Contents of the S data system Shorelines and county boundaries as a network Bruce Peterson Oak Ridge National Laboratory 2003 January 21, rev 2003 July 20 The S-network contains shorelines and North American political boundaries represented as a link-node network. As a network, analysis and maintenance may be facilitated by the use of techniques and programs developed for other types of networks. The network is analytic, and should produce polygons of any feature in the attribute set. Except for minor editing and edge matching, the US portion of the network was extracted from TIGER/Line files of approximately 1993. Hydrologic features without width, that is, represented by a single polyline with land on both sides, are not included. No hydrologic features are named. FIPS codes identifying counties are those of 2000. However, county sub-division numbers allow a mapping into 1990 FIPS codes. The Canadian provinces and Mexican states were drawn from the Bureau of Transportation Statistic's NORTAD dataset of 1997 (which was in turn derived from the Digital Chart of the World) and were integrated with the US portion. Canada and Mexico contain outside shorelines only, and Nunavut has not been separated from Northwest Territories. All US coordinates are in decimal degrees, datum NAD27. Longitudes are all west of Greenwich in the range [-240, -30]. I generally regard the data as having 100 m locational accuracy. However, there are some areas in TIGER where accuracy is worse, including many urban areas and some counties where a consistent internal offset was observed. The accuracy in Canada and Mexico is suspected to be approximately 1000 m, and the datum of the DCW (whatever it was) was not disturbed except in the immediate vicinity of the US border. The principal purposes of the data system are to provide background displays for context when plotting other geographic data, and to segment other types of objects into counties. The polylines in this network run from an A- to a B-endpoint node, and note in the attribute set the county FIPS codes on the left and right sides, and whether there is land or water on each side. Many of the links outlining islands and lakes are loops where the A- and B-nodes are the same. These links have no compass heading, and instead are marked as single link polygons with an area, positive or negative depending on whether the line circles counter-clockwise or clockwise, respectively. These polygons may be deleted without corrupting the topology of the remaining network, provided that nested polygons are deleted in the order smaller first. (A single link polygon contains no objects except smaller loops. It is not unusual to have 3 levels: a lake within an island within a lake.) A small number of links separate nothing, that is, their left and right attributes are identical. Object IDs, link and node, begin with a 2-digit state FIPS code followed by a sequence number, and in fact objects are routinely stored in state files. State boundaries are composed of chains of links with different states on either side. Those links and their connecting nodes have IDs beginning '00' and are stored in a boundary link or node file, called by convention state 00. New "FIPS" codes between 80 and 90 were invented and arbitrarily assigned to Canadian provinces. The listing in in an associated file of county codes. Currently, all of Canada is in a single file labelled "S88." Additional invented codes are 91 for Mexico and 99 for international ocean waters. The county boundaries are political, and run out to the middle of the Great Lakes and into the oceans for approximately 3 miles. In most cases the type of water body is not characterized, and a single link may include lake, river, channel, estuary, bay, etc, shorelines. It is typical for a single link to run both up and down the left and right banks of the same river. A consistent set of ocean and Great Lakes shorelines have been identified so that they may be mapped as the "outside" boundary of the US rather than the offshore political boundary to make a more conventional map. These "ocean" shorelines have been edited in Washington, Michigan, and Delaware. In all other states, the ocean shorelines are merely the boundaries of polygons that include open waters, and therefore include many rivers and streams that run several miles inland. The typical procedure for drawing a map with county boundaries that end at the ocean shoreline would be 1) Eliminate all loops (lakes and islands) with enclosed areas under some threshhold (say 0.1 sq km). 2) Draw lines with water codes of 7 and above on one side, and less than 7 on the other. This is the ocean shoreline. 