Get Weather Data

Historical daily weather for any US ZIP code or latitude/longitude.

Prefer clicking to coding? A Streamlit app gives the same data

result = weather . get (( 40.7484 , - 73.9967 ), "2024-01-15" )

phenomena (fog, thunder, hail, freezing rain, ...) from GHCN WT**

it (or a glob of many files) with query_weather(...) via DuckDB

df = weather . get_frame ( "90210" , "2024-07-01" , "2024-07-07" )

# one row per day; columns = metadata + tmax/tmin/prcp/... in your units

directly queryable) and run SQL over them with DuckDB. Needs the

# Streaming Parquet output (inferred from the .parquet suffix; no pandas

Weather (). process_csv ( "locations.csv" , "weather.parquet" , date_column = "date" )

the same on the CLI. The daily and hourly DataFrames also write Parquet

cov = weather . coverage ( "59718" , "2024-01-01" , "2024-12-31" )

print ( cov . station_name , cov . station_distance_meters , "m away" )

hours = weather . get_hourly ( "11371" , "2023-07-16" ) # LaGuardia, one UTC day

print ( h . observed_at , h . temp , "F" , h . wind_speed , "mph" , h . wind_direction , "deg" )

df = weather . get_hourly_frame ( "11371" , "2023-07-16" , "2023-07-17" )

CLI: get-weather hourly 11371 2023-07-16 [--end 2023-07-17] [--units imperial] .

With include_flags=True , result.flags maps each field to its GHCN

quality-control flag (blank = passed all checks; GHCN stations only).