DATABASE NAME

     ORNHN  Oak Ridge National Highway Network (1993)

DATABASE DESCRIPTION

     Approximately 400,000 miles, or 10%, of the Nation's public
     roads, the original release of the National Highway Planning
     Network used for the Commodity Flow Survey

     Originally published by BTS on Transportation Data Sampler
     (BTS-CD-01)

ORIGINATING AGENCY

     Transportation Planning and Policy Group
     Energy Division
     Oak Ridge National Laboratory

AGENCY RESPONSIBLE FOR MAINTAINING DATA

     Bureau of Transportation Statistics

CONTACT

     National Transportation Atlas Databases
     Bureau of Transportation Statistics
     U.S. Department of Transportation
     Washington, DC  20590
     (202) 366-DATA
     Internet:  info@bts.gov

SOURCE

     1:2,000,000 U.S.G.S. National Atlas Digital Line Graphs,
     1980 supplemented by editing by Oak Ridge National
     Laboratory for the Federal Highway Administration
     
ACCURACY

      1,200 m (0.75 mi) root-mean square positional accuracy

TYPE OF GIS DATABASE

     Line database with nodes and links

FILES/SIZE/NUMBER OF RECORDS

     Two ASCII files

     1)   orhcfs.pnt          node file contains node locations
                              and attributes
     2)   orhcfs.lin          link file contains link attributes
                              and shapes for links

     Approximate size of files:         18,717,000 bytes
     Approximate number of nodes:       32,000
     Approximate number of links:       

FILE STRUCTURE

     The files are ASCII-fixed format.  Character fields are
     enclosed in quotation marks.

     The Node File

     The node file contains a single 80-byte logical record for
     each node referenced in the link file as a link endpoint. 
     Each record contains the 7-digit node number, the node's
     location, and optionally an alphabetic node name and other
     characteristics.  The longitude and latitude of each node
     will be the same as the first or last digitized point in the
     location chain of every link that uses the node as an
     endpoint. Node record fields and locations are:

       (1) Type               1: 1  (A1)    Always blank
       (2) Node ID            2: 8  (I7)                 
       (3) Longitude          9:20  (F12.4) Explicit decimal pt
       (4) Latitude          21:32  (F12.4)   
       (5) Name              33:60  (A28)  

     As is true for link IDs, the first two digits of the node
     number are the FIPS code of that node's state.  However,
     nodes on state and international boundaries begin with a
     prefix of '00'.  No individual link will cross a state
     boundary, and most (but not all) boundary nodes will connect
     links in two separate states.  All Canadian and Mexican
     boundary nodes have IDs of the form '0009xxx'.

     The longitude and latitude of each node will be the same as
     the first or last digitized point in the location chain of
     every link that uses the node as an endpoint.

     The Link File

     The link file contains a variable number of contiguous
     records describing each link.  Each record is an 80-byte
     card image.  Every link starts with a header record
     containing link attributes except for location.  The fields
     are described in the following section.  Following the
     header are as many card image records as necessary to hold
     the sequence of longitude/latitude pairs that describe the
     shape or alignment of the link, directed from the A-node to
     the B-node.  There are four pairs per record which should be
     read in format (8F10.6), although explicit decimal points
     are used when there are fewer than 6 decimal positions.  The
     number of pairs associated with the link is the last field
     on the header record.  Non-digitized links without alignment
     information will have only two pairs of coordinates.

     Link header record field names and locations

             Name                Format Columns  
                                                 
     ( 1)  Record type            (A1)   1       
     ( 2)  Link ID                (I8)   2:9
     ( 3)  Sign route             (A6)  10:15
           Sign route 2           (A6)  16:21
           Sign route 3           (A6)  22:27
     ( 4)  Length                 (F5)  28:32
     ( 5)  Heading                (A1)  33
     ( 6)  Urban flag             (A1)  34
     ( 7)  One-way flag           (A1)  35
                                        36
     ( 9)  Median                 (A1)  37
     (10)  Access control         (A1)  38
     (11)  Number of lanes        (I2)  39:40
     (12)  Traffic restriction    (A1)  41
     (13)  Toll flag              (A1)  42
     (14)  Bit field              (Z1)  43                        
              3 - Temporary Interstate
              4 - Interstate connector
     (15)  Truck route flag       (A1)  44
     (16)  STRAHNET flag          (A1)  45
     (17)  Bit field              (Z1)  46
              1 - Major highways subsystem inclusion
              2 - Major highways exclusion     
              4 - Tunnel flag
     (18)  Pavement type          (A1)  47
     (19)  Administrative class   (A1)  48
     (20)  Functional class       (A2)  49:50
     (21)                               51
                                        52:54
     (24)  Special systems        (Z1)  55
              3 - Trans-America corridor
              4 - NHS high priority corridor
     (25)  Status                 (A1)  56
                                        57:58
     (27)  A-node                 (I7)  59:65
     (28)  B-node                 (I7)  66:72  
           Auxiliary records     (3I1)  73:75  
           Number of points       (I4)  76:79  

          The columns denoted as "bit fields" each contain a
          single hexadecimal digit representing four bits for
          four binary (yes/no) flags, with the bits numbered from
          left to right.  For example, the digit "3" represents
          0011 [that is, 1-No, 2-No, 3-Yes, 4-Yes], and "D"
          represents 1101.  A blank is the same as a zero.  In a
          low-level programming language it can be read directly
          as a bit string;  otherwise, you must read it as
          character data and enumerate each of the 17 possible
          values ("0" - "F" and blank) and their translations
          into the four binary variables it may include.

     Link data element descriptions

     (1)  LINK TYPE
               Always blank.

     (2)  LINK ID
               Conceptually, the link ID is an 8-digit integer
               that uniquely identifies a link.  Analysis and
               editing software will generally use link IDs as
               pointers and cross references.  The link ID is
               composed of three parts:
                    (2a)  State prefix.  The first 2 digits are
                    the FIPS code of the state the link is
                    contained in, from 01 to 56, and therefore
                    provide a means of dividing the national
                    network into subnetworks by state.

