Calorbis is not an official emergency service. Always follow guidance from your local authorities and emergency services.
Every hazard layer on the map, with its real provider, resolution, refresh cadence, coverage, license, and known limitations - written directly from the code that produces it, not marketing copy. This page exists because a platform combining observations, forecasts, satellite detections, experimental models, news, and community reports has to say which is which before it can honestly call itself dependable.
| Layer | Type | Provider / product | Resolution | Update cadence | Coverage | License | Known limitations |
|---|---|---|---|---|---|---|---|
| Earthquakes (M2.5+, 24h) | Observed | USGS Earthquake summary GeoJSON feed | Point epicenter (network-dependent precision) | ~5 min | Global, but seismic network density varies by region | US Government public domain | Preliminary magnitude/location can be revised after initial detection. |
| GDACS alerts (flood, cyclone, volcano, drought, wildfire) | Observed | GDACS (UN OCHA / European Commission) Event list API | Single representative point per event | ~5 min, provider-driven expiry (event dropped once GDACS marks it ended) | Global, limited to events meeting GDACS’s own significance threshold | GDACS terms of use (disaster-awareness use) | A regional event (e.g. a multi-country drought) currently renders as one point, not its real extent - tracked as CAL-108. |
| Extreme heat/cold, high wind, heavy precipitation, UV index (7-day) | Forecast | Open-Meteo (blends ECMWF/GFS/ICON model output) Hourly forecast API | ~9-11km model grid | 6h cron sweep | A curated ~4,000-city catalog, not a full grid sweep | Open-Meteo free tier | Forecast uncertainty grows with lead time; not shown as a probability. |
| Severe-convection environment (7-day) | Experimental | Open-Meteo (CAPE from the same forecast blend above) Hourly forecast API | ~9-11km model grid | 6h cron sweep | Same ~4,000-city catalog | Open-Meteo free tier | Raw CAPE, not a calibrated hail-probability forecast - see CAL-79. Renamed from "hail risk" for this reason. |
| Temperature / precipitation (right now) | Observed | Open-Meteo Current-conditions endpoint | Nearest model grid cell to each catalog city | 6h cron sweep | Same ~4,000-city catalog | Open-Meteo free tier | A snapshot at last cron run, not a live per-second reading. |
| Unhealthy air quality (PM2.5) | Observed | OpenAQ Aggregated sensor network API | Point station - marker drawn at the station itself, which can be up to 25km from the catalog city it was looked up for (CAL-107) | Live-fetched, throttled to ~5 requests/batch | ~35 selected cities across 17 countries (free-tier rate limits, not a global sweep) | OpenAQ (CC BY 4.0) | Coverage reflects rate-limit-driven city selection, not where air quality is worst. |
| Active fire detections (satellite) | Satellite | NASA FIRMS (VIIRS) Area CSV API | 375m pixel | Hourly cache (source updates per satellite overpass) | Global, filtered to high-confidence detections with FRP ≥ 50MW, top 300 by power | NASA FIRMS (public, attribution requested) | A pixel-center thermal anomaly, not a confirmed fire boundary or perimeter - can include non-fire heat sources; cloud/smoke cause missed detections. See CAL-70’s sibling finding on this pattern. |
| Coastal storm surge (US stations) | Observed | NOAA CO-OPS Tides & currents API | Point tide station | ~15 min | US coasts/territories only | NOAA public domain | No global coastal coverage - NOAA’s long-term network is US-only. |
| Coral bleaching heat-stress alert | Satellite | NOAA Coral Reef Watch Virtual station daily product | ~5km grid, represented as station points | Daily (12h cache); tiered freshness - current <36h, delayed 36-60h, hidden past 60h (CAL-106) | Fixed set of monitored reef locations, not every reef worldwide | NOAA public domain | A rolling 7-day heat-stress maximum, not confirmation that bleaching has occurred. |
