Introduction

tig2aprs reads US Census TIGER/Line(R) 1994 files and generates DOS/APRS maps. tig2aprs reads the collection of TIGER/Line files as a single standard input stream and creates one or more output maps plus one or more maplists. This is well suited to dealing with the TIGER/Line data on CD-ROM or even across the network since you can "unzip -p" the files directly from the CD and pipe them into tig2aprs.

About the US Census TIGER/Line(R) 1994 Files

What Are They

The best thing about TIGER/Line files is that they are in the public domain, so you can redistribute APRS maps made from them (unlike copyrighted maps such as those found with commerical packages such as Delorme Street Atlas -- which are generally derived from TIGER/Line in the first place!).

The map data in TIGER/Line come from US Geological Survey 1:100K Digital Line Graph (DLG) maps which are supplemented with USGS 1:24K, and US Census GBF/DIME and other sources.

Unlike USGS DLG's which are organized by 7.5 or 15 minute map quadrants, TIGER/Line's basic file unit is the county (or county-equivalent).

A new set of TIGER/Line maps appears to come out from the US Census every couple of years (they've published them in 1992, 1994, and expect to be releasing the next set in 3rd quarter 1996). Besides being used for the decennial census, other US government agencies have collaborated with the Census Bureau to augment the maps. For example, the US Postal Service added geographic linking of ZIP+4 codes.

Documentation

You will need to be familiar with that documentation to understand what I am babbling about when I talk about "record types" below.

Where to find TIGER/Line files

But, before you change channels, there's hope! Many libraries are official depositories of US government documents and data. There is likely a library near you that has a set of TIGER/Line CD's. Whether or not access to them is practical is another question. An example of a library that does it "right" is Columbia University's Electronic Data Service which provides access to the CDs to members of the general public on Internet-connected PCs. The CDs are not available to be borrowed but you can copy the files you care about onto floppies or FTP them somewhere. Check your local large public or research university library to see if they have a similar program in place.

Choosing TIGER/Line files of interest

36/tgr36119.zip

TGR36119.F61
TGR36119.F62
TGR36119.F63
TGR36119.F64
TGR36119.F65
TGR36119.F66
TGR36119.F67
TGR36119.F68
TGR36119.F69
TGR36119.F6A
TGR36119.F6C
TGR36119.F6H
TGR36119.F6I
TGR36119.F6P
TGR36119.F6R
TGR36119.F6S
TGR36119.F6Z

The complete list of state and county codes is in Appendix A of the TIGER/Line 1994 documentation file, app_a.asc.

Using tig2aprs

Order of Input Data Matters

The tig2aprs file input ordering requirements are:

• All record types 1 have to be read in first. These records are the key to deciding whether to ignore or save selected information from the other record types.
• All record types 2 are expected to be grouped by like TLID and sorted by seqno if there are multiple type 2 records for the same TLID. The TIGER/Line data is documented as being supplied with this ordering on the CDs.
• After type 2, the ordered set of expected record types is 'P', 'C', 'S', '7', '8', and '9'. Don't worry if you get the order wrong, tig2aprs will complain (or at least dump core:-).
You can omit some of the types. For example, type 2 records fill in the way points between the ends of road "segments" defined by type 1 records. You can make a more jagged map by not supplying type 2 records.

Example

$ cat domap
#!/bin/sh
#36119: Westchester (= ../*1, etc.)
#36087: Rockland
#36071: Orange
#36079: Putnam
#36027: Dutchess
#36111: Ulster
#36061: New York (Manhattan)
#36081: Queens
#36047: Kings (Brooklyn)
#36085: Richmond (Staten Island)
#36005: Bronx
#36059: Nassau
#36103: Suffolk
#09001: Fairfield
#09005: Litchfield
#09009: New Haven
#34003: Bergen
#34031: Passaic
#34013: Essex
#34023: Middlesex
#34039: Union

#more=${more:-""}
two=${two:-"2"}
counties="36005 36087 36071 36079 36061 09001 34003 36059 36081 $more"

for t in 1 $two P C S 7 8 9
do
# I have Westchester already unzipped in the .. directory
   if [ $t = 1 ]; then
     ../src/fixup <../*1
   else	
     cat ../*$t
   fi
# the other Counties' zip files are copied off CD into /other/tmp.
   for c in $counties
   do
     st=`echo $c|sed -e 's/^\(..\).*$/\1/'`
     if [ $st = 34 -a $t = C ]; then
	:
     else
        unzip -p /other/tmp/$st/tgr$c TGR$c.F6$t
     fi
   done
done | $*

$ domap tig2aprs -ao -r 16 -d 4 -t 4 -p nywc4 -l maplist.wc4 2>log.wc4

Data Reduction

The data reduction techniques tig2aprs uses include:

Feature Cutoffs

TIGER/Line identifies all features with a Census Feature Class Code (CFCC) which distinguish Interstate highways from dirt roads, for example. tig2aprs has a table of feature cutoffs to use based on the level of detail requested.

Water Filtering

There are many "trivial" water features such as small lakes and ponds that are unfortunately not properly identified as such with a CFCC. These features lead to noisy maps and waste precious data points. The tig2aprs water filter will drop all water features that cover less than a specified percentage of the map area.

Joining segments

Pieces of roads or segments (TIGER/Line jargon for these is Complete Chains) are joined together, eliminating unneccesary duplicate data points at each intersection.

Line Smoothing

tig2aprs includes an algorithm to "straighten" out wiggly lines, thereby eliminating intermediate wiggles and their associated data points.

Point Fuzzing

tig2aprs allows you to control the "focus" such that data points that are "near" each other are considered equivalent, allowing for more point elimination.

Tiling

If the above-mentioned point reduction techniques still fail to get down below 3000 points in a map, tig2aprs is able to recursively split a map into quadrants and so on until each resulting map covers a small enough area that the maximum points limit is achieved.

Seamless Map Transitions

Labels

Labels for areas (towns, villages, parks, etc.) are automatically centered within the visible portion of an area on a given map. Multiword labels (names longer than 12 characters) are broken up and stacked to get around the DOS APRS limit.

Runtime Options

Usage: tig2aprs [-vjnoaCTD] -c lat,lon -r range -d detail -t min_tile
       -f fuzz -F maxfuzz -w filt% -W maxfilt% -p map_prefix
       -s slopefuzz -S maxslope -l maplist -m maxmaplist -M maxAPRSpoints
       -R resolutuion -L places|landmarks|kgls 
 -v = verbose (more -v's for more verbose)
 -j = flip and join road segments
 -n = match segments by name as well as CFCC
 -o = make three overlapping maps
 -a = all the usual (same as -vjnC -w .05 -W .1 -s .05 -S .2)
 -T = print all a road's TLIDs in comment
 -D = create a *.dat file before fuzzing a map
 -c = map center lat,lon in decimal degrees
 -r = map radius in miles
 -d = level of map detail in miles
 -t = make tiles no smaller than this radius in miles
 -f = initial map fuzziness (reducing # of points)
 -F = worst map fuzziness
 -w = toss lakes smaller than x% of map
 -W = worst lake fuzz%
 -s = line smoothing factor
 -S = worst line smoothing factor
 -p = filename prefix for map files (default is 'map')
 -l = maplist filename
 -m = max map names per maplist before splitting up
 -M = max APRS/DOS map points (default 2999)
 -R = resolution (a/k/a pixels per degree)
 -L = label places, landmarks, or key geographic locations (kgls)

Customization

aprs.tk