                    (2b)  Link sequence number.  Originally,
                    within each state, the links as provided from
                    the USGS file were simply numbered from 1 on,
                    in the same order (excluding links entirely
                    outside the state boundaries).  New links
                    which were not subdivisions were added to the
                    bottom of the file, incrementing the sequence
                    number by one.  The leading digit of the
                    sequence number will be 0 for original NHPN
                    links;  other leading digits are used for
                    links incorporated from other sources so that
                    their original ID or sequence numbers may be
                    preserved.

                    (2c)  Subdivision number.  During the editing
                    process when a parent link is subdivided into
                    several daughter links, those daughters will
                    retain the same sequence number, but will be
                    assigned unique subdivision numbers from 1 to
                    9 in the last character position of the link
                    ID.  Links which have not been subdivided
                    will have a subdivision number of 0.  The
                    subdivisions are not necessarily in order
                    relative to the original parent.  Many parent
                    links have been (or will be) divided into
                    more than 9 daughters.  When this happens two
                    new "daughter" links with different sequence
                    numbers are established from the parent, and
                    they are subdivided.

     (3)  SIGN ROUTE
               There are three sign route fields, in case
               multiple routes jointly use the same road.  The
               first character indicates the type of sign route,
               the next five the route number or road name, left
               adjusted with blank fill on the right.  A
               qualifier may follow the route number.  Normally,
               multiple sign routes are listed in order of
               priority of the type, then the qualifier, then the
               route number, unless there is evidence that the
               state recognizes a different cardinal route on a
               link.  [For instance, I-59 takes precedence over
               I-20 in Mississippi and Alabama by state choice,
               and the mileposts follow the I-59 milepoint
               sequence.]
                    Sign route types:
                    I    Interstate.  Posted as an Interstate
                         route, but not necessarily on the
                         Federal-Aid Interstate system.
                    U    U.S. route.
                    S    State route.
                    T    Secondary state route.  Some states have
                         a second-level state-wide route
                         structure with an independent numbering
                         scheme.  Used in Texas for loop and spur
                         routes.  Used in Pennsylvania for "state
                         routes" (as distinguished from state
                         highways ["traffic routes"]), even
                         though route numbers are county
                         specific.
                    J    Interstate related route, such as
                         business routes and ramps.
                    C    County route number or road name.  Note
                         that there is sometimes ambiguity about
                         whether a route should be designated
                         county or state secondary.  Used in
                         Texas for farm (FM) routes.
                    L    Local road name.  Usually abbreviated,
                         as LOGLES for Oglesby Road.  Local
                         directional qualifiers are usually not
                         added:  North 4th Street would be
                         represented as L4-ST.
                    P    Parkway.
                    R    Inventory route number (not normally
                         posted).  Also used for Federal-aid
                         route identifiers.
                    V    Federal reservation road.  Used in
                         national parks and forests, Indian
                         reservations, and military reservations.

               Commonly used route qualifiers trailing the route
               number: 
                    N, S, E, W.  Directional qualifiers are used
                    to distinguish between distinct mainline
                    routes, and not to indicate direction of
                    travel on divided highways or couplets.
                    A  - Alternate.
                    BR - Business route or business loop.
                    BY - Bypass.
                    SP - Spur.
                    LP - Loop.
                    RM - Ramp.
                    PR - Proposed.
                    T  - Temporary.
                    TR - Truck route.
                    Y  - Wye.
               The state may incorporate letters into the sign
               route;  our dual intent is to make the physically
               posted route identifier recognizable and to have
               one of the sign routes match the state route
               identifier used in its inventory system.
               Sign routes or inventory routes following couplets
               may use a lower-case qualifier of "n", "s", "e",
               or "w";  "p" or "q" (cardinal or reverse);  or "l"
               or "r" (left or right) in the last character
               position of the field.

               In many cases the route or name could not be
               identified from the maps we were working with.  An
               "X" or "?," possibly with a qualifier, was then
               used, e.g., CX-MAD for unidentified Madison County
               road.  An "X" may also be used if the link is a
               collection of several differently named roads,
               particularly in cities.

               A trailing dash may follow the route name if that
               name applies to only part of the link.  A trailing
               plus sign may be used for the two street names of
               a couplet represented as a single bi-directional
               network link.

               Reliability:  Varies from state to state depending
               on the quality of the state highway map and
               whether we had county-level maps to work with.  In
               general, state route numbers and above have been
               carefully checked in all states as of 1986 or
               later.

               Sources:  Primary:  State highway and highway
               district maps.  Secondary:  County maps, USGS
               large scale maps,CA route log, HPMS.

     (4)  LENGTH
               The length of the link, in miles.  While indicated
               in tenths with an explicit decimal point,
               accuracies are generally no better than a mile. 
               Distances were estimated by first calculating an
               imputed distance from the sum of individual
               digitized segments of the link.  That imputed
               distance was then compared to an indicated
               distance on the state highway map.  If
               significantly different, the indicated distance
               was substituted.  "Significant" usually meant
               about a mile and a half.  The preponderance of
               links met the comparison test because, even though
               locational accuracy is only 3/4 of a mile at each
               end, there is a high local correlation in the
               direction and magnitude of errors.

               Initial imputed distances included an adjustment
               factor to compensate for the generalization of
               curves by chords during digitization and for
               slopes.  The most commonly used adjustment factors
               ranged from 0% to 4%.  These were determined from
               tests on Interstate highways using HPMS, where it
               was possible to obtain precise distances between
               county boundary crossings.  This was compared to
               imputed distances using the digitized segments,
               and the nationwide average discrepancy was about
               2.5%.  The factor varies due to the amount of
               curvature of the road, which in turn depends on
               topography and road class.  In the Midwest, the
               average is lower and in mountainous areas much
               higher.

