NASA Space Apps Challenge 2026 · Seoul
Two satellites count the same fires very differently: VIIRS reports roughly 4.6 to 5 times more detections in Indonesia. We map both, month by month, and put them on one scale with a simple baseline, with its error stated. Underneath: forest loss, oil palm, peat and burned area.
What the app shows now
A full-screen map of Indonesia with real MODIS (2001 on) and VIIRS (2012 on) fire detections from NASA FIRMS, one month at a time, and harmonized monthly counts for each province. The harmonized counts come from a province-by-month ratio baseline: before 2012 we scale MODIS by the VIIRS/MODIS ratio of each province and calendar month. On years held out of the fit its error is 12.6% for the national monthly count and 28.9% for the median province. It is a baseline, not our model.
Under the fires you can switch on forest loss (Hansen), oil palm by planting year (Descals), peat, burned area, forest history since 1990 (JRC) and tree-plantation conversion. Forest alerts and pulpwood are still placeholder layers and say so.
The model research
We are training a shared encoder for MODIS and VIIRS on pairs of images taken minutes apart over the same ground. We built 767 such pairs. This is research in progress: the first experiments used a smaller set, and the model will replace the baseline on the map only if it beats it. Nothing on the map comes from it today.
What we found about forest, fire and palm
About 24% of the forest lost in Indonesia since 2001 was oil palm by 2021, mostly planted within three years of the loss. Forest inside a MODIS fire footprint was about 1.5 to 2 times as likely to become oil palm within ten years as similar forest without one. This is an association, not proof that the fire caused the planting; the size depends on how controls are chosen and on the sensor.
The team
AI / Robotics Engineer at CarbonSix
PhD in Aerospace Engineering, KAIST
Recognition