This folder holds two zip archives of the challenge data, for anyone who cannot install a command-line tool:
Direct links (plain HTTP, no account):
Each archive has the full project README at its top level, and unzipping it yields the same folder tree as the bucket (reference/<region>/..., strata/national/..., strata/<region>/...).
Browser downloads of multi-GB files do not resume, so if one fails partway the archive is corrupt and has to be fetched again: check the byte count against the size shown in the file browser before unzipping.
If you can install the aws CLI, use it instead of the zips. It needs no account, resumes if the connection drops, and re-running it only fetches what has changed, so a re-issued file is a one-line update instead of a fresh multi-GB download. The steps are below; they are the same ones in the project README.
Install the aws CLI if you do not have it (aws --version tells you). Any one of these:
No aws configure step. The bucket is public and --no-sign-request below tells the CLI not to look for credentials.
Open a terminal in the project folder you are using for the challenge. The files land in the directory you run the command from (./), so cd there first.
Run the line for the region you want. Point the CLI at the bucket directly, no --endpoint-url.
Or everything at once, into a folder tree that mirrors the bucket:
One layer across all regions also works, because the self-describing names make --include match at any depth:
Sizes per region, reference/ only: Maricopa about 1.0 GB, Northern CA 0.45 GB, Eastern OK 0.9 GB, South Texas 3.9 GB. The strata tables (parquet + csv) add about 1.7 GB, and the boundaries file is small.
Check the byte counts, not just the file names. A parallel download can exit cleanly and still leave truncated files behind. Compare local sizes to the bucket listing:
Any file that is smaller locally is incomplete. Re-running the same aws s3 sync line fixes it: sync compares size and timestamp and re-fetches only the files that differ.
None of the query examples in the project README need a local copy: DuckDB, GeoPandas and the aws CLI all read straight from the bucket. See the project README, section "Access".