               When new non-digitized links were added, distances
               were most commonly transcribed from a highway map
               or other source.  Sometimes, especially if the
               link was short and if there was confidence about
               endpoint absolute locations, a circuity factor for
               the link would be estimated and a geographically
               imputed distance used instead.

               Many other supplemental sources of distance
               information were consulted in special
               circumstances that can give accuracies of better
               than a half mile.  In many areas, particularly
               urban, distances were measured on large scale
               maps.  Additional accuracy was possible where we
               had county maps with indicated distances.  In
               California, state route logs were used to check or
               transcribe distances.  In other limited
               circumstances, it was possible to unambiguously
               determine distances from HPMS.

               In Idaho and South Dakota where no source of
               individual link distance information existed that
               was not privately copyrighted, links with imputed
               distances that failed comparison tests were
               calculated by regression equations using location
               in the state, the degree of curvature, and roadway
               characteristics, or else circuity factors were
               visually estimated.  In New York, most of these
               roads were either manually measured from 1:250,000
               scale maps or else circuity factors were
               estimated.

               Unfortunately, the pedigree of the distance data
               cannot be determined from the link file.  It
               should also be recognized that state highway map
               distance indicators are often ambiguous.  In
               general, users should count themselves lucky if
               the indicated distance on any particular link is
               within one mile or 3% of the true value, whichever
               is worse.  Although no systematic study has been
               done, our best guess is that the nationwide
               average error for non-Interstate distances is an
               underestimate near one-quarter percent, while the
               standard error for a link is between one-half and
               one mile.  Interstate distances are overestimated
               by an average of near 0.7 percent.  Clearly, this
               data is not adequate for rating purposes.

     (5)  HEADING
               This is the direction (N, S, E, or W) that one
               travels along the link from the A-node to the
               B-node.  Generally, it will match the heading
               physically posted along with the sign route at the
               roadside rather than a geographic heading, but the
               geographic heading may be used to avoid ambiguity
               when multiple sign routes use the link or when
               geographic and posted directions are reversed. 
               For instance, a link record will indicate you are
               heading east on I-81 near Wytheville, Virginia,
               when the roadside signing will indicate north on
               I-81 and south on I-77.

     (6)  URBAN FLAG
               This flag will indicate a subjective judgement
               about the degree of urban congestion which we use
               to influence routing choices.
                    U    Urban.
                    V    Urban bypasses.  Generally
                         circumferential routes.
                    S    Small urban or towns.
                    T    Partially urban.  That is, part of the
                         link is subject to congestion effects
                         limiting speed.  Operationally, we treat
                         links flagged "T" as having 0.5 miles of
                         their length subject to urban speed
                         reduction.

     (7)  ONE-WAY INDICATOR
               1    Traffic flow permitted from A- to B-node
                    only.  Most parallel one-way streets are
                    presently generalized as single two-way
                    links.
               2    Bi-directional link.
               3    Traffic flow from B- to A-node only.

     (9)  MEDIAN
               M    Divided highway (with median).
               C    Undivided (i.e., "centerline").
               F    Ferry.
               Source:  State highway maps, supplemented by HPMS.

     (10)  ACCESS CONTROL
               U    Uncontrolled access.
               G    Partially controlled access (with some
                    at-grade intersections).
               I    Fully controlled access.  All intersections
                    are grade-separated interchanges.
               F    Ferry.
               Source:  State highway maps, supplemented by USGS
               large scale maps and HPMS.  There is often
               ambiguity about the definition of partially
               controlled access.

     (11)  NUMBER OF LANES
               Generally either 2 or 4, representing all multi-
               lane roads.  Reported as 0 for ferries.  In some
               instances there will be links marked 2-lane
               divided representing a collection of access ramps.

               Source:  State highway maps, supplemented by HPMS. 
               Most maps will not distinguish between 2-lane and
               multi-lane undivided highways, hence many roads
               can be expected to be misreported.  In rare cases
               an actual value greater than 4 was reported by a
               state DMC or transcribed from HPMS, but for
               consistency it would be wise to treat all
               otherwise identical roads with 4 or more lanes the
               same way.

     (12)  TRAFFIC RESTRICTION
               In order of priority:

               Z -  No vehicles (generally a passenger ferry).
               P - Closed to public use.
               C - Commercial traffic prohibited.
               H - Hazardous materials prohibited.
               T - Large trucks prohibited.
               Q - Occasionally closed to public.
               L - Local commercial traffic only.
               W - Normally closed in winter.

               Completeness:  Not systematic in any state;  a
               blank field is no indication that a traffic
               restriction does not exist.  This information has
               usually been added on an exception basis;  that
               is, to solve a detected routing problem.  Parkways
               and national parks were assumed to be restricted. 
               Roads flagged in HPMS were also included.

     (13)  TOLL FLAG
               T    Toll road.
               B    The link contains a toll bridge (or tunnel).
               F    Free passage.  Roads not flagged can be
                    assumed toll free, but ferries are uncertain
                    unless explicitly marked.

               Source:  State highway maps, supplemented by HPMS.

     (14-3)    Temporary Interstate.  Temporary connecting routes
               for uncompleted Interstate sections.

     (14-4)    Interstate connector.  These Interstate ramps and
               connecting surface streets, while not part of
               Interstate mainline mileage, are necessary to
               maintain the connectivity of the Interstate
               system.

     (15)  DESIGNATED TRUCK ROUTE
               A    State designated route for STAA-dimensioned
                    vehicles. Consult appendix for completeness. 
                    When complete, all roads in the state
                    subnetwork which have been designated will be
                    flagged, but in many states there are
                    additional designated roads not included in
                    the network.
               B    Federally designated National Network routes
                    for large commercial vehicles (a subset of
                    State designated routes).
               C    While nominally a Federally designated route,
                    there may be construction-related
                    restrictions for an extended period.

               Source:  Federal Register (23 CFR 658), HPMS, and
               subjective assumptions when new parallel roads
               have been opened.  All Interstate routes are
               assumed to be included unless Federally exempted
               (regardless of State intent).

