1 · Choose an area
Draw a rectangle or polygon with the tools on the map, pick a country, or type coordinates.
No area selected yet.
Try one that is quick to load:
2 · Build the calendar
Selected month
Which months burn most?
Which years burned most?
Day by day
How the two sensors were made comparable
Before and after, yearly totals
Test: MODIS alone predicting VIIRS, 2012–2020
Method in brief
- Common unit. Everything is expressed in S-NPP VIIRS-equivalent hotspots: the number of detections the 375 m VIIRS instrument on Suomi NPP reports, or would have reported before it launched.
- Learned from the overlap. For this area, each other satellite gets a conversion factor from the months it flew together with S-NPP. Factors are fitted per calendar month, because small cool burns that 1 km MODIS misses dominate some seasons and large wildfires dominate others. Sparse months lean on the area's overall factor.
- Stable calibration window. MODIS factors use 2012–2020 only; Terra and Aqua have drifted to different overpass times since.
- Best instrument each day. When any VIIRS satellite is flying its detections are pooled; before 2012, Terra and Aqua are pooled and scaled. Every hotspot carries a weight, so maps, CSVs and the calendar add up to the same totals.
- Checked, not assumed. Each overlap year is held out in turn, the factors refitted, and MODIS asked to predict what VIIRS saw. The scores above are from those held-out years.
Known limits: dates are UTC; satellite outages of a few days are not gap-filled; near-real-time detections carry no fire-type label; hotspots count detections, not fires or burned area.