This code works on the bulk NCEI climate division files at the NCEI for counties. For an interactive way to view this data, try the County Time Series page.
This is not a particularly hard script to write, but it was somewhat tedious so hopefully this saves someone some time. I needed historical average temperatures for US counties, and I found the data but had to write this to get it into a useful dataframe.
It should convert files in that link that have the filenames:
climdiv-pcpncy-vx.y.z-YYYYMMDD
climdiv-tmaxcy-vx.y.z-YYYYMMDD
climdiv-tmincy-vx.y.z-YYYYMMDD
climdiv-tmpccy-vx.y.z-YYYYMMDD
climdiv-cddccy-vx.y.z-YYYYMMDD
climdiv-hddccy-vx.y.z-YYYYMMDD
In addition to converting from a fixed width formation to CSV, the script creates a 5 digit FIPS that combines state FIPS and county FIPS. This is available as an integer (4 or 5 digits) as FIPS_int (5001) as as five character string with leading zeros ("05001") in FIPS_str. The combined state+county FIPS code should make it easier to merge into other data sources that have a five digit FIPS.
The element (and other columns) are described in the county readme, but is generally the same for every row for the same filename (tmpccy files wil be all be a "2" element code, for example.)
01 = Precipitation
02 = Average Temperature
25 = Heating Degree Days
26 = Cooling Degree Days
27 = Maximum Temperature
28 = Minimum Temperature
The script requires pandas. If you want to export to parquet, you'll need something like pyarrow or a parquet backend for pandas.