     (17-1)    Major highways subsystem.  A "major intercity
               highways" sub-network was constructed by using all
               Interstate links, most (but not all) rural
               principal arterials and urban express-ways, and
               subjectively selected "gap fillers" deemed to be
               logical extensions of rural principal arterials
               (most of which are urban principal arterials). 
               This subnetwork depends on proposed routes (not
               yet open to traffic) for its connectivity.  Many
               of the included routes are parkways closed to
               commercial traffic, particularly in New York.

     (17-4)    Tunnel flag.  A tunnel is known to exist on the
               link, generally because it was shown on a highway
               map.  Absence of a flag is no guarantee of the
               absence of a tunnel.

     (18)  PAVEMENT TYPE
               P    Paved.
               Q    Secondary paved.  Generally a subjective
                    determination that through traffic should be
                    discouraged, because of the road's local
                    nature or poor quality.
               G    Gravel, or otherwise indicated as below paved
                    quality.
               D    Dirt or unimproved.
               F    Ferry.

               Source:  State highway maps, for unpaved roads. 
               The P/Q distinction considered in part average
               daily traffic and lane width from HPMS where
               available or an indication of a lower road class
               on the state highway map or USGS map. 
               Occasionally the distinction was made to correct a
               routing a state DMC declared unrealistic.  But
               there were no hard rules covering the choice.

     (19)  ADMINISTRATIVE CLASS
               I    Federal-Aid Interstate.
               P    Federal-Aid Primary.
               S    Federal-Aid Secondary.
               U    Federal-Aid Urban.
               T    Combination of "S" and "U".  That is, part of
                    the road is FAU inside an urbanized area
                    boundary and the rest is FAS.  In some cases,
                    it is a combination of "N" on one side and
                    "S" or "U" on the other side of an urbanized
                    area boundary (common in Arkansas and North
                    Carolina).
               N    Not on a Federal-Aid system.
               F    Direct Federal system.

               All FAI roads are marked.  If blank,
               administrative class is unknown.  The confidence
               that the road is not FAP depends on the state. 
               With minor (mostly urban) exceptions, the FAP
               system now open to traffic is completely
               represented in the following states, as determined
               from examination of HPMS, FHWA-supplied large
               scale state maps, or county maps:
                    AL, AK, AR, CA, CO, CT, DE, DC, GA, ID, IL,
                    IA, KS, KY, LA, MD, MN, MS, MO, MT, NE, NV,
                    NH, NJ, NY, NC, ND, OH, OK, OR, RI, SC, SD,
                    TN, TX, UT, VT, VA, WA, WV, WY
               FAP coverage is believed near complete and
               correct, as determined from a national map of the
               FAP system, in the following states:
                    AZ, FL, HI, IN, ME, MA, MI, NM, WI

               Administrative and functional classes are
               generally not indicated on a small number of short
               connecting links identified as ramps or local
               surface streets in the sign route field. 
               Consequently, when constructing subnetworks based
               on these attributes, connectivity at some
               legitimate junctions may be lost unless tests for
               the existence of these connectors are made.

               There are a number of cases where a portion of an
               existing open route is of a lower administrative
               or functional class because a new alignment is
               planned.  Rather than explicitly add a new
               parallel projected link, the existing link may be
               flagged with the administrative and functional
               classes of its proposed replacement (except that
               surface streets will not be flagged as
               expressways).

     (20)  FUNCTIONAL CLASS
               Functional class identifiers use the standard
               codes in HPMS, plus special codes with a leading
               blank (represented below as an underscore, "_") to
               indicate combinations where part of the link is
               inside and part outside an urbanized area
               boundary.  As a rule of thumb, if less than 10 to
               20 percent of a link's length AND less than 2
               miles of a link is of a different functional
               class, a combination code will not be used.  [The
               same rule applies to administrative class.]  It
               should be noted that it is often difficult to be
               precise about the location of urbanized area
               boundaries without a current large-scale map that
               will show them.  The typical situation is that
               boundary locations were estimated from milepoints
               in HPMS.

               Rural                         Urban
          01 - Interstate.              11 - Interstate.
          02 - Principal arterial.      12 - Other expressway.
                                        13 -    
                                        14 - Principal arterial.
          05 - Major arterial.          15 -                    
          06 - Minor arterial.          16 - Minor arterial.
          07 - Major collector.         17 - Collector.
          08 - Minor collector.
          09 - Local.                   19 - Local.
                         Combination
                    _1 -  01 & 11
                    _2 -  02 & 12
                    _3 -  02 & 14
                    _4 -  06 & 12       05 & 14
                    _5 -  06 & 14
                    _6 -  06 & 16       07 & 14
                    _7 -  07 & 16       07 & 17

          The functional class "05" is used in states which
          distinguish an intermediate rural arterial class, and
          in some other states which indicate minor arterials of
          different importance.  These roads are reported as "06"
          in HPMS.

          Some states use codes of "13" and "15" in HPMS.  The
          former are generally urban extensions of rural
          principal arterials, but both probably fall under the
          FHWA definition of urban principal arterials (class
          "14").
          All roads which are in whole or part rural principal
          arterials will have functional class indicated, and
          coverage is complete as of 1984.  That is, there are no
          rural principal arterials open to traffic in the
          continental U.S. that are missing from the network. As
          a practical matter, rural principal arterials can be
          identified by codes of "02", "_3", and "_2".

          Beyond that, the completeness of functional class
          indicators varies by state (consult appendix). 
          Functional classes below "07" and "16" are rarely
          indicated because, even if reported in HPMS, their
          location and identity are usually impossible to
          estimate.  There are very few roads in the network in
          these low functional classes anyway.  In many states,
          functional class is not reported for Interstates even
          when other arterials are marked.  Therefore,
          administrative rather than functional class should be
          used to identify Interstate highways.

