Burn Calendar

One consistent fire-hotspot record from NASA's MODIS and VIIRS sensors, January 2001 to December 2025.

Or a point and radius
km

Which months have the most fires?

Which years were the worst?

Is unusual?

Pick an area

Drag on the map to draw a box. Click a county to select it. Cell colour shows hotspots in the chosen years.

Burning activity calendar

Click a month to test it. Values are VIIRS-equivalent hotspots.

Before and after harmonization

Harmonized recordNaive join: raw MODIS to 2011, raw VIIRS after

Each year against the multi-year average

Seasonal pattern

Bars show the median for each calendar month, whiskers the middle half of years.

Median month

Download

The calendar file has one row per month with the harmonized value, its 90% range, the raw count from each satellite, the median and the anomaly. The hotspot file has the same counts for every 0.1° grid cell in the selected area.

How the two sensors were made comparable, and the evidence

Held-out test. Each year from 2012 to 2025 was predicted from MODIS alone, using exchange rates learned from the other years, then compared with what VIIRS saw.

VIIRS observedPredicted from MODIS
  1. Keep vegetation fires only, at nominal or high confidence, from MODIS Collection 6.1 (Terra, Aqua) and VIIRS 375 m (Suomi NPP).
  2. Remove industrial heat sources: 500 m spots that VIIRS flags in 5 or more years, mostly at night.
  3. Count hotspots per month in each county and 0.1° cell.
  4. On the overlap, February 2012 to December 2025, learn one exchange rate per calendar month: VIIRS hotspots divided by MODIS hotspots.
  5. From 2012 use VIIRS as observed. Before that, multiply the MODIS count by the rate. Before July 2002 only Terra flew, so a Terra-only rate applies.
  6. The 90% range on converted months comes from the held-out errors, and is wider for small counts.

Limits. The record ends in December 2025; 2026 needs a FIRMS key. Converted months are estimates, and small areas are noisy: MODIS misses many small fires, so a converted zero can hide real burning. Satellite outages and cloud cover are not corrected. A hotspot is a detection, not a fire count or burned area. Data: NASA FIRMS.