earths-hidden-signals

Live site: https://earths-hidden-signals.vercel.app

Earth's Hidden Signals: decoding NASA data before disaster. Our project for the NASA Space Apps Challenge 2026 challenge Be An Earth System Trend Detective!

The same global warming shows up differently across South Asia: heavier rain in some places, drying in others, more heat everywhere. We use NASA Earth observation data to show what is changing, where, by how much, and whether the change is statistically significant. Then we link those trends to the floods, landslides and wildfires that follow.

What's inside

Folder	What it is
web/	Next.js + TypeScript website: Overview, Map Explorer (NASA GIBS imagery + Landsat forest loss), Trend Analysis, Hazard Signals, Methods
analysis/	Python pipeline: downloads NASA records, runs the trend tests, writes JSON for the website

Data

• Temperature: NASA GISTEMP v4 (GISS)
• Rainfall: GPCP v2.3 (NASA GSFC / NOAA)
• Map imagery: NASA GIBS, from MODIS land surface temperature, MODIS NDVI, MERRA-2 and GLDAS
• Forest loss: Hansen/UMD Global Forest Change, built from NASA/USGS Landsat
• Hazard events: NASA FIRMS MODIS fires, NASA Global Landslide Catalog, GDACS flood alerts

Method

• Autocorrelation-corrected (Hamed–Rao) Mann–Kendall test for significance
• Sen's slope with a 95% confidence interval for the rate of change
• Benjamini–Hochberg false-discovery-rate control for the map-wide results

Details are in analysis/README.md.

Key findings (1981–2025)

• Warming everywhere: all 234 land cells warmed significantly, by about +0.31 °C per decade (+1.4 °C since 1981).
• Monsoon rain splits: the Indus plain got wetter, while the Bengal delta, Central Himalaya, Western Ghats and Northeast hills got drier. Most cells show no detectable trend.
• 2025 preparedness signals: Indus plain floods and Western Himalaya landslides. In both regions, 2025 monsoon rain resembles past high-event years (evidence for readiness, not a forecast).

Run it

# analysis (optional: results are already in web/public/data)
cd analysis
pip install -r requirements.txt
python build.py
python build_hazards.py

# website
cd web
npm install
npm run dev    # http://localhost:3000

This project shows evidence-based trends to support preparedness. It is not a disaster prediction system.