converting ZIP codes to latitude and longitude

Latitude and longitude are calculated as population-weighted averages of the latitude and longitude of individual blocks. Similar methods can be used to obtain the average latitude and longitude of other geographic areas, for example block groups.

The example given is for a UNIX computer which has an "unzip" utility and "cut" program for extracting files from a comma-delimited file. (At Lawrence Berkeley National Laboratory (LBNL), type "module load unzip" to access unzip.)

• USE FTP to download the compressed file ??.zip for one state, for example ak.zip for Alaska.
• Unzip the file and then use "cut" to extract the ZIP code, population, latitude and longitude of each block; then delete the original downloaded file ak.zip.

  unzip -p ak.zip |cut -f32,34-36 -d, >ak.tmp; rm ak.zip
  

• Obtain a sorted list sortzip[j] of unique zip codes that are present in the input file ak.tmp. Define vectors nrecs[j], sumpop[j], sumlat[j] and sumlong[j]. In each case j=1,2,...n where n is the number of unique zip codes. Initialize nrecs, sumpop, sumlat, and sumlong to zero.
• For each line i in ak.tmp, do the following:
  1. read a line containing zip[i], pop[i], lat[i], and long[i].
  2. rlat=lat[i]-lat[1]; rlong[i]-long[1]. To avoid rounding errors, latitude and longitude values relative to those of the first record are used.
  3. find the value of j such that zip[i]==sortzip[j].
  4. nrecs[j]=nrecs[j]+1
  5. sumpop[j]=sumpop[j]+pop[i]
  6. sumlat[j]=sumlat[j]+pop[i]*rlat[i]
  7. sumlong[j]=sumlong[j]+pop[i]*rlong[i]
  For each value of j from 1 to n, do the following:
  1. if sumpop[j].gt.0 {sumlat[j]=sumlat[j]/sumpop[j] + lat[1]}
  2. if sumpop[j].gt.0 {sumlong[j]=sumlong[j]/sumpop[j] + long[1]}
  3. to an output file ak.out, write a line containing sortzip[j], nrecs[j], sumpop[j], sumlat[j], and sumlat[j].
• The output file ak.out contains five columns: zip code, number of blocks, population, latitude (population-weighted), and longitude(population-weighted).
• A perl algorithm for producing a population-weighted file of zip code coordinates has been kindly contributed by Dan Kegel (dank@alumni.caltech.edu).