3) Draw all county boundaries with water codes less than 7 on both sides. The S-network topology is intended to be strictly planar. That is, no two polygons will share any points in their interiors, and they will share points on their boundaries only if they share common links or nodes in their boundary chains. Most likely, there are no cases of corrupt topology in the native format files. However, there are numerous cases where one polygon may approach to within 1 m of a non-adjacent polygon. A cleaning operation in a GIS system may regard these as within its tolerance buffer, and insert a node at the near-approach. Also, the polygon shapefiles included in this distri- bution were constructed using 4-byte floating point coordinates, and there is a small chance that this could convert a near-approach into an overcross, since the resolution of lon/lat coordinates in REAL*4 is slightly worse than 1 m. Users should also note that minor modifications of shape, say from thinning or even re-projection, may disrupt the original topology. Each state has 3 files: A. Nodes (susXX.ndr, where 'XX' is the state's FIPS code) contain only node ID and longitude and latitude, in format (1X,I7,2F12.6). Examples: 0504643 -93.209672 34.947608 2406863 -79.241416 39.560860 The only functions of a node are to anchor incident links and establish their topology (connections). B. Link attributes (susXX.llr) contain Attribute Format Columns (1) Link ID I8 1-8 Link type A1 9 always blank unused 10 (2) A-node I7 11-17 (3) B-node I7 18-24 (4) Length (km) F7.2 25-31 explicit decimal point edit field A2 32-33 ignore (5) Heading A2 34-35 link: 8 pt compass loop: '+p', '-p' (6) Enclosed area F6.0 36-41 sq km, explicit sign in col 36 (7) Source A2 42-43 T=Tiger J=subjective H=freehand M=BTS NORTAD (DCW) blank=original Tiger (8) Water type left I2 44-45 1=land (9) Water type right I2 46-47 2=water (undifferentiated) 5=river 6=lake 7=Great Lakes 8=marine harbor 9=Ocean (10) county FIPS left I6 48-53 (11) county FIPS right I6 54-59 (12) county subdiv left I2 60-61 (13) county subdiv right I2 62-63 Examples: 05037450 05088770508877 0.198km-p-0.000 2 1 05095 05095 05037510 05093410509341 1.04 +p+0.063 2 1 05095 05095 3 05040180 05046430504633 0.823kmW 0 2 2 05105 05149 2 05040190 05046700504643 2.33 SW 0 1 1 05105 05149 ---------1---------2---------3---------4---------5---------6---------7 Length, heading, and area are clearly derivable from the link's polyline, so are technically redundant, but are included to simplify operations in application programs. Water types greater than 4 are intended only for major water bodies that would be shown on a national map. FIPS codes allow 6 digits in anticipation of international use. County subdivisions are usually blank, but will be used when different parts of a current county had different FIPS codes in a previous year. For example, Yakutat borough in Alaska was formed from parts Skagway and Valdez, so those two parts have different suffixes. The mappings between subdivisions and previous codes are contained in the ancillary file of county codes. In GIS versions, a derived character variable called LTYP (or ILTYP if integer) is used to facilitate mapping. Types are: 0 - Identical left and right attributes 2 - Shoreline only 4 - Ocean/GrtLk bdry with internal water 5 - with land 6 - County bdry, in water or shoreline 7 - County, on land 8 - State boundary, water or shore 9 - State, on land A - International, water or shore B - International, land C - International waters boundary C. Link location (susXX.lcr). Each polyline has a header record containing Link ID and the number of vertices in format (I8,I4), followed be as many card image records as necessary to hold the vertices in format (8F10.6), two coordinates per vertex, with implicit decimal point. Examples: 05040170 5 -93217208 34946516 -93216536 34946584 -93212288 34947224 -93210368 34947548 -93209672 34947608 05040180 3 -93209672 34947608 -93217200 34947700 -93217208 34946516 All objects are in ID order in their files. No polyline object contains more than 2400 vertices. Mexico (S91) has the only state file that contains its offshore international waters boundaries. The program RHA reformats raw files into a variety of GIS formats (Arc/Info generate, Maptitude, MapInfo, and shapefile) under the control of a configuration file. The Fortran source code can be found in the railroad section. The CFG commands that produced the shapefiles displayed in the illustration are reproduced below: # RHA config file for boundaries type V-shapefiles (A-arc P-maptitude M-mapinfo T-transcad2) # AREALMT 0.2 If not commented out, would exclude lakes <= 0.2 sqkm DSYSTM s THIN D=1000 E=21 21 m thinning to reduce file size # lincl LTYP c 1 lincl BHDNG c 2 lincl WTRLF i 1 lincl WTRRT i 1 lincl FIPLF i 6 lincl FIPRT i 6 # output szs output files will be: .\szsl.shp, szsl.shx, szsl.dbf # input ..\s\sus00.llr input ..\s\sus05.llr input ..\s\sus47.llr END Unified national covers derived from the state files are included in this distribution, and are in ESRI shapefile format which most commercial GIS systems can load. Covers SCUL and SCUO contain political boundaries only in polyline and polygon forms, respectively. SCUO is most appropriate for determining which county contains an object, and SCUL for cutting objects at borders into county sections. SDUL and SDUO further subdivide counties into inland and ocean portions, and are most appropriate for displays. This dataset was produced under the sponsorship of the Federal Highway Administration in 1993.