Space Weather Observatory

Source and method version: space-weather-method-v1 · October 2026.

This workspace brings together measured/estimated solar-terrestrial conditions, official NOAA forecasts, research models and NASA/JPL orbital data. Each retains its own units, valid time, uncertainty and coverage. There is no blended disaster probability or invented system-failure confidence score.

What is collected

Family	Products and interpretation
Geomagnetic environment	Planetary Kp, minute-estimated Kp, local Boulder K, GOES magnetic components, official G scale, NOAA Kp forecasts and separately labelled Geospace modeled Kp/Dst. Local K, planetary Kp, ap/A and Dst are different quantities.
Solar wind	Active and comparison L1 source streams, magnetic magnitude/components in GSE and GSM, plasma speed/density/temperature, supported velocity components and spacecraft ephemerides. Rejected overall-quality samples are gaps; inactive streams are not silently substituted for the designated active stream.
Radiation / solar emissions	GOES X-rays, integral and differential proton/electron/alpha channels, daily fluence, EUVS irradiance/Mg II, F10.7 and observatory/frequency-specific radio flux. Channels and integration periods are retained; overlapping thresholds are not summed.
Solar activity and forecasts	Regions, sunspot reports, edited events, detected flares, official three-day forecasts/discussion, 27-day and 45-day outlooks, monthly observed solar-cycle history and updated cycle forecasts, NASA DONKI event/linked-model context and source imagery.
Aurora / ionosphere / induction	Complete 65,160-node OVATION grid, hemispheric power, global GloTEC model grid, regional US–Canada 1D and CONUS EMTF geoelectric grids. Geoelectric fields are regional model estimates in mV/km, not measured induced currents or a power-grid failure map.
Planetary defence	Near-Earth asteroid orbital catalogue, predicted Earth close approaches, Sentry monitoring, Scout candidates and reported historical fireballs. These datasets describe different stages and kinds of evidence.

Times, missing values and quality

Retrieval/check time is distinct from observation time, model valid time, reporting period, orbital epoch and analysis time. Native source timestamps remain visible. Forecast rows are not relabelled observations.
Missing, invalid and expired values remain unavailable. Genuine zero values remain zero. Negative solar-cycle fill values are gaps, including in the historical series beginning in 1749.
GOES X-ray electron-contamination and magnetometer arcjet flags are retained; rejected samples remain gaps. EUVS flags, including geocorona, are disclosed. Source-specific radio-quality qualifications remain inspectable.
The ≥2MeV electron channel can be contaminated during significant proton events. Its 1000particles/(cm²·s·sr) alert threshold is an environmental indicator, not a satellite-failure probability or radiