides a crosswalk from the three new tables, starting January 2007, to all the previous data tables before 2007.90 • For this data source, mode of transport is recorded as the mode in use when the shipment enters or exits the United States. For example, if a shipment originates from Dallas, Texas, by rail but transfers to truck in Austin, Texas, and arrives in the Port of Laredo to cross the U.S.–Mexico border by truck, the mode of transport for that shipment is “truck.”91 • Further reporting changes related to rail can be found in the Transborder Freight Data Documentation.92

74 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.7.4 Water 6.7.4.1 Taxonomic Differences Taxonomic and methodological differences among and within water mode data sources fall under the following categories: • Unique data elements • Inclusion/exclusion of data 6.7.4.1.1 Unique Data Elements. Data sources often use unique data elements to identify and differentiate attributes related to water modes of transport. Data users are advised to note these differences when performing analysis with data elements from multiple data sources. Carload Waybill Sample • The Confidential Carload Waybill Sample contains the data element TYPE OF MOVE VIA WATER, which provides data on water movement within the United States. This data element includes the following distinctions, coded by number: – 0 = Not a water movement – 1 = Ex-Lake (from Great Lakes to reporting railroad) – 2 = Lake Cargo (from rail to Great Lakes) – 3 = Intercoastal (a continuous movement by U.S. rail that is part of an Atlantic Ocean [or Gulf of Mexico] and Pacific Ocean movement, in either direction) – 4 = Coastwise (a continuous movement involving rail at either end of a coastwise move- ment between ports on the East Coast, including the Gulf of Mexico, or between ports on the West Coast) – 5 = Inland Waterways (a rail movement in combination with a barge movement on rivers and canals other than on the Great Lakes that is not considered a part of the rail movement [e.g., rail-car ferry]) – 9 = Unknown – Blank (not reported on hardcopy waybills)93 Commodity Flow Survey (CFS) • The CFS contains the data element MODE within the ORIGIN BY DESTINATION BY MODE data table. This data element reports the mode of transport used to move a shipment to its domestic destination. For water moves, the CFS includes the following mode categories: – Inland water—This data element is used to report vessels or barges operating primarily in navigable waters, both within and along the borders of the United States, including rivers, lakes, vessels moving along the shoreline but actually in the ocean (e.g., on the Intracoastal Waterway along the Atlantic and Gulf coasts, Inside Passage of Alaska), canals, harbors, major bays, and inlets. – Great Lakes—This data element is used for vessels or barges operating on the Great Lakes. – Deep sea—This data element is used for vessels or barges operating primarily in the open waters of the ocean, outside the borders of the United States. – Multiple waterways—This data element is used for shipments sent by any combination of Inland water, Great Lakes, and Deep sea, and which usually involve a transfer between vessels.94 North American Transborder Freight Database (Transborder) • The DISAGMOT data element uses numerical codes to identify mode of transport for shipments entering and exiting the United States. DISAGMOT 1 signifies “Vessel” (indicating water mode).95

Differences in Data Element Definitions 75 U.S. Waterway Data • The data element VTCC (Vessel Type, Construction, and Characteristics) contains a four- character alphanumeric code that describes in general terms the vessel type, construction, and characteristics of its use. For example, a VTCC code of 2A22 represents the code for a self-propelled tanker constructed of steel that is being used as a liquid bulk tanker. A full list of vessel types can be found in the Navigation Data Center User’s Guide under Appendix 4.96 6.7.4.1.2 Inclusion/Exclusion of Data. Certain data elements are included in some water data sources and excluded in others. Data users are advised to note these gaps in the data when using data sources. Foreign Trade Statistics (FTS) • Data for “Vessel and Air” exports and for general imports represent waterborne and airborne shipments only (i.e., merchandise actually leaving or arriving in the United States aboard a vessel or an aircraft). • Imports and exports moved by vessels moving under their own power or afloat and by aircraft flown into or out of the United States are included in the “All Methods” data but excluded from the “Vessel and Air” statistics. • Mail and parcel post shipments, including those transported by vessel or air, are included in the “All Methods” data, but are excluded from the “Vessel and Air” statistics. • Low-value shipments are included in the “All Methods” data but are excluded from the “Vessel and Air” statistics.97 Freight Analysis Framework (FAF3) • In the FAF, the water mode includes shallow draft, deep draft, Great Lakes, and intra-port shipments. Data users are advised to note that water mode data does not include shipments that are classified under “multiple modes and mail.”98 6.7.4.2 Methodological Differences Foreign Trade Statistics (FTS) • Statistics related to water modes are based on the method of transportation by which the mer- chandise arrived in or departed from the United States. Some shipments between the United States and other countries enter or depart the United States through Canada or Mexico. Such shipments are recorded under the method of transportation by which they enter or depart the United States regardless of the transportation mode between Canada or Mexico and the country of origin or destination.99 Freight Analysis Framework (FAF3) • Changes in the way the 2002 versus 2007 Commodity Flow Survey (CFS) assigned water-only versus water-inclusive intermodal shipments (typically, truck-water combinations) make direct comparisons of water-only traffic volumes and modal shares problematic. Appendix A of the FAF3 User Guide shows the modal class changes between 2002 and 2007 and provides definitions for the modes.100

76 Implementing the Freight Transportation Data Architecture: Data Element Dictionary North American Transborder Freight Database (Transborder) • With the release of January 2004 statistics, the Bureau of Transportation Statistics (BTS) began incorporating vessel data provided by the U.S. Census Bureau into the Transborder data. The vessel data provided information on U.S.–North American transborder trade simi- lar to U.S.–North American transborder surface freight. Thus, for the first time, additional information such as U.S.–North American transborder trade by port and commodity became available. Further reporting changes related to vessel data can be found in the Transborder Freight Data Documentation.101 6.7.5 Pipeline 6.7.5.1 Temporal Differences Temporal differences among and within pipeline mode data sources occur as a result of changes in methodology over time. North American Transborder Freight Database (Transborder) • Before 1993, the U.S. Census Bureau only provided mode of transport information for air, water, and “Other.” No detail was available for surface trade. Since 1993, however, North American freight transportation data has been made available for all modes of transportation, including pipelines. • Beginning in January 1997, the Bureau of Transportation Statistics (BTS) restructured the Transborder freight data files to simplify the table structure and improve usability of the data. Land mode tables that had previously been separate from the air and vessel tables were combined, and now all modes of transportation are covered by the data element DISAGMOT. DISAGMOT uses numerical codes to identify mode of transport for shipments entering and exiting the United States. For example, DISAGMOT 7 signifies pipeline mode.102 6.7.5.2 Methodological Differences Commodity Flow Survey (CFS) • Pipeline data in the CFS includes movements of oil, petroleum, gas, slurry, and so forth through pipelines that extend to other establishments or locations beyond the shipper’s estab- lishment; however, aqueducts for the movement of water are not included.103 Freight Analysis Framework (FAF3) • The FAF3 definition of pipeline mode (coded using the number 6) includes crude petro- leum, natural gas (NG), and product pipelines. Data users are advised to note that products shipped via pipeline include flows from offshore wells to land which USACE counts as water moves. Pipelines that are part of “multiple modes and mail” are not included in the FAF3 pipeline data.104 6.7.6 Multimodal/Intermodal 6.7.6.1 Taxonomic Differences Taxonomical differences among and within multimodal mode data sources involve the inclu- sion or exclusion of data in element definitions.

Differences in Data Element Definitions 77 Commodity Flow Survey (CFS) • The CFS does not report on shipments weighing 150 pounds or less, which are typically clas- sified under “multiple modes and mail.”105 Freight Analysis Framework (FAF3) • Multiple Modes and Mail—This value includes shipments by multiple modes and by parcel delivery services, U.S. Postal Service, or couriers. This category is not limited to trailer-on- flat-car or container-on-flat-car (TOFC/COFC) shipments.106 Intermodal Terminals Database • Because an intermodal terminal may connect more than one pair of modes or transfer more than one type of cargo between one or more pairs of modes, it may have multiple records in the intermodal connections file.107 6.7.6.2 Temporal Differences Temporal differences among and within data sources related to multimodal freight move- ment are a result of changes in methodology over time. Carload Waybill Sample • To provide more complete data, in 1994 Railinc began flagging privately owned intermodal units; however, although reporting of private intermodal units has since increased, many of these units are still not reported in the Universal Machine Equipment Register (UMLER) computer platform.108 • Railinc’s UMLER database is used by railroads, rolling stock owners, and repair shops to share a wealth of rail-car information, which is used to interchange cars, pool traffic, and issue blocking requests.109 • To reduce the possibility of confusion, the UMLER database maintains only the most recent car initial/number/type assignments for TTX equipment. (The TTX Company assigns car initials and car type by car number and, based on need, frequently and repeatedly reassigns series of car numbers to different initials and car types. The original number assignment usually refers to intermodal flatcars, but subsequent assignments have often related to multi-level flatcars.) • Because the UMLER locates flatcars by comparing the car number with its assigned car initial and car type, reassignment of series numbers can complicate data analysis and lead to report- ing errors in the edited database. For example, the car initial and car type currently assigned to a particular car number are written onto edited waybill records. An error flag, “14” will be appended to the record in UMLER if the car type no longer corresponds to certain codes (P, Q, or S). In many cases, however, at the time of the waybill movement, the car number was most likely assigned to a different car initial and to car type P, Q, or S. • To reduce the number of waybill errors generated by this issue, intermodal waybills processed after September 1, 1995, have used the dummy car initial number GBRX 091193 in instances of traditional trailer-on-flat-car/container-on-flat-car (TOFC/COFC) movements.110 Intermodal Terminals Database • The amount of effort required to keep the database current depends on the volatility of the data, which in turn depends on how much and how fast the intermodal infrastructure is

78 Implementing the Freight Transportation Data Architecture: Data Element Dictionary changing. Changes to the intermodal infrastructure are being driven by the continued growth of containerized traffic, recent and proposed railroad mergers, the formation of ocean carrier alliances, technological advances, freight rate incentives, the availability of federal funding for intermodal projects, and many other events and factors. Although the number of intermodal terminals and intermodal connections added to or removed from the transportation system over the course of a year may be relatively small compared to the number of terminals in existence, it is nevertheless significant.111 6.7.6.3 Methodological Differences Freight Analysis Framework (FAF3) • Differences in the way the 2002 versus the 2007 Commodity Flow Survey (CFS) assigned water-only versus water-inclusive intermodal shipments (typically, truck-water combina- tions) make direct comparisons of water-only traffic volumes and modal shares problematic. Appendix A of the FAF3 shows the modal class changes between 2002 and 2007 and provides definitions for the modes.112 • For multiple-mode shipments, if a respondent has reported a shipment’s mode of transport as both parcel and air, CFS treats the shipment as parcel only. Vehicle Inventory and Use Survey (VIUS) • The 2002 VIUS dropped the intermodal question (railroad, maritime, or domestic contain- ers; piggyback trailers; or conventional trailers). The U.S. Census Bureau had requested that questions be considered for deletion to make room for the questions being added to the 2002 VIUS, and data users agreed that this question was either of limited use or the quality was questionable.113 6.7.7 Unknown/Other 6.7.7.1 Taxonomic Differences Taxonomical differences among and within unknown/other mode data sources are a result of unique items included in the data element definitions. Carload Waybill Sample • For the data element corresponding to the TOFC/COFC service code, the code for the Inter- modal Service Code (ISC) must be entered in the first position of the field. Three blanks in this field indicate that the movement is not intermodal in nature. Unknown ISCs are indi- cated by an X. • For the data element corresponding to “All Rail/Intermodal,” the number 9 indicates an unknown mode. • For the data element corresponding to “Type of Move Via Water,” the number 9 indicates an unknown mode.114 Fatal Analysis Reporting System (FARS) • FARS uses a data element called “Transported to Medical Facility By,” which reports details on travel to a medical facility via unknown modes using the following number codes: – 3 = EMS (emergency medical services), unknown mode – 4 = Transported by unknown sources – 9 = Unknown mode.

Differences in Data Element Definitions 79 • Other mode travel is reported using the code “6—Other.”115 • NHTSA updates the FARS Analytical User’s Manual116 every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. Freight Analysis Framework (FAF3) • The data element corresponding to “Other and Unknown” includes movements not elsewhere classified, such as flyaway aircraft and shipments for which the mode cannot be determined.117 North American Transborder Freight Database (Transborder) • DISAGMOT uses numerical fields to identify the surface mode or other mode of transport of shipments entering or exiting the United States. DISAGMOT 8 (“Other and unknown”), includes “flyaway aircraft, or aircraft moving under their own power (i.e., aircraft moving from the aircraft manufacturer to a customer and not carrying any freight), powerhouse (elec- tricity), vessels moving under their own power, pedestrians carrying freight, unknown, and miscellaneous other.”118 6.7.7.2 Temporal Differences Temporal differences among and within data sources related to unknown/other freight move- ment are a result of changes in methodology over time. Data sources may change their data collection or reporting methods over time, making it difficult to compare data elements across multiple years within a single data source, or across data sources. Commodity Flow Survey (CFS) • For the 2012 CFS, a change was made relating to mileage processing. Mode of transportation is now imputed whenever a respondent has provided a mode of “other,” or “unknown,” or otherwise failed to provide a modal response (“missing mode”) for a shipment. • During the 2007 CFS mileage processing, 2.4% of shipments had a respondent-provided mode of “unknown” or “other,” and an additional 2.1% had no reported mode at all. Since all shipments must be properly routed to calculate a distance traveled, imputations were made. For 2012 CFS mileage processing, if the shipment weighed less than 80,000 pounds, it was routed via highway mode as a for-hire truck; if the shipment weighed 80,000 pounds or more, it was routed via rail mode.119 Fatal Analysis Reporting System (FARS) • FARS uses a data element known as “Transported to Medical Facility By,” which reported unknown mode travel under different codes before 2010. Data users should be aware of the changes and consult the appropriate FARS analytical reference guide for the proper codes.120 North American Transborder Freight Database (Transborder) • DISAGMOT (the data element corresponding to mode of transport) uses number codes to identify the mode of transport of shipments entering or exiting the United States. Since April 1995, in response to inquiries from data users and a U.S. Census Bureau investigation, the Transborder database added DISAGMOT 9, signifying foreign trade zones, as a mode of transport. Data users are advised to keep in mind two things: (1) before April 1995, such

80 Implementing the Freight Transportation Data Architecture: Data Element Dictionary imports were included under DISAGMOT 8 (“Other and unknown”); and (2) The actual mode of transportation is not available for imports coded under “foreign trade zones” using DISAGMOT 9. Although foreign trade zones are treated as a mode of transportation in this dataset, the actual mode for a specific shipment into or out of a foreign trade zone remains unknown because U.S. Customs and Border Protection (CBP) does not collect this information. 6.8 Safety Data Elements Keywords: safety, risk, accident, incident, crash, collision, fatality, injury, property damage, hazardous material, driver, vehicle unit, first harmful event 6.8.1 Taxonomic Differences The definitions of “fatal crash” and “fatality” were found to be consistent in the FARS, MCSMS, and FRA Safety Database; no taxonomic differences were identified. Fatal Analysis Reporting System (FARS)121 • A fatal crash is a crash that involves a motor vehicle traveling on a trafficway customarily open to the public, and results in the death of an occupant of a vehicle or a non-occupant within 30 days (720 hours) of the crash. • NHTSA updates the FARS Analytical User’s Manual122 every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. Motor Carrier Safety Measurement Systems (MCSMS)123 • A fatality is any person killed in or outside of any vehicle (e.g., truck, bus, car) involved in a crash or who dies within 30 days of a crash as a result of an injury sustained in the crash. Federal Railroad Administration (FRA) Safety Database124 • A fatality is defined as an individual who is confirmed dead within 30 days of a rail-transit- related incident. 6.8.2 Temporal Differences Temporal differences among and within safety-related data elements are a result of changes in definitions and codes over time. Data sources sometimes change the data element defini- tions and codes over time to accommodate changes in the type of data collected and the way the data is presented. Caution should be exercised to ensure that the correct definitions and codes are used. Fatal Analysis Reporting System (FARS)125 • FARS, which became operational in 1975, is a nationwide census providing the National Highway Traffic Safety Administration (NHTSA), Congress, and the American public yearly data regarding fatal injuries suffered in motor vehicle traffic crashes. A comprehensive cod- ing manual has been produced each year. In addition, NHTSA updates the FARS Analytical User’s Manual126 every year to summarize the evolution of coding. When conducting analysis

Differences in Data Element Definitions 81 across years, data users should check every data element of interest in each year’s coding manual. • P22/NM21—This data element identifies the method of transportation provided to transport a person to a hospital or medical facility. Although this field exists in the 1975 and 1976 files, it is not initialized (i.e., it has no values in those years). This variable was expanded to include non-motorists in 2010. • HAZ_CARG—From 1982 to 2006, this data element was used to identify the presence of hazardous cargo for a vehicle and to record information about the hazardous cargo when available. Since 2007, however, HAZ_CARG has been replaced with the following five data elements: – HAZ_INV—This data element identifies whether the vehicle was carrying hazardous materials. – HAZ_PLAC—This data element identifies the presence of hazardous materials for the vehicle and whether the vehicle displayed a hazardous materials placard. – HAZ_ID—This data element identifies the four-digit hazardous material identification number for the vehicle. – HAZ_CNO—This data element identifies the single-digit hazardous material class number for the vehicle. – HAZ_REL—This data element identifies whether any hazardous cargo was released from the cargo tank or compartment of the vehicle. • Data users should be cautious about changes in attribute codes over time. For instance, the data elements BODY_TYP and TOW_VEH define vehicle categories such as passenger cars, pickups, buses, trucks, and so forth. These fields help differentiate freight-related fatal crash records from fatal crash records of other motor vehicles types. Table 6-1 summarizes tempo- ral differences in truck-related codes. • For additional examples, please see Appendix C of the FARS Analytical User’s Manual,127 which tabulates changes made to all FARS data elements since 1975. 6.8.3 Methodological/Reporting Differences Methodological/reporting differences among and within safety-related data elements fall under the following categories: • Methodological differences within a data source • Reporting differences within a data source Classification (BODY_TYP) Data Year and Code 1975–1981 1982–1990 1991–Later Pickups 50 50, 51 30-39 Large Trucks 53-59, or (60 and tow_veh=1) 70-72, 74-76, 78, or (79 and tow_veh in 1-5) 60-64, 66, 67, 71, 72, 78, or (79 and tow_veh in 1-4) Light Trucks & Vans 43, 50-52, or (60 and tow_veh=0) 12, 40, 41, 48-51, 53-56, 58, 59, 68, 69, or (79 and tow_veh=0 or 9) 14-22, 24, 25, 28-41, 45-49, or (79 and tow_veh =0 or 9) Medium Trucks 53, 54, 56 70, 71, 75, 78 60-62, 64, 67, 71 Heavy Trucks 55, 57-59, or (60 and tow_veh=1) 72, 74, 76, or (79 and tow_veh in 1-5) 63, 66, 72, 78, or (79 and tow_veh in 1-4) Combina€on Trucks ((53-56, 60) and tow_veh=1), or 57-59 ((70-72, 75, 76, 78, 79) and tow_veh in 1-5), or 74 ((60-64, 71, 72, 78, 79) and tow_veh in 1-4), or 66 Single-Unit Trucks (53-56, 60) and tow_veh =0 (70-72, 75, 76, 78, 79) and tow_veh in (0,9) (60-62, 63, 64, 67, 71, 72, 78, 79) and tow_veh in (0,5,6,9) Table 6-1. NHTSA’s vehicle body type classification.

82 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.8.3.1 Methodological Differences Within a Data Source Data users should be aware of methodological differences within a single data source that make certain types of analysis difficult. Motor Carrier Safety Measurement Systems (MCSMS)128,129 • The MCSMS methodology is frequently updated by the Federal Motor Carrier Safety Admin- istration (FMCSA) to include the most current set of violations being recorded from inspec- tions. The original MCSMS methodology was developed based on the SafeStat measurement system. In January 2008, FMCSA started an Operational Model Test of the Compliance, Safety, and Accountability program. Notable milestones of the methodology changes in the Carrier Safety Measurement System (CSMS) portion of the MCSMS (as opposed to the Driver Safety Measurement Systems) are as follows: – CSMS130 Methodology Changes from Version 1.2 to 2.0 (Implemented August 2010) – CSMS Methodology Changes from Version 2.0 to 2.1 (Implemented December 2010) – CSMS Methodology Changes from Version 2.1 to 2. 2 (Implemented January 2012) – CSMS Methodology Changes from Version 2.2 to 2.2.1 (Implemented August 2012) – CSMS Methodology Changes from Version 2.2 to 3.0 (Implemented December 2012) – CSMS Methodology Changes from Version 3.0 to 3.0.1 (Implemented August 2013) – CSMS Methodology Changes from Version 3.0.1 to 3.0.2 (Implemented June 2014) 6.8.3.2 Reporting Differences Within a Data Source For some data sources, reporting mechanisms might change over time. Caution should be exercised when using or interpreting data in certain types of safety analyses. Pipeline and Hazardous Material Safety Administration (PHMSA) • The accident reporting criteria for hazardous liquid pipeline systems were revised in 1990, 1991, 1994, 1996, and 2002. For example, beginning in 1991, a release of carbon dioxide (50 or more barrels) was added as a type of hazardous materials (hazmat) accident. In addi- tion, incident reporting criteria for gas transmission, gas gathering, and gas distribution pipeline systems were revised in 1990 and updated in 2011. For more information on these and other methodological changes, see the PHMSA Reporting Criteria Changes—1990– Current.131 • Beginning in 2005, incident reporting criteria were modified to include the discovery of un- declared hazmat. It is reported that these types of incidents consist of approximately 8% of total reported incidents, although about half of them do not indicate a release of hazmat or any other criteria for incident reporting. More information on data quality assessment can be found in the PHMSA publication, A Data Quality Assessment: Evaluating the major safety data programs for pipeline and hazardous materials safety (November 10, 2009).132 Federal Railroad Administration (FRA) Safety Database • The reporting threshold for the Rail Equipment Accident/Incident Reporting Threshold table is updated annually. Starting with $750 for data released between 1957 and 1974, the reporting threshold increased to $6,700 for data released 2002–2005 and to $10,500 in 2014.133 • In response to changes associated with the Occupational Safety and Health Act, FRA amended its accident/incident reporting rules so that the data on occupational fatalities, injuries, and illnesses in the railroad industry is comparable with such data for other industries The changes were implemented beginning May 1, 2003.134,135

Differences in Data Element Definitions 83 Motor Carrier Management Information System (MCMIS)136 • Beginning January 1, 1994, states participating in the Motor Carrier Safety Assistance Pro- gram were required to report through the SAFETYNET system a standard set of data items on all trucks and buses involved in traffic crashes that met a specific severity threshold. Report- able crashes include one or more of the following vehicle types: – A truck (used primarily for the transportation of property) having at least six tires in contact with the road surface – A vehicle displaying a hazardous material placard – A bus with seating for at least nine people (15 people before 2001), including the driver • The Federal Motor Carrier Safety Administration (FMCSA) uses data from both the Fatal Analysis Reporting System (FARS) and MCMIS. The two databases may report different fatal crash counts because of variations in the way they define reportable vehicle configurations. FMCSA provides the FARS/MCMIS Fatal Crash Record Matching Tool to help reconcile dif- ferences between the FARS and MCMIS databases. The tool matches fatal large truck and bus crash records between the databases by comparing several key fields (e.g., county, date, time, VIN, DOT #) of large truck or bus fatal crash records.137,138 6.9 Units of Measurement Data Elements Keywords: length, width, volume, depth, height, capacity, distance, monetary, passengers, time, weight Definitions of these commonly used units of measurement vary among freight data sources: • Distance • Monetary data • Passenger movements • Time • Volume: Traffic • Volume: Water/vessels • Weight • Geospatial data Applicable taxonomic, temporal, and methodological differences identified as part of NCFRP Project 47 are detailed in the following sections. 6.9.1 Distance 6.9.1.1 Taxonomic Differences Air Carrier Statistics • DISTANCE GROUP—This data element measures the distance of a flight segment in 500-mile increments using code numbers 1–17. 139 Federal Railroad Administration (FRA) Safety Database The FRA Safety Database reports distance using several distinct data elements. Data users should be aware of the differences to ensure that the correct distance measure is being used for analysis.140 • LOCOMI—This data element reports the number of locomotive train-miles traveled in the month. A train-mile is defined as the movement of a train for a distance of 1 mile. Data users

84 Implementing the Freight Transportation Data Architecture: Data Element Dictionary should note that the presence of multiple locomotives in the train does not affect the mileage calculation. • MTMI—This data element reports the number of motor train-miles for the month. • YSMI—This data element reports the number of yard-switching train-miles for the month, which represents the miles traveled while the train is engaged in yard-switching service. • TOTMI—This data element indicates the total miles as reported on Form FRA F6180.55, Railroad Injury and Illness Summary. • PASSMI—This data element reports the number of passenger-miles for the month. A passenger- mile is defined as the movement of a passenger for a distance of 1 mile. • FRTRNMI—This data element reports the number of train-miles in freight service during the month. • PASTRNMI—This data element reports the number of train-miles in passenger service during the month, defined as the movement of a passenger for a distance of 1 mile. • OTHERMI—This data element reports any other train-miles not included in freight, passenger, or yard-switching train-miles. Vehicle Inventory and Use Survey (VIUS) • MILES_ANNL—This data element reports the number of miles a vehicle was driven in the reporting year without adjusting for partial-year ownership of the vehicle. Data users should be aware that this data element may reflect additional miles traveled when vehicles were not owned by the respondents. • MILES_ANNLNOIMP—This data element reports the number of miles a vehicle was driven in the reporting year as adjusted for partial-year ownership of the vehicle. • TAB_MILES—This data element indicates the weighted annual truck-miles driven during 2002 after applying the expansion factor for trucks (the TAB_TRUCKS data element).141 6.9.1.2 Methodological Differences Federal Railroad Administration (FRA) Safety Database • YSMI—This data element reports the number of yard-switching train-miles. The FRA Guide for Preparing Accident/Incident Reports advises that, if actual mileage is not known, YSMI can be computed at the rate of 6 mph for the time actually engaged in yard-switching service.142 • FRTRNMI—This data element reports the number of freight train-miles run by a railroad on its own track during the month. Data users should be aware that FRTRNMI reports freight train-miles by railroad, not by track; it does not aggregate train-miles reported by the railroad that owns the track together with train-miles that may be reported by another railroad, which may occur if one railroad’s equipment is being operated over the track by a different railroad’s crew. In such cases, the railroad of the crew operating the equipment enters the freight train- miles on their own FRA form.143 6.9.2 Monetary Data 6.9.2.1 Taxonomic Differences Air Carrier Financial Report • Beginning on October 18, 2006, numbers reported in the Schedule B-1, B-1.1, P-1.1, and P-1.2 data tables began following the format of common public financial documents, such as reports filed with the Securities and Exchange Commission or company financial statements. This for- mat reverses signs from the accounting format in which numbers appeared before that date.144

Differences in Data Element Definitions 85 6.9.2.2 Methodological Differences Carload Waybill Sample • The Surface Transportation Board (STB) classifies railroads based on their annual operating revenues as either Class I ($250 million or more), Class II ($20 million or more), or Class III ($0–$20 million). The average index (deflator factor) is based on the annual average Railroad Freight Price Index for all commodities. The formula below is used to adjust a railroad’s oper- ating revenues to eliminate the effects of inflation. Current Year’s Revenues × (1991 Avg. Index / Current Year’s Avg. Index) • EXPANDED TOTAL REVENUE—This data element indicates the total freight revenue (item 15 in the STB Reference Guide145) multiplied by the expansion factor (item 88). Rev- enue splits are calculated by dividing the waybill’s expanded freight revenue figure by the number of 100-mile blocks traveled by each railroad in the route. The origin railroad is apportioned revenue for an additional block to allow for pickup and switching expenses. Likewise, the termination railroad is credited with revenue for an additional block, to allow for delivery expenses.146 • TOTAL VARIABLE COST—This data element indicates the expanded variable cost for all rail- roads in the waybill computed using the Uniform Railroad Costing System (URCS). The URCS produces average variable costs for Class I railroads using railroad-specific accounting and oper- ating data. Costs for local and regional railroads use URCS regional data. See the STB Reference Guide for more details on the methodology used to calculate TOTAL VARIABLE COST.147 Commodity Flow Survey (CFS) • VALUE (MILLION $)—This data element reports the dollar value, in millions of dollars, of the entire shipment. This is defined as the net selling value, exclusive of freight charges and excise taxes. Data users are advised to note that the total value of shipments as measured by the CFS and the U.S. gross domestic product (GDP) provide different measures of economic activity in the United States and are not directly comparable. GDP is the value of all goods produced and services performed by labor and capital located in the United States. As mea- sured by the CFS, the value of shipments is the market value of goods shipped from manufac- turing, mining, wholesale, and select retail and service establishments, as well as warehouses and managing offices of multiunit establishments.148 Table 6-2 highlights three important differences between GDP and CFS value of shipments. GDP CFS Captures goods produced by all establishments located in the United States Measures goods shipped from a subset of all goods-producing establishments Measures the value of goods produced and of services performed Measures the value of goods shipped Counts for only the value added at each step in the producon of a product Captures the value of shipments of materials used to produce or manufacture a product, as well as the value of shipments of the finished product itself* *This means that the value of the materials used to produce a particular product contributes multiple times to the value of the commodity in the CFS. Table 6-2. Differences between GDP and CFS value of shipments.

86 Implementing the Freight Transportation Data Architecture: Data Element Dictionary North American Transborder Freight Data (Transborder) • Although Transborder contains data on exports to and imports from Canada and Mexico, all data elements that report monetary information (e.g., FREIGHT, VALUE) are reported in U.S. dollars.149 6.9.3 Passenger Movements 6.9.3.1 Methodological Differences Air Carrier Statistics • In the T-100 Market Data, a passenger is “enplaned” and is counted only once as long as he or she remains on the same flight. In the T-100 Segment Airline Traffic Data, a passenger is “transported” and is counted for each leg of the trip. Therefore, the num- bers in the segment data will tend to be higher than those in the market data (except for international flights). The Bureau of Transportation Statistics (BTS) generally uses market data for passenger, freight, or mail totals, as shown in this example provided by the U.S. DOT:150 For example, 250 people take a flight from JFK (Point A) to BWI (Point B), where 200 passen- gers deplane and the other 50 passengers, along with 70 additional passengers, continue on to MIA (Point C), where all passengers deplane. In [the market dataset, (T-100 Market Data)], Point A to Point B would be counted as one market of 200 and Point A to Point C would be counted as another market of 50. Point B to Point C would be a market of 70 people. In [the segment dataset (T-100 Segment Airline Traffic Data)], Point A to Point B would be counted as one segment of 250 and Point B to Point C would be counted as another segment of 120. A passen- ger from A to B to C would be counted for both legs. A to B: 200 Market, 250 Segment B to C: 70 Market, 120 Segment A to C: 50 Market (no Segment)151 Border Crossing/Entry Data • A passenger is defined as a person entering the United States at a particular port in a privately owned vehicle, pickup truck, motorcycle, recreational vehicle, taxi, ambulance, hearse, trac- tor, snowmobile, or other motorized private ground vehicle. • A pedestrian is a person arriving on foot or by certain conveyance (such as a bicycle, moped, or wheelchair) requiring U.S. Customs processing.152 6.9.4 Time 6.9.4.1 Taxonomic Differences Air Carrier Statistics • Air Carrier Statistics reports quarterly data using specific timeframes; however, as shown in Table 6-3, calendar quarters as defined by Air Carrier Statistics may differ from calendar quarters as defined by other data sources, including the U.S. government:

Differences in Data Element Definitions 87 • To ensure that they are using the desired unit of measurement for analysis, data users need to recognize the difference between the data elements RAMPTORAMP (or RAMPTIME) and AIRTIME. – RAMPTORAMP (or RAMPTIME)—This data element reports the time computed from the moment the aircraft first moves under its own power for purposes of flight until it comes to rest at the next point of landing.155 – AIRTIME—This data element, on the other hand, reports the airborne hours of the air- craft, computed from the moment it leaves the ground until it touches the ground at the end of a flight stage.156 Carload Waybill Sample • Some data elements in the Carload Waybill Sample report dates using different codes, which could create difficulty in making direct joins with data element within the data source, as well as with data elements from other data sources. The data element WAYBILL DATE, for example, uses the data coding system mmddccyy (month, day, century, year), while the data element DEREGULATION DATE uses the coding system ccyymmdd.157 6.9.4.2 Temporal Differences Fatal Analysis Reporting System (FARS) • The 2010 FARS incorporated many changes, most of which resulted from efforts by the National Highway Traffic Safety Administration (NHTSA) to standardize variables in FARS and the National Automotive Sampling System’s General Estimates System (GES).158 Three substantial changes regarding FARS data elements related to time were: – CRASH DATE—This data element added GES element information, including new GES Special Instructions. The new reporting system removed Attribute 98, “Not Reported for Both Month and Day.” – CRASH TIME—This data element added GES element information, including new GES Special Instructions. The new reporting system removed Attribute 9988, “Not Reported.” – DEATH_TM—This data element records the hour and minute of a person’s death using a four-digit coding system and the 24-hour clock format. Data from 1975 to 2008, however, followed a slightly different reporting format than did data from 2009 and later, as shown in Table 6-4: • NHTSA updates the FARS Analytical User’s Manual159 every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. Table 6-3. Calendar quarter definitions. Air Carrier Statistics153 U.S. Government154 Q1 January 1–March 31 October 1–December 31 Q2 April 1–June 30 January 1–March 31 Q3 July 1–September 30 April 1–June 30 Q4 October 1–December 31 July 1–September 30

88 Implementing the Freight Transportation Data Architecture: Data Element Dictionary 6.9.5 Volume: Traffic 6.9.5.1 Taxonomic Differences Freight Analysis Framework (FAF3) • Data users should be aware that local truck traffic that is not part of FAF3.1 (FAF version 3.1) truck estimates is provided under two data elements: – NONFAF07 This data element is used for current traffic. – NONFAF40 This data element is used for forecast traffic.160 Highway Performance Monitoring System • AADT_SINGLE_UNIT—This data element represents the annual average daily traffic (AADT) for single-unit trucks and buses, which are defined as vehicle classes 4 through 7 (buses through single-unit trucks with four or more axles). • AADT_COMBINATION—This data element represents the AADT volume for combination unit trucks. Combination trucks are defined as vehicle classes 8 through 13 (single-trailer trucks with four or fewer axles through multi-trailer trucks with seven or more axles).161 6.9.5.2 Temporal Differences Freight Analysis Framework (FAF3) • FAF3 uses 2008 Highway Performance Monitoring System (HPMS) data to determine annual average daily traffic (AADT) for the year 2007. Data users are advised to note the temporal difference between the HPMS data and the FAF3 reporting year.162 Highway Performance Monitoring System (HPMS) • FUTURE_AADT—This data element represents a 20-year forecast annual average daily traf- fic (AADT), which may cover a period of 18 to 25 years from the year of the data submittal.163 6.9.5.3 Methodological Differences Highway Performance Monitoring System (HPMS) • Data users are advised to note that the annual average daily traffic (AADT) for the National Highway System, Interstate, Principal Arterials (OFE, OPA), and HPMS Sample Panel sec- tions are typically based on traffic counts taken on a minimum 3-year cycle, while AADT for the Non-Principal Arterial System and Non-Sample Panel sections are typically based on a minimum 6-year counting cycle.164 • HPMS guidance requires that growth factors be applied if the AADT is not derived from cur- rent year counts. For specific guidance on factor development recommended for HPMS data, see the Traffic Monitoring Guide.165 1975–2008 2009 and Later Midnight 2400 0000 Time of Death (hhmm Format) 0001–2359 0001–2359 Not Applicable (Non-Fatal) -- 8888 Unknown 9999 9999 Table 6-4. Changes in format of DEATH_TM records.

Differences in Data Element Definitions 89 • HPMS requires that vehicle classification counts be adjusted to represent average conditions as recommended in the FHWA’s Traffic Monitoring Guide; see that guide for specific guidance on count adjustments used in the HPMS.166,167,168 6.9.6 Volume: Water/Vessels 6.9.6.1 Taxonomic Differences U.S. Waterway Data • ACTUALCY—This data element reports the actual cubic yards dredged. • NRT—This data element reports vessel net tonnage, defined as the volume of space available for the accommodation of passengers and the stowage of cargo, expressed in units of 100 cubic feet for each net ton. Data users are advised to note the difference between NRT and tonnage capacity, which simply expresses a volume capacity for passengers and cargo. For a more detailed discussion of how to calculate vessel net tonnage, see the 2012 Waterborne Transportation Lines of the United States.169 6.9.7 Weight 6.9.7.1 Taxonomic Differences Foreign Trade Statistics (FTS) • AIR_SWT_MO—This data element, representing air shipping weight, reports the gross weight in kilograms of shipments made by air, including the weight of moisture content, wrappings, crates, boxes, and containers (other than cargo vans and similar substantial outer containers).170 North American Transborder Freight Data (Transborder) • SHIPWT—This data element, representing shipping weight, reports the gross weight of ship- ments of imports (and some exports) in kilograms, including the weight of moisture content, wrappings, crates, boxes, and containers (other than cargo vans and similar substantial outer containers). SHIPWT does not include data for exports shipped by land modes of transpor- tation and reported using paper Shipper’s Export Declarations documents; however, export weight (SHIPWT) is required to be filed for all modes of transportation using the Automated Export System. Vehicle Travel Information System Documentation • TOTAL WEIGHT OF VEHICLE—This data element reports the gross vehicle weight to the nearest tenth of a metric ton (100 kilograms). Data users are advised to note that this measure- ment differs from measurements based on a short ton (2,000 pounds), which are often used by other similar data sources.171 Vehicle Inventory and Use Survey (VIUS) • WEIGHT_SIZE—This data element reports the average weight of the vehicle or vehicle/ trailer combination grouped into the following ranges: – Light—The average vehicle weight is 10,000 pounds or less. – Medium—The average vehicle weight is 10,001 to 19,500 pounds