          Functional class will be indicated for all links that
          are rural arterials in the following states.  Coverage
          of rural arterials is also substantially complete in
          these states, but most urban minor arterials and many
          urban principal arterials off the FAP system may have
          no network representation.
               AL, AK, AR, CA, CO, CT, DE, DC, GA, ID, IL, IA,
               KS, KY, LA, ME, MD, MI, MN, MS, MO, MT, NE, NV,
               NJ, NY, NC, ND, OH, PA, RI, SC, SD, TN, TX, VT,
               VA, WA, WV, WY

          Because of the special meaning of a leading blank, this
          field cannot be treated as a numeric variable without
          prior user modification.

          Source:  HPMS and large scale state or highway district
          level maps supplied by FHWA.

     (24)  SPECIAL SYSTEMS

     (24-3)    Trans-America Corridor.  An Oak Ridge subjective
               interpretation.

     (24-4)    NHS High Priority Corridor.  An Oak Ridge
               representation of corridors included in the
               proposed National Highway System by legislative
               mandate.

     (25)  STATUS
               O    Open to traffic.
               U    Under construction.
               P    Proposed.  [Often reasonable guesses for
                    route alignments that will match distances in
                    HPMS.]
               X    Speculative.

DISCUSSION


                         DESCRIPTION OF THE
             OAK RIDGE NATIONAL HIGHWAY PLANNING NETWORK
                            Version  1.1 


                           Bruce Peterson
              Transportation Planning and Policy Group
                           Energy Division
                    Oak Ridge National Laboratory
                           22 December 1992


                              ABSTRACT

     The Oak Ridge Highway Planning Network is a database of
     major highways in the United States.  It is a foundation for
     analytic studies of highway performance, for vehicle routing
     and scheduling problems, and for mapping purposes.  The
     network is based on a set of roadways digitized from the
     National Atlas by the U.S. Geological Survey.  It has been
     enhanced at Oak Ridge National Laboratory by adding
     additional roads and attribute detail and correcting
     topological errors to produce a true analytic network.  This
     documentation is intended primarily to assist users of this
     database by describing its structure, data elements, and
     development.

     OVERVIEW

     The Oak Ridge Highway Planning Network is a geographically
     based analytic network of the major highways in the United
     States.  It was developed at Oak Ridge National Laboratory
     to support analyses of a wide variety of highway
     transportation issues that require use of a network. 
     Potential uses include studies of highway performance and
     network design, vehicle routing and scheduling, and social
     and environmental impacts of transportation.  It presently
     contains approximately 400,000 miles of roadway and will,
     with varying degrees of accuracy, show the location of these
     roads and attribute detail about their characteristics.

     The checking and correction of information in this network,
     and the addition of new information, is an ongoing activity,
     and the completeness of information varies significantly
     from state to state.  This version is therefore a snapshot
     of a constantly evolving database.  Future versions of this
     network can be expected to contain more complete attribute
     detail and additional roadway attributes, as well as
     additional roads.  The appendix shows the current
     development status of each state subnetwork.  Descriptions
     under individual data items should be consulted to better
     appreciate data reliability and completeness in areas of
     interest.

     While we have naturally attempted to reduce the incidence of
     errors, it is unrealistic to expect that all have been
     caught, especially considering the vagueness of much of the
     source information and the need for subjective
     interpretation.  Further, the data is not warranted in any
     way.  Therefore, no user should rely on any data without
     doing his or her own validity checking.  We at Oak Ridge
     would appreciate reports of errors that users discover, and
     we will endeavor to correct them in future versions.  Error
     reports may be sent to Bruce Peterson at telephone
     615/574-4419, or through E-mail on Internet <gto@ornl.gov>
     or Bitnet <gto@ornlstc>.

     SOURCES

     Data included in this network come from several sources. 
     All data has been derived from or checked against
     information obtained from state and Federal governmental
     agencies.

     (1)  USGS National Atlas Digital Line Graphs

     The foundation data was the set of roadways digitized by the
     U.S. Geological Survey (USGS) from 1:2,000,000 scale plates
     of the National Atlas, circa 1980.  The geographic accuracy
     of the road alignments in these digital line graphs (DLGs)
     was very high considering the scale, but it was primarily a
     cartographic network, intended for automated mapping.  Much
     of the information about road attributes important in
     analytic studies was limited, missing, or obsolete, such as
     sign routes.  Furthermore, topological details (the way
     roads are connected) were sometimes excessively generalized
     for our purposes, especially in urban areas.

     (2)  State Maps

     We obtained state published highway maps dated some time
     between 1983 and 1990 as the primary data source for most
     non-locational data.  The typical state highway map shows at
     least all state highways, indicates distances between major
     junctions, has urban area inserts, and was published by a
     state agency.  Exceptions are California, Idaho, New York,
     and South Dakota, where the standard highway maps are
     privately copyrighted.  In New York, detailed Department of
     Transportation highway maps (1:250,000) were used, but which
     lacked distance indicators.  In California, highway district
     maps and route logs were used, augmented by USGS 1:250,000
     and 1:100,000 scale maps.  In Idaho and South Dakota, state
     highway agency maps were supplied by the Federal Highway
     Administration (FHWA) showing routes and functional classes,
     but not distances.  When "state maps" are mentioned in
     previous discussions of individual data items, it is to this
     set that they refer.

     (3)  Other Maps

     We had county road maps and highway district maps from a
     number of states, which were used for whatever detail they
     could supply, topological and attribute.  USGS 1:250,000 and
     1:100,000 scale maps were frequently consulted to resolve
     locational and attribute details.  We considered them
     supplemental sources to be used when the state maps were
     inconclusive or not detailed enough.  FHWA supplied maps for
     every state showing highway functional classes.

     (4)  HPMS

     FHWA allowed us access to the Highway Performance Monitoring
     System (HPMS), which is logically an inventory of highways
     in the United States supplied by individual state highway
     agencies.  In many states it is possible to translate
     between individual records and the real roadways they
     describe.  When it was possible, selected attribute
     information in HPMS was transcribed to network links.

     (5)  Geoplex Corporation

     Under contract with FHWA, the Geoplex Corporation digitized
     most Federal-Aid Primary (FAP) roads in each state that were
     not included in the USGS network.  They used 1:500,000 scale
     USGS state maps as their source.  In addition, they marked
     links on their own network (which also used the USGS DLGs as
     its base) which were designated truck routes and FAP routes. 
     We are in the process of incorporating this digitization
     into the network;  a few states (consult appendix) are
     complete in this version.

     (6)  Defense Movement Coordinators

     One of the functions of this database is to assist military
     personnel in the management of military movements.  State
     National Guard offices are now being staffed with Defense
     Movement Coordinators (DMCs) who are familiar with their own
     states' highway systems and who maintain contact with both
     state highway agencies and local units.  Because of their
     management responsibilities, they have an interest in
     correcting errors and maintaining the accuracy and currency
     of the data.

     CONSTRUCTION

     The base data for this highway network were the 1:2,000,000
     USGS digital line graphs.  The locational data were edge
     matched, scissored into state subnetworks, and element
     identifications were assigned.  As later described, initial
     geographically calculated distances were assigned to each
     link.  In addition, other attribute details that could be
     deduced from the USGS attributes (e.g., most sign routes,
     access control, and divided highway flags) were also
     incorporated.

     An initial clean-up pass was made through every state to
     check or add sign routes and check distances.  The most
     serious topological problems were also resolved at this
     time.  Except for location, attribute editing was done with
     a text editor.  For location, a set of programs were locally
     written to move links and junctions, adjust their shapes,
     subdivide original links by establishing new junctions along
     their length, add new non-digitized (straight line) links
     between nodes, and incorporate digitized chains from other
     sources.  At this stage, new junction locations were
     estimated visually, either relative to an existing link or
     at a free-standing new absolute location.  Initially, the
     network contained approximately 320,000 miles.

     From this point on, the treatment of each state was
     individual.  Several states were test sites for the Army
     Forces Command convoy management program, and many new links
     were added at their request as potential convoy routes. 
     Access roads to major military bases were also added,
     although they have been excluded from the public version of
     the network.  In most cases, detailed descriptions of these
     roads were supplied by DMCs.

     We have also attempted to include in the network the entire
     Federal-Aid Primary system (essentially all rural arterials,
     though only a small proportion of urban arterials).  Again,
     new link locations were not digitized, although more care
     was taken in the location of new junctions.  During this
     second pass through the states, road locations were compared
     to large scale USGS maps and the more egregious problems
     were corrected, so many of the road alignments have been
     changed from the original DLGs.  Insofar as our data sources
     permitted, we attempted to add administrative and functional
     class indicators to network links.  Only Hawaii has not had
     this information checked, although an initial assignment of
     classes has been made.

     The third stage, which began in 1988, is the inclusion of
     Geoplex digitized alignments and identifiers for designated
     truck routes.  Consult the appendix for states where most of
     our straight-line added links from stage 2 have been
     augmented with Geoplex alignments.

     Finally, most link alignments in Washington State are the
     result of an experimental incorporation of alignments from
     the 1:100,000 USGS digital line graphs.  Users are cautioned
     that they may find odd shapes in the immediate vicinity of
     interchanges, due to the integration procedure following
     particular ramps when matching links.  Also, short portions
     of distinct NHPN links (on the order of 200m) may directly
     overlay each other in these vicinities.

     LOCATIONAL ACCURACY

     Although we have not conducted a systematic study, our
     general impression based on comparison with large scale maps
     is that geographic accuracy of the original USGS National
     Atlas DLG data is about 1500 meters (root mean square). 
     This average includes a number of rural areas where more
     significant systematic errors occurred.  Urban areas, with
     their high levels of generalization and many competing
     features causing displacement on the original map plates,
     also tended to have less than average accuracy.

     The first pass of the states, where urban topological
     problems were untangled, often purchased greater clarity at
     the cost of minor reductions in locational accuracy, because
     new junction locations were visually estimated relative to
     points or links which already had locational problems.

     Naturally, new straight line links have much worse accuracy
     away from their endpoints.  And their endpoint locations,
     insofar as they were visually estimated subdividing points
     on existing links, can also be expected to be slightly worse
     than the original USGS standard.  Both problems are
     gradually being overcome as more accurate digitized shapes
     are incorporated, resulting in accuracies significantly
     superior to the original (excepting the remaining straight
     line links, of which there will be many).

     Although variation is very high because of the diversity of
     sources and integration techniques, we believe that 1200 m
     RMS is a fair overall indicator of locational error.  In
     evaluating locational accuracy, it is important to remember
     that our primary goal was a topologically correct analytic
     network.  We did not wish to throw away locational
     information given to us by USGS, but it was usually a
     secondary concern.  While this network can certainly be
     mapped, it would be misleading to regard it as a
     cartographic database because of the large number of
     non-digitized links.  Also, because many roadway alignments
     have been modified, locational errors in this database
     should not automatically be assumed correlated with those of
     the USGS National Atlas DLGs when overlaying these roads
     with other features from that source.  For instance, it is
     quite possible that a road and an adjacent USGS-digitized
     shoreline will each have 1200 m errors in the opposite
     direction, making the road appear to be 2 km offshore.

     TOPOLOGICAL ACCURACY

     One of the purposes of this network is to serve as the base
     of a roadway inventory system.  The accurate representation
     of roadway connectivity is therefore very important to
     facilitate the automated assignment of inventory data to the
     network.  There are, however, limits on the fidelity of the
     network's topological representation imposed both by source
     material and by our desire to reduce maintenance problems by
     avoiding complexity.  In general, our topological resolution
     standard was 0.2 miles.  That is, if two junctions are
     closer than 0.2 miles apart, they will be represented by a
     single node in the network.  Network matching procedures
     which attempt to construct equivalency tables between this
     network and finer resolution geographic roadway data must
     take this generalization into account.

