How many streets are there in the USA?

Hi All,

I'm about to run through all 24GB+ of TIGER Line files and convert the
data to a different format. The size of my target data structure can
be computed as a function of the number of streets in the TIGER data.

Does anyone have a rough idea of how many streets/roads/etc. to
expect? I need to know so I can determine whether my process will have
enough virtual memory to hold the entire data structure or if I will
have to do some trickery.

TIA,

-Le Chaud Lapin-

Somewhere between 30 and 40 million streets.

Yup.

I just processed District of Columbia and all 50 US states except
Connecticut, because the table I copied from a US government web site
with state codes was missing CT. Unfortunately, I can't find the
link, but I am pretty certain CT was missing because I checked and
rechecked when I noticed the fault.

Anyhow, after processing DC and these 49 states, looking for cases
where census feature class code (CFCC) began with 'A', I got around
27.1 million streets, which would, indeed, make the answer 30-40
million streets if Connecticut were included, along with Puerto Rico
and U.S. territories, taking into acount that CFCC 'A' is not the only
indicator of an addressable feature.

I guess I see why people like GIS so much. There are so many data-
intensive things you can do with it, and it seems relevant to just
about everything. ;)

Much thanks.

-Le Chaud Lapin-

Found it: http://www.itl.nist.gov/fipspubs/fip55-3.htm

Look at Table 1 in this document. Notice that Connecticut is missing.

Or it could just be my misreading.

I find it hard to believe that such an important omission would be
overlooked for so long.

-Le Chaud Lapin-

It is not overlooked. The GNIS people have been aware of it for quite
a while. The erroneous data is simply stale and will not die. They
pointed me to the correct list at USGS:

domestic, and then either download (scroll to FIPS 55), or click
search for interactive retrieval. "

-Le Chaud Lapin-