90 Implementing the Freight Transportation Data Architecture: Data Element Dictionary – Light-heavy—The average vehicle weight is 19,501 to 26,000 pounds – Heavy-heavy—The average vehicle weight is 26,001 pounds or more.172 This classification may be different from that used by other data sources. 6.9.7.2 Temporal Differences Fatal Analysis Reporting System (FARS) • GVWR—This data element reports the gross vehicle weight rating. In 2007, GVWR was modi- fied to allow gross combination weight rating (GCWR) to be recorded for combination vehi- cles to match the nationally accepted reporting criteria for GVWR (i.e., FMCSA’s SAFETYNET and Model Minimum Uniform Crash Criteria). Use of GCWR instead of GVWR will impact only these vehicles: – Light trucks, 10,000 lbs. or less, pulling trailers (truck/trailers) (greater than 10,000 pounds GCWR) – Single-unit trucks, less than 26,000 lbs., pulling trailers (truck/trailers) (greater than 26,000 pounds GCWR)173 • NHTSA updates the FARS Analytical User’s Manual174 every year to summarize the evolution of coding. When conducting analysis across years, data users should check every data element of interest in each year’s coding manual. 6.9.7.3 Methodological Differences Carload Waybill Sample • TARE WEIGHT OF CAR—This data element reports the light weight for each car (i.e., not an average) in hundreds of pounds. Data users are advised to note that, if articulated, the tare weight represents the sum of the light weight vehicles for the total number of units of the consist (the set of vehicles forming a complete train).175 • Freight weight statistics in the Carload Waybill Sample are based on billed rather than actual lading weights. Even though the overall difference between billed and actual weights is small, statistically significant variation does exist among many individual commodities. Consequently, the use of billed weights in certain types of waybill analysis can lead to biased conclusions for a variety of reasons. The Surface Transportation Board (STB) therefore advises that it is unwise to extrapolate weight-related calculations to multiple decimal point levels of precision.176 • EXACT EXPANSION FACTOR—Each waybill uses an expansion factor (EXACT EXPANSION FACTOR) to expand car, ton, trailer/container, and revenue statistics to 100% levels. For example, the data element EXPANDED TONS reports the billed weight in tons multiplied by the expansion factor. The expansion factor is calculated according to the following formula: Factor = (Population count / Sample count) Commodity Flow Survey (CFS) • Data users are advised to note that the ton totals in the CFS represent the sum of separate ship- ments of a commodity as it moves through the production and consumption segments of the supply chain; hence, the tonnage of goods may be counted more than once in the production life cycle (e.g., goods that are moved through distribution centers).177 North American Transborder Freight Data (Transborder) • SHIPWT—This data element reports shipping weight for all imports but only certain exports in the Transborder database. Historically, shipping weight information from the U.S. Census Bureau has been available for shipments by vessel and air only. In the Transborder database,

Differences in Data Element Definitions 91 shipping weight data is available for all import modes. For exports, Transborder SHIPWT data is available for air and vessel modes but not for surface modes.178 6.10 Geospatial Data The two main sources of geospatial transportation data are the National Transportation Atlas Database (NTAD) and the Topologically Integrated Geographic Encoding and Referencing (TIGER). The NTAD is a compilation of multiple transportation data sources provided by the U.S. DOT and other federal agencies. The Bureau of Transportation Statistics (BTS) maintains and distributes the NTAD. However, the contributing agencies are responsible for the maintenance and accuracy of the data. TIGER, which is maintained by the U.S. Census Bureau, is made up of severable file types containing census geographic data and information such as geographical boundaries, roads, rivers, lakes, cities, census blocks groups, and census tracts. Geographical features contained in these two data sources may sometimes overlap; however, the attributes (or geographical information) contained in each data source may vary. Some of the geospatial data provided by TIGER is made available in the NTAD and vice versa. Data sources compiled within the NTAD and TIGER are available at their respective websites and the Freight Data Dictionary web application. For additional information, please review the data- base’s metadata. Identifying the differences within the attributes of each geospatial data sources was beyond the scope of NCFRP Project 47, as was merging and combining of the data elements contained in these data sources. The study team recommends that additional information on geospatial data integration be sought from other well-versed sources. Other publicly available freight-related data sources not included in the NTAD and TIGER are: • Cropscape,179 which is provided and maintained by the National Agricultural Statistics Service (NASS), and • National Corridors Analysis and Speed Tool (N-CAST),180 which is administered by the American Transportation Research Institute (ATRI) through an agreement with FHWA. Private-sector geospatial data sources containing commodity flow information include: • Transearch by I Global Insight181 • vFreight by the Economic Development Research Group182 Additional information on these data sources is available on their respective websites and the Freight Data Dictionary web application. Endnotes for both Chapter 6 and Chapter 7 are listed in the References section.