How it works
Five satellites have watched Myanmar's fires, each with a different eye. Mee turns them into one record you can compare year to year.
- 1
Collect every fire detection
Mee downloads all active-fire detections over Myanmar from NASA FIRMS: MODIS on Terra and Aqua since 2000, and VIIRS on Suomi NPP, NOAA-20 and NOAA-21 since 2012.
- 2
Keep only real vegetation fires
Detections flagged as gas flares, volcanoes or other static heat are removed. Low-confidence detections are dropped. Repeats are counted once.
- 3
Put every satellite on one scale
VIIRS sees far more small fires than MODIS. On days when both fly, Mee measures the gap for each state and month, then scales every satellite to match NOAA-20.
- 4
Build the calendar
The scaled counts are added up per township, per day. When several VIIRS satellites cover a day, Mee uses their average. Before 2012, scaled MODIS fills in.
- 5
Define normal
For each date, Mee looks at the 7-day total on the same dates in every past year. The middle value is normal. Above the 75th percentile is above normal. Above the 90th is far above normal.
Limits
- Satellites only see fires burning when they pass over. Cloud and smoke can hide fires.
- One count is one hot pixel, not one fire. A large fire can show as many pixels.
- Terra and Aqua will stop working in 2027, and Suomi NPP data stopped in November 2026. The scaling step is what keeps the record consistent after that.
Data
- NASA FIRMS active fires: MODIS Collection 6.1 and VIIRS 375 m (Suomi NPP, NOAA-20, NOAA-21). firms.modaps.eosdis.nasa.gov
- Township and state borders: geoBoundaries, CC BY 4.0, from the Myanmar Analytics Project. geoboundaries.org
- Code is open source under the MIT license. github.com/thesithunyein/mee