Storm reports¶
Severe-weather reports (hail, wind, tornado) live on client.storm_reports. They need the
read:observations scope, like observations.
from stormlore import Client
client = Client()
catalog = client.storm_reports.catalog()
catalog.report_types # e.g. ["hail", "tornado", "wind"]
catalog.to_dataframe() # one row per source, with its span
Reports in a box¶
plains = {"west": -104, "south": 30, "east": -90, "north": 45}
page = client.storm_reports.list(
plains, start="2026-05-01", end="2026-06-30", report_type="hail"
)
page.to_dataframe()
list is paged exactly as observation listings: one
request per call, cursor=page.next_cursor for the next page, or iter_list to walk them:
for page in client.storm_reports.iter_list(plains, start="2026-05-01", report_type="hail"):
...
Filters: start, end, report_type, state (two-letter code), source, limit. The
box may not cross the antimeridian.
Each report is an Observation; what the source wrote (report_type, magnitude, county,
remarks …) is in properties and becomes columns of the frame, served as written.
One report¶
report = client.storm_reports.get(page[0].id)
An unknown id raises ObservationNotFoundError.
Counts per county¶
counts = client.storm_reports.by_county(plains, start="2022-01-01", report_type="hail")
counts.to_dataframe() # fips, county, state, total, then one column per year
With no start, the window is the 2 years before end (default now). The span may not exceed
5 years: a longer one answers 400 (InvalidRequestError).
A county with no reports in a given year has 0 in that year's column.
Draw it¶
page.plot() # a map, one colour per source
Every chart, drawn: charts and maps (pip install 'stormlore[viz]').