PerryGeo

Thousands of Mexicans come to the United States every year and, besides the legal troubles of border crossing, they face a tough journey across the desert in order to reach their destination. They have very little information to go on and many die from dehydration as they attempt to find their way through the vast deserts of the american southwest.

A faith-based organization called Humane Borders is trying to help the situation. They have produced a number of maps documenting town locations, roads, water stations, walking distances, cell phone towers and even places where other immigrants have died along the way. This was made possible in part by GIS software donated by ESRI.

I heard today on CNN with Lou Dobbs that the Mexican government is now printing and distributing these maps to citizens. Though the maps will clearly state “Don’t Do It! It’s Hard! There’s Not Enough Water!”, critics are saying the maps aid criminals and will enourage illegal aliens to cross the border. Others have pointed out that, from an economic standpoint, this may benefit the US border patrol since so much of their budget is devoted to aiding sick and injured imigrants and properly taking care of the dead. Humane Borders is hoping to make people aware of the risks so that they can either choose not to go or be better prepared should they decide to cross.

In any case, it is an interesting example of how geographic information is still so important (and controversial) in our society.

Most commercial software comes with fairly elaborate geocoding engines and there are nice geocoding services on the web that can do one-at-a-time geocoding but the recent post at Spatially Adjusted pointed out a great free resource for batch geocoding named, conveniently enough, Batch Geocode. Just give it a list of tab or pipe delimited addresses and it outputs a table with your original data plus a lat/long for every row.

I have been working on a python script to convert text files into point shapefiles and thought this would be a great chance to put it to work. The only dependency is a recent version of python with the ogr module (see FWTools for an easy to install package for windows or linux).

First, I take a list of cities and feed it to batchgeocode.com (a very nice feature is that the yahoo geocoder, on which batchgeocode is based, does not require street level addresses):

City|State
Santa Barbara|CA
Arcata|CA
New Milford|CT
Blacksburg|VA

After running the geocoder, I get back a table with lat/longs:

City|State|lat|long|precision
Santa Barbara|CA|34.419769|-119.696747|city
Arcata|CA|40.866261|-124.081673|city
New Milford|CT|41.576599|-73.408821|city
Blacksburg|VA|37.229359|-80.413963|city

Copy and paste that into a text file and add a second header row that defines the data type for each column. It would be possible to autodetect the column types but there are cases where a string of numeric digits should be kept as a string (for instance the zipcode 06776 would become 6776 if it was read as an integer).The possible column types are string, integer,real, x and y with x and y representing the coordinates.

City|State|lat|long|precision
string|string|y|x|string
Santa Barbara|CA|34.419769|-119.696747|city
Arcata|CA|40.866261|-124.081673|city
New Milford|CT|41.576599|-73.408821|city
Blacksburg|VA|37.229359|-80.413963|city

Now run the txt2shp.py utility. The input and output parameters are self-explanatory and the d parameter defines the string used as a delimiter. Notice that the syntax follows the GRASS standard of parameter=value:

txt2shp.py input=cities.txt output=cities.shp d='|'

And now you’ve got a shapefile of the geocoded cities!

The txt2shp.py script can be downloaded here. Try it out and let me know how it’s working for you.

Update: In order to generate a .prj file for your output shapefile, you can use the epsg_tr.py utility if you know the EPSG code. Batch Geocoder returns everything in lat/long (presumably with a WGS84 datum?) so you can use EPSG code 4326:

epsg_tr.py -wkt 4326 > cities.prj