| Lake surface water temperature (curated largest lakes) | Satellite | Copernicus Land Monitoring Service (University of Reading / ESA CCI LAKES) Lake Surface Water Temperature NRT v1 (SLSTR L3C, 10-daily 1km global mosaic) | 1km (0.01°) grid, but only within the product’s own ESA CCI LAKES mask (~2,000 predefined lakes worldwide) - a point outside that mask returns no reading, not a zero | 10-daily (a dekad); this pass checks daily for the freshest available dekad, since the product’s own "within 3 days" publication claim doesn’t always hold (a real ~20-day gap was observed live 2026-07-31) | Top 1,500 largest lakes by area from the HydroLAKES catalog (see waterBodies.ts) - about 40% return a valid reading in a typical dekad (591/1,500 on the 2026-07-11 to 2026-07-20 dekad), the rest are outside the product’s own lake mask or cloud/quality-masked that period - a real, reported gap, not papered over | Copernicus Land Service (free and open) | A 10-day average skin temperature, not an instantaneous reading. Quality-filtered (only "low quality" or better per the product’s own QLEVEL band is kept) but not independently recalibrated. Feeds into the experimental land-weather stress proxy below as an additional signal where a matched lake has a current reading, see CAL-70 phase 3. |
| Low dissolved oxygen (observed sensors, US) | Observed | USGS Water Services instantaneous-values API | Point sensor | Live-fetched, ~1h cache | 8 warm-water US states (FL, TX, LA, MS, AL, GA, SC, CA) | USGS public domain | US-only; sensor calibration/QC is USGS’s, not independently verified by Calorbis. |
| Experimental land-weather stress proxy | Experimental | Calorbis (derived) Composite of 3 land-weather Calorbis layers, plus a 4th real satellite lake-temperature enrichment where available (see below) | City-point | Inherits its inputs’ mixed cadence (6h to ~10 days) - flagged in the popup when signals of different ages combine | Wherever its 3 land-weather input signals fire at the same location; the lake-temperature enrichment only reaches locations that both match a known lake AND that lake has a current reading (~40% of the curated lake list per dekad) | Internal derived signal, no formal license | Not a fish-kill or hypoxia forecast (CAL-70) - primarily built from rainfall/river-discharge/soil-saturation signals at land-based city coordinates, with no dissolved-oxygen, salinity, ocean, or ice-cover data. CAL-70 phase 3 added a real lake-surface-temperature reading as an enrichment (never a standalone trigger) where a matched lake has one - see the lake-temperature layer above. Model-validity decision pending, see CAL-70. |
| High soil-moisture / runoff risk | Experimental | Calorbis (derived from Open-Meteo soil-moisture + precipitation) Hourly forecast API, aggregated to a 30-day trend | ~9-11km model grid | Cron-sharded across 4 runs/day, ~24h full-sweep cadence | Same ~4,000-city catalog | Open-Meteo free tier (Calorbis-derived threshold logic on top) | A modelled trend, not a direct soil-moisture measurement. |
| River gauges (observed, Germany) | Observed | Pegelonline (German federal waterways service) REST API | Point gauge | ~30 min | 5 major German rivers (Elbe, Rhein, Donau, Weser, Main), 8 stations each | Pegelonline open data | Germany only - not a substitute for a global observed-gauge network. |
| River flood risk — GloFAS ensemble (curated major rivers) | Forecast | Copernicus Early Warning Data Store (EWDS), GloFAS v4 operational forecast Real 51-member ensemble (50 perturbed + 1 control), 15-day daily river-discharge forecast | 0.05° (~5.5km) grid, discharge is model-simulated, not a gauge reading | ~12h (GloFAS itself refreshes ~2x/day) | 20 curated, real, named major rivers at real cities (e.g. Danube at Budapest, Mississippi at Memphis, Mekong at Phnom Penh) - not the full global river network. See the Open-Meteo fallback below for everywhere else. | Copernicus/EWDS licence (CEMS-FLOODS datasets licence, accepted per-account) | The dataset’s own EWDS page states: "accessing this dataset via CDS for time-critical operation is not advised or supported" - shown here anyway because Calorbis sends real flood warnings from it, disclosed rather than hidden. Return-period thresholds (2/5/20-year) come from a separate static GloFAS v4 grid, snapped to each point’s real river-channel pixel (not always the nominal city coordinate - see extract_thresholds.py). No per-point forecast-skill/verification figures are shown - not fabricated when not available. |
