Pulling Fires for an Area
The Area API returns detections inside a bounding box as CSV. The country endpoint FIRMS also lists is marked as currently unavailable, so a box is the dependable choice.
The Request
/api/area/csv/[MAP_KEY]/[SOURCE]/[AREA_COORDINATES]/[DAY_RANGE]
/api/area/csv/[MAP_KEY]/[SOURCE]/[AREA_COORDINATES]/[DAY_RANGE]/[DATE]
| Part | What to pass |
|---|---|
SOURCE |
A source name such as VIIRS_NOAA20_NRT. Use VIIRS by default from here on. |
AREA_COORDINATES |
west,south,east,north in decimal degrees, or world |
DAY_RANGE |
1 to 5 days per request |
DATE |
Optional, as YYYY-MM-DD. Leave it out for the most recent data. |
Without a date, you get today back to today minus DAY_RANGE - 1, and with a date, you get that date forward to DATE + DAY_RANGE - 1. Coordinates must follow the west, south, east, north order, and FIRMS rejects words like north in their place.
Latest Detections
from firms import get_fires
# Southeastern Oregon and the Idaho border, west, south, east, north
BOX = (-118.3, 41.9, -116.4, 44.6)
latest = get_fires("VIIRS_NOAA20_NRT", BOX, day_range=2)
print(len(latest), "detections in the last 2 days")
print(latest[["latitude", "longitude", "acq_date", "acq_time", "confidence", "frp"]].head())
A Past Fire
The Big Grass Fire started during thunderstorms on July 23, 2026, in Malheur County, and it became the largest wildfire in Oregon history. The box above covers the county and the Idaho side of the border.
Older dates sit in the Standard Processing archive, so ask for an _SP source and run data_availability() first if you aren't sure a date has moved there yet.
first_days = get_fires("VIIRS_NOAA21_SP", BOX, day_range=5, date="2026-07-23")
Longer Periods
Each request covers at most 5 days. For a longer window, step through it in 5-day blocks and join the results.
import pandas as pd
from datetime import date, timedelta
def fires_between(source, box, start, end):
frames = []
day = start
while day <= end:
span = min(5, (end - day).days + 1)
frames.append(get_fires(source, box, span, day.isoformat()))
day += timedelta(days=span)
return pd.concat(frames, ignore_index=True)
july = fires_between("VIIRS_NOAA20_SP", BOX, date(2026, 7, 23), date(2026, 7, 31))
Every block is at least one transaction against your 10-minute limit, which is worth checking before you loop over a whole season.