Pick a province, enter a point, or draw an area. The calendar snaps to whole 0.25° cells (about 25 × 25 km).
—
—
Burning-activity calendar
Color = nivel del GeoFuego Index 3.0 que el motor hubiera indicado ese mes (sequía acumulada ERA5 por zonas de fuego + fase ENSO vigente entonces, con la temporada de cada provincia) · número = focos del histórico. Hindcast: 2012–2024 es el período de calibración; desde 2025 es fuera de muestra; desde noviembre de 2026, prospectivo.
Counts are satellite fire detections expressed in VIIRS-equivalent pixels (375 m). They are not a number of fires and not burned area. A fire that burns for several days is counted on every satellite pass.
From 2012 the values are observed by VIIRS. Before 2012 they are estimated from MODIS, a coarser sensor, and shown with an 80 % range: about 1 month in 5 falls outside it.
MODIS missed many small fires that VIIRS detects. In an area this size, a month with no value before 2012 means “MODIS detected nothing”, not “nothing burned”. Do not compare how often it burned before and after 2012.
This month vs the historical median
In 2012 the satellite changed. The record should not jump with it.
Table · same numbers, downloadable
How well it works — tested on years the model never saw
Country-level tests, out of sample. Factors fitted on 2012–2023 only; 2024 and 2025 were used neither for the factors nor for the range. Details and every script: Method & code.
Monthly series with 80 % range
Which years burned most
Which months of the year
Where it burns — share of hotspots by province
Fixed heat sources (gas flares, industry) — separate series, VIIRS counts since 2012
Never added to the calendar above. In the sensor's own units: VIIRS counts since 2012; the MODIS-era values are not shown.
How to read this
- Unit: the 0.25° cell. Drawn areas snap to whole cells; an area smaller than one cell becomes that cell.
- Size tiers: ≥ 100 cells full calendar · 25–99 full, pre-2012 labelled reconstructed · 4–24 baseline 2012–2025, pre-2012 zero = no MODIS detection · 1–3 pre-2012 greyed and excluded unless you toggle them on. Chubut and Santa Cruz: one tier stricter. Tierra del Fuego: from 2012 only.
- 80 % range: calibrated out of sample per fire-regime region, size class and predicted count, widened in strong ENSO months and in cloudier-than-usual months. Only reconstructed months carry it.
- Markers: ▲ cloudier than usual at overpass time (the count may be low) · ◆ partly modelled (satellite outage) · hatch: Terra-only era (Nov 2000 – Jun 2002), shown for ≥ 100 cells.
- Preliminary years (from 2026, marked *): near-real-time, NOAA-20/21 converted to S-NPP-equivalent (± 5 %), month to date in the current year. They are not part of the harmonized record and never enter medians or rankings: a year joins the record only when NASA publishes its reprocessed archive and the harmonization is re-run. This is not an alert system: an empty cell does not mean there is no fire.
