Live Data

Be An Earth System Trend Detective!

Real NASA data. Pick a place and find out what’s actually changing.

Global Trends

Explore the Planet

Click the map to analyse a place

Key Insights

Compare Regions

Open

Future Outlook

Explore Data

Discover Environmental Changes

Visualize NASA satellite data and explore trends across the globe.

Time Range:

Quick Actions

Full report

Time series and trend

Signal or noise?

Bright line: this place. Grey: fake worlds with the same noise and no trend.

Trend, season and noise

Trend, seasons, and what’s left.

Future outlook

Forecasts are ranges, tested on past years before they are shown.

10-year outlook

Ensemble median with 50% and 90% ranges

Back-test: forecasts vs what happened

Each orange line: a 5-year forecast from its start year.

Climate scenarios to 2100

NASA NEX-GDDP-CMIP6 climate models under two emissions pathways.

Compare Regions

Compare Our Planet

Select two regions to compare environmental trends and see the difference.

Region A
VS
Region B

NDVI Comparison

Key Differences

Side-by-side statistics

All five datasets, same tests
Insights

Key Insights

Explore important findings from NASA data and research.

Trend Detective

One cause can push two measurements in opposite directions.

ScanEvery location and every pair of variables.
FilterKeep pairs where both trends pass the significance test.
ContrastRank pairs that move in opposite or surprising directions.
LinkMeasure how closely the two records move together, with delays.
ExplainMatch the pattern to a known driver from published research.

Both records, year by year

Annual means, each on its own axis

How tightly linked?

Does one lead the other?

Does the second record follow the first, 0–6 months later?

About

About EarthScope

Is the change real, and how big is it? EarthScope answers that from open NASA data. Made for NASA Space Apps 2026.

Team

Datasets

Trend and significance

1. Test. Months become yearly averages. The Mann–Kendall test asks whether they keep rising or falling.

S = Σi<j sign(xj − xi) · Z = (S − sign S) / √Var(S)

2. Correct for memory. Wet years follow wet years. We widen the test for that, so memory doesn’t pass as a trend.

neff = n (1 − r1) / (1 + r1)

3. Measure. Sen’s slope gives the rate, with a 95% range.

β = median[(xj − xi) / (tj − ti)]

4. Map. Benjamini–Hochberg keeps false alarms near 5%.

Forecasts

Ensemble. Five models forecast each record. Each is tested on years it never saw.

Skill gate. A model that can’t beat the recent average is dropped.

skill = 1 − MAEmodel / MAEclimatology

Ranges. The bands come from how wrong past forecasts were.

Beyond ten years. A trend line can’t see 2100. NASA’s climate projections can. They’re next.

Limits

  • Vegetation and rain start in 2010. Sixteen years is short.
  • MODIS is retiring. VIIRS can take over.
  • Two trends moving together suggest a cause. They don’t prove one.
  • Rain data is weaker over snow, mountains and coasts.