HOTSPOT
Satellite-proxy model inferring nutrient-driven ecological change in marine environments from remote sensing and time-series data.
- Won the Stockholm Regional Water Prize
- 3rd at regionals, and advanced to the state science fair
- Constraint filters and time-series interpolation to reconstruct missing data
The problem
Nutrient pollution changes marine ecosystems faster than field sampling can measure it. Satellites see the whole ocean, but their record is full of gaps, and a forecast built on interpolated nonsense is worse than no forecast.
What I built
Ecological constraint filters and local time-series interpolation that reconstruct missing data while enforcing physically consistent predictions, so the model cannot invent states the ocean could not actually be in.
What happened
Focused on reducing false positives so environmental forecasts stay usable in real monitoring settings. Won the Stockholm Regional Water Prize, placed 3rd at regionals, and advanced to the state science fair.