     Resolution can, of course, be much worse than this standard. 
     Most state highway maps will typically not resolve junctions
     less than a half mile apart;  if that map was our best
     source material they will not be resolved in the network
     either.  Most interchanges are represented by single nodes
     as well, even though they may contain several ramps longer
     than 0.2 miles.  A very small number of interchanges do have
     connecting links in the network representing a collection of
     ramps, most commonly on toll roads, where a single node
     would have caused excessive displacement of the through
     routes.  A related consequence of this topology standard
     occurs when a route is carried by parallel one-way streets,
     known as couplets.  This is a common occurrence in urban
     areas, but because the two real components of a couplet are
     seldom more than two city blocks apart, they fall within the
     resolution standard, and so are represented by a single
     link. 

     COMPATIBILITY

     In general, both link and node sequence numbers will be
     preserved between successive releases.  That is, the same
     sequence number will never point to logically different
     network elements in different versions, and logically
     equivalent elements will usually have equivalent or related
     identifications.  However, there will still be many
     circumstances where equivalencies will not be transparent. 
     Establishing equivalencies between elements may be required
     by any users who modify or add attribute information to
     their own network copy and wish to transfer it to a new
     version.  The problem will also arise when users make their
     own structural changes, and wish to transfer their data onto
     another user's network.  It is recommended that these users
     follow the same Oak Ridge conventions listed below in their
     own modifications if they wish to maintain transferability,
     although they must also explicitly mark new element IDs they
     establish so they may be distinguished from the same IDs
     which may be added in a future Oak Ridge version.

     Several different types of structural changes can be
     expected to occur, each with different considerations for
     maintaining compatibility.

     (1)  Link subdivision.  An existing link may be subdivided,
          usually in order to add a new roadway junction or
          interchange, but sometimes to add control points,
          traffic generators, or to indicate a change in roadway
          attributes.  In most cases the resulting links will
          have the same link sequence number as the parent, but
          will differ in subdivision number.  When the parent
          link itself had a non-zero subdivision number, one of
          the daughter links may retain the same number.  [In
          fact, there is no intention that subdivision numbers
          will be retained between releases, even when the links
          are physically identical.  Endpoint nodes of
          subdivisions must be examined to establish this
          equivalency.]  In rare cases, this convention will be
          ignored because more than nine subdivisions would
          otherwise result.  When this happens, a new link
          sequence number has to be assigned to one or more of
          the daughters, and equivalencies can be recognized only
          by following a sign route chain between the endpoint
          nodes of the group of parent links with the original
          sequence number.  This is unusual enough that it is
          probably best done manually.

     (2)  Link combination.  There are many nodes which serve no
          purpose, that is, which divide a portion of a route
          into two links with identical attributes, but which are
          not junctions themselves.  These links may be combined
          into a single link, retaining the sequence number of
          one of the incident links, and eliminating the node,
          which is then deleted from the node list. 
          Alternatively, that node may be moved to a new location
          along one of the incident links, in effect establishing
          a new dividing point between the same consecutive links
          along the route.  The node is then available to be used
          as a junction for a completely new link intersecting
          the old route without the necessity of establishing new
          subdivisions.  Topologically, the route's structure
          will not have changed, since it will have the same
          sequence of links and nodes, although physically the
          node may be several miles from its original location.

     (3)  Junction detail.  A common structural enhancement is to
          resolve detail in a junction formerly represented by a
          single node.  In this case a group of closely spaced
          nodes will be established.  Only one of them will have
          the original node number;  the others will have new
          numbers.  To the extent possible, the new short links
          connecting these nodes will logically be subdivisions
          of the original incident links, but new subdivisions
          will always form a linear sequence along the original
          route.  Therefore, there will often be some completely
          new link sequence numbers within the group.  It also
          follows that the endpoint node of the original parent
          link sequence may be changed to another node within the
          junction group.  [The ideal way to handle this problem
          would be to recognize daughter nodes, either by
          explicitly noting a parent node number or by having
          node "subdivision" numbers as is done for links.  At
          present, it has not been considered worthwhile enough
          to add this complication to network editing procedures,
          but it may be adopted in the future, especially if
          interchanges are resolved into explicit ramps.

     (4)  New alignment.  When a new roadway is constructed to
          carry an existing route, the ORNHN may be updated in
          either of two ways.  If the new route is close to the
          old (including reconstruction), or if the old route is
          removed or otherwise closed to traffic, the same link
          IDs will simply be transferred to the new alignment. 
          Logically, existing NHPN links will be physically moved
          over the map surface with few if any topological
          changes.  [Of course, links attributes like access
          control may be changed.]  When this happens there
          should be no logical problems in conversion. 
          Alternatively, if the old route remains a significant
          traffic artery, it will remain in the network with the
          previous ID (but a different sign route), and the new
          alignment will be added as a completely new link and
          ID.

     ACKNOWLEDGEMENTS

     The preponderance of the funding and assistance for
     development of this network has been from the U.S. Army
     Forces Command to support convoy movement management and
     from the Federal Highway Administration (FHWA) to support
     national planning and policy studies.  We at Oak Ridge
     gratefully acknowledge the help we have received from many
     individuals at FHWA, FORSCOM, and in National Guard state
     offices, especially program managers Bob Proferes and Steve
     Lewis (FHWA), MAJ William Fulmer and LTC Dale Easterling
     (FORSCOM), and MAJ Jim Massfeller and LTC Woodrow Sullivan
     at the National Guard Bureau.  Additional development and
     maintenance support has also come from the Military Traffic
     Management Command and the Department of Energy in order to
     include data relevant to their own operational and policy
     interests.  Special thanks are due to the research
     assistants who have painstakingly added and checked data
     over the years:  Marilyn Osterlund, Tom Sills, Carolyn
     Bolton, Mike DeVivo, and Joy Medford, and to the staff of
     the University of Tennessee Transportation Center.

