IR–042
Los Angeles County, CA
Two decades of urban warming
- Finding
- +0.42 °C / decade
- Confidence
- 99.86%
NASA POWER / MERRA-2 · 2005–2025
DISASTER EVIDENCE / NEW INVESTIGATIONS
From a past event to a better question.
Trace sourced event evidence, inspect NASA precipitation context and explore a current point outlook.
2024 · Historical investigation
Nepal floods · September 2024
Investigate rainfall context and the documented warning timeline around the September 2024 floods. This point is not a measurement of the whole Bagmati basin.
Open investigation →2025 · Historical investigationCyclone Ditwah · Sri Lanka 2025
Examine rainfall context around Cyclone Ditwah and separate historical observations, archived reports and today's rainfall forecasts. Colombo is one sample point, not the cyclone's footprint.
Open investigation → Curated solved case files
Start with a story that already has evidence.
Each case bundles a representative request, source, and analysis state for an instant frontend demonstration.
IR–038
Central Valley, CA
The signature of a historic drought
- Finding
- −0.07 index / decade
- Confidence
- 98.2%
NASA POWER / MERRA-2 · 2001–2024
IR–031
Amazon Basin, Brazil
Forest resilience under pressure
- Finding
- −0.08 NDVI / decade
- Confidence
- 95.4%
MODIS MOD13C2 · 2003–2024
IR–024
Koshi Basin, Nepal
Himalayan snow retreat
- Finding
- −3.8% cover / decade
- Confidence
- 97.7%
MODIS MOD10C1 · 2001–2024
02 SRS Sequenced Guided Investigations (Pinned Real NASA Snapshots)
Step-by-Step Forensic Walkthroughs with Interactive Checkpoints
Each guided investigation uses a pinned real NASA dataset snapshot with verifiable SHA-256 provenance, three audience explanation levels, and interactive methodological checkpoints.
Core Definitions Used Across Globe & Case File
Hamed–Rao Mann–Kendall TestStatistical Test
A variance-corrected non-parametric trend test (Hamed & Rao, 1998) that accounts for serial autocorrelation in environmental time series by computing the effective sample size (n_eff).
Sen’s Slope Estimator (Theil–Sen)Effect Size
The median of all pairwise slopes between observations (xj − xi)/(tj − ti). Highly resistant to isolated outliers or extreme weather spikes.
Benjamini–Hochberg FDR (q-value)Multiple Testing
False Discovery Rate adjustment applied when scanning many grid cells on the Living Globe so false positives stay bounded at α = 0.05.
Effective Sample Size (n_eff)Autocorrelation
The number of statistically independent observations in an autocorrelated series. Positive lag-1 autocorrelation reduces n_eff below the raw month count N.
Evidence Grade (A–D)Verification
Deterministic quality rubric combining record length (>=15 yr), valid monthly completeness (>=85–90%), Hamed-Rao p/q significance, Sen/OLS concordance, and sensor caveat severity.
Trend TriangleRobustness
A matrix of Sen's slopes and p-values computed across all combinations of start year and end year to verify that a trend is not an artifact of endpoint selection.