| River flood risk — anomaly-ratio fallback (global) | Forecast | Open-Meteo Flood API (wraps Copernicus GloFAS’s deterministic mirror) Daily river-discharge forecast | The GloFAS ~5km (0.05°) grid cell containing the city coordinate - NOT snapped to the nearest major river. Every GloFAS cell carries a discharge value, so this can be a minor stream rather than the river the city is known for; the popup says so (CAL-107) | Cron-sharded across 4 runs/day, ~24h full-sweep cadence | Same ~4,000-city catalog, minus any city within 60km of one of the 20 curated GloFAS-ensemble points above (that river reach is covered by the real ensemble instead). | Open-Meteo free tier | A single discharge-anomaly ratio against the river’s own recent baseline, not a real ensemble exceedance probability - kept only as broad-coverage fallback for rivers the curated GloFAS list doesn’t reach yet. |
| News mentions | News | GDELT Doc 2.0 API | Country centroid | ~15 min | Global, English-language sources only | GDELT (free, terms of use) | Pinned to the reporting outlet’s estimated home country, not the event’s actual location - an international wire story about a disaster elsewhere still shows at the outlet’s country. Not a verified hazard signal. |
| Community reports | Community | Calorbis users User-submitted reports | User-specified point | Real-time | Wherever users choose to report | N/A | No formal verification/trust system yet (duplicate detection, moderation, confirmed/disputed status) - tracked as CAL-93. |
These risks matter, and Calorbis has no layer for them. That is not a backlog item waiting on engineering: as of the last review, no public, real-time (or near-real-time) data supplier exists for them anywhere. We would rather say so here than let an empty map imply they are fine.
| Risk | What exists instead | Why it is not a live layer |
|---|---|---|
| AMOC (Atlantic overturning circulation) weakening or collapse risk | The RAPID-MOCHA array at 26.5°N measures the overturning transport directly, and Copernicus Marine models it. Tipping-point studies (e.g. early-warning statistics on sea-surface temperature and salinity fingerprints) are published papers, not feeds. | RAPID data are released in batches roughly a year or more after measurement, and no agency publishes an operational AMOC risk index. A collapse early-warning signal exists only as research output with wide disagreement on timing; showing it as a live layer would present a scientific debate as a monitoring product. |
| Absolute groundwater volume / remaining aquifer storage | GRACE and GRACE-FO satellites give monthly total-water-storage anomalies (change relative to a baseline) at roughly 300 km resolution; national well networks (e.g. USGS, BGR) give point levels for the countries that publish them. | Satellites measure change in mass, not how much water is left, and the ~300 km footprint cannot separate one aquifer from its neighbours or from soil moisture and surface water. Latency is one to several months. Depletion rate is trackable (a future layer, CAL-20); absolute remaining volume is not observable from any public source. |
| Real-time vector-borne disease range shift (malaria, dengue, chikungunya, Zika) | WHO and ECDC publish periodic surveillance reports and outbreak notices; research groups publish modelled Aedes/Anopheles suitability maps updated on annual or multi-year cycles. | Case surveillance is reported by countries with weeks-to-months delay and very uneven completeness, and mosquito range is not remotely sensed. No public feed gives a current, location-specific transmission risk; a climate-suitability model on the map would look like surveillance it is not. |
| Live internal climate-driven migration and displacement | IDMC's Global Report on Internal Displacement (annual, disaster-vs-conflict split) and the World Bank's Groundswell scenario projections for 2050. | Both are periodic retrospective or scenario products, not tracking. Displacement is counted after the fact through humanitarian reporting, and no source distinguishes climate-driven movement in real time. The climate report on this site cites IDMC's disaster-displacement total for that reason - as a published figure, not a live count. |
Last reviewed 2026-07-30, alongside the reliability audit that produced most of the fixes referenced above (CAL-70 through CAL-108); the “not monitorable” list was added 2026-08-28 (CAL-55). If a layer’s real behavior ever drifts from what’s written here, that’s a bug in this page, not just the layer - open an issue.