     Appendix
     STATUS OF STATE SUBNETWORK UPDATES
     21 December 1992
     State    Adm   Func          Urb      Links      Update
          Algn   PO_    1357_  Trk    Miles     Notes
     AL 01  G  H 820  H 12510   T  P   9.6   13        91 94
     AK 02  - HM 352 HM 1.116   U  S   9.7    4        12
     AZ 04  .  N 811  . 02413   U  .   6.0    3        8  11
     AR 05  G  H 82.  H 1352.   U  U   8.2   10        06 1B
     CA 06  G  H 910  H 23510   T  S  16.1   16        91 94
     CO 08  G  H 540  H 12340   T  S   8.8    7        92 OC
     CT 09  G  H 910  H 23510   T  S   2.0    5        99
     DE 10  G HM 730 HM 12520   U  U   0.7    2        07
     DC 11  G HM A0. HM 1180.   U  U   0.1    2   DC   06
     FL 12  .  N 7.3  . .21.7   B  .  11.9   17        8  13
     GA 13  G  H 730  H 12430   T  S  16.0   27        93 22
     HI 15  -  H 432 HM 00332   .  U   1.3    3 active 2B
     ID 16  G  H 730  H 13230   T  U   5.0    4        04
     IL 17  G  M 91.  M 1071.   B  U  13.4   25        17
     IN 18  .  N 8.2  . .20.8   P  .   7.4   10        8  16
     IA 19  .  H A00  H .3601   B  .   9.9   11        8  9C
     KS 20  G  H 810  H 13410   T  S  10.6   11        94
     KY 21  .  H 630  H 12330   T  P   7.3   10        8  15
     LA 22  G  H 55.  H 12340   P  U   7.9   12        2B
     ME 23  G  N 641  M 12331   U  P   4.3    6        18
     MD 24  .  H 911  H .2502   U  P   2.9    6        8  0A
     MA 25  .  N 7.3  . 222.6   U  .   2.6    5        8  11
     MI 26  G HN 901  H 13321   T  S   9.8   12   MI   95 13
     MN 27  G  H 91.  H 1351.   T  U  11.8   13        01
     MS 28  .  H 811  H 02511   P  P   8.3   10        8  12
     MO 29  .  H 811  H .2512   U  P   9.7   11        8  0A
     MT 30  G  H 910  H 23410   U  U   7.8    5        13
     NE 31  .  H 910  H 13510   P  P   8.6    8        8  0C
     NV 32  G  H 55.  H 11241   T  U   5.4    3        04
     NH 33  .  H 7.3  H 124.3   T  P   1.8    3        8  18
     NJ 34  .  H 730  H 12510   T  P   2.9    7        8  0B
     NM 35  .  N 5.5  . .2..8   U  .   8.6    5        8  11
     NY 36  G  H 820  H 12520   T  U  12.4   25   NY   03 0A
     NC 37  .  H 640  H 13331   U  S   9.7   15        8  11
     ND 38  .  H A00  H 12700   U  P   6.4    5        8  0A
     OH 39  .  H 820  H .2512   P  P  10.2   16        8  0A
     OK 40  .  N 703  . .2107   P  .   9.2    9        8  9A
     OR 41  .  N 8.2  . 041.6   B  .   6.9    4        8  0A
     PA 42  G  H 91.  H 1261.   U  U  13.1   32        0C 19
     RI 44  G  S 721  S 12511   T  S   0.6    2        94
     SC 45  .  H 910  H 12510   U  P   7.4   12        8  11
     SD 46  .  H 9.1  H 135.1   P  P   6.9    5        8  9C
     TN 47  .  H 901  H .16.3   T  P   8.3   12        8  26
     TX 48  G  H 73.  H 1334.   T  U  30.1   30        98
     UT 49  .  H 712  . .1..9   P  .   5.1    4        8  0C
     VT 50  .  H 712  H .2414   U  P   2.0    3        8  0A
     VA 51  .  H 812  H 02413   T  P   8.6   14        8  14
     WA 53  D  M 811  M 13411   T  P   7.2    7        8A 2C
     WV 54  .  H 621  H .1422   U  P   4.5    5        8  05
     WI 55  .  M 8.2  M 14501   T  .  10.4   13        8  9C
     WY 56  G  H 64.  H 12331   U  U   6.7    5        11
          -------------------------------------------
     Algn      G  Geoplex digitization included
               D  USGS 1:K DLG shapes incorporated
     Adm       Administrative class
               H  From HPMS
               M  From special state map or county maps
               N  From national system map
               S  From state publication
               .  No reliable information
               PO_: percentage of mileage (multiple of 10) in
                    FAI/FAP;  other classes; and unknown.  ["A"
                    is > 95%; "." is < 0.5%; "0" is 0.5 to 5%]
     Func      Functional class
               H  From HPMS
               M  From special state map
               S  From state publication
               1357_:    percentage of mileage in functional
                         classes ending in 1; 2 and 3; 4, 5 and
                         06; 16, 7, 8 , and 9; and unknown.
     Trk       T  State designated truck routes flagged
               U  Federal and state designated routes flagged
               B  Only Federal routes flagged
               P    Truck routes have not been explicitly
                    examined, but all FAI/FAP roads are
                    Federally-designated routes, and have
                    derivative flags. Most other state highways
                    are state designated routes, but are
                    generally unflagged.
     Miles     Road mileage in subnetwork, in thousands.
     Links     Number of links in subnetwork, in hundreds
     Update    Year/month of last major structural revision and
               last minor revision [94=1989/April;
               0B=1990/November]
               -------------------------------
     Notes:    MI   Many roads indicated as FAS or FAU with
                    functional classes of 07 or 16 in HPMS (1984)
                    are shown as FAP on the national map.  They
                    are marked FAP with their HPMS functional
                    codes in the link list.
               DC   The District contains many additional urban
                    minor arterials on the FAP system which are
                    not included.
               NY   All urban principal arterials are indicated,
                    except for many non-FAP local streets in the
                    five boroughs of New York City (because HPMS
                    uses grouped data there).
               active    The state is currently undergoing major
                         structural revision, and much attribute
                         data is likely to be incomplete.