[Mapserver-users] Deriving TIGER extents

[Marvin Humphrey]

Greetings,

My ultimate goal is to generate a few thousand maps via offline batch 
processing, using TIGER data.  I don't need to set up a production 
server, so performance isn't an issue at this time and I'm not worried 
yet about converting to shapefile format for improved speed.

Unfortunately, I can't even get one TIGER file to display. I managed to 
grok GRASS, so I *ought* to be able to figure this out.  But I'm stuck. 
  It seems that I have some sort of extents problem. Maybe I just have 
my numbers wrong, but if so, I'm not sure how to correct them or to 
ensure that I get them correct in the future.

To start with, I downloaded a random county in Michigan and hacked up a 
bare-minimum .map file (see below).  Then I set up a LAYER according to 
the instructions at: http://mapserver.gis.umn.edu/doc36/ogr-howto.html

According to http://mapserver.gis.umn.edu/doc40/mapfile-reference.html 
Mapserver will attempt to derive the extents of a map if none are 
specified in the .map file. However, when I tried to omit the extents 
from the .map file, the browser displayed this error message:

  msCalculateScale(): General error message. Invalid image extent.
  msCalculateScale(): General error message. Invalid image extent.
  mapserv(): Web application error. No way to generate map extent.

Can it be that TIGER files have no extents data in them? Skimming the 
technical spec of the TIGER format didn't reveal anything obvious 
(http://www.census.gov/geo/www/tiger/tiger2002/tgr2002.pdf).  So how do 
people actually use them?  Does the tiger2shape utility figure out what 
the most extreme elements are in the file and define the extents that 
way?  Is there some sort of auxiliary data repository at www.census.gov 
that I have yet to uncover? Trial and error?

What would be great is if I could find a utility like ogrinfo that can 
print the extents to STDOUT.  Then I could grep the output to snag only 
what I need and autogenerate a bunch of map files using Perl.  It seems 
like ogrinfo ought to do what I need.  The ogrinfo manpage has an 
example from an NTF file illustrating typical output

% ogrinfo -ro -where 'GLOBAL_LINK_ID=185878' wrk/SHETLAND_ISLANDS.NTF 
BL2000_LINK
INFO: Open of `wrk/SHETLAND_ISLANDS.NTF'
using driver `UK .NTF' successful.
Layer name: BL2000_LINK
Geometry: Line String
Feature Count: 1
Extent: (419794.100000, 1069031.000000) - (419927.900000, 
1069153.500000)
...

Additionally, the manpage indicates that with the -q flag enabled, the 
output will include extents:

        -q:
            Quiet verbose reporting of various information, including
            coordinate system, layer schema, extents, and feature count.

Unfortunately, when I run ogrinfo specifying any one layer of the TIGER 
datasource and -q, I get a torrent of output, which seems to include 
coordinates for each individual element in the datasource... but no 
summary data.  When I run with the -summary option, all I get is a 
couple line breaks then the layer name:

# ogrinfo -ro -q -summary TGR26049 CompleteChain

Layer name: CompleteChain

(FWIW, I am specifying the directory holding all the related files and 
not an individual TIGER file, so I'm not being bitten by that bug.)

And when I run ogrinfo with just the -ro flag -- (read-only, otherwise 
I get an error saying ERROR 4: Tiger Driver doesn't support update. Had 
to open data source read-only.) -- all I get is a list of layers.  No 
extents.

# ogrinfo -ro TGR26049
INFO: Open of `TGR26049'
using driver `TIGER' successful.
1: CompleteChain (Line String)
2: AltName (None)
3: FeatureIds (None)
4: ZipCodes (None)
5: Landmarks (Point)
6: AreaLandmarks (None)
7: Polygon (None)
8: PolygonCorrections (None)
9: EntityNames (Point)
10: PolygonEconomic (None)
11: IDHistory (None)
12: PolyChainLink (None)
13: PIP (Point)
14: TLIDRange (None)
15: ZeroCellID (None)
16: OverUnder (None)
17: ZipPlus4 (None)

Stymied!  Very well, I thought, before I write to the list I'll try to 
get one county to display -- then I'll inquire about tools which I can 
use to derive extents automatically.  I'll hack away at the file until 
I get something to display... Nope, nothing doing!

I read in a few archived emails that the extent must be specified in 
the same units as the datasource.  In the case of TIGER, that's 
apparently decimal degrees (according to 
http://www.census.gov/geo/www/tiger/tigermap.html#TOWN ). Since I'm not 
sure which county in Michigan I downloaded, I entered LatLong for the 
entire state. Even if the county was small, the CompleteChain layer 
ought to be visible.  But the map was solid white.

Eventually I know I will need to get into joining multiple counties in 
a con mapfile, but for now, I'd just like to get a single file to 
display.

Any suggestions?

-- Marvin Humphrey


MAP
IMAGETYPE      PNG24
SIZE           400 300
SHAPEPATH      "data"
FONTSET        "fonts/fonts.list"
SYMBOLSET      "symbols/symbols35.sym"
UNITS DD

# Start of LAYER DEFINITIONS 
---------------------------------------------

EXTENT -91 41 -81 48
#Longitude: 82° 26' W to 90° 31' W
#Latitude: 41° 41' N to 47° 30' N

LAYER
     NAME "tigertest"
     TYPE line
     CONNECTIONTYPE OGR
     CONNECTION "../TGR26049"
     DATA "CompleteChain"
     STATUS ON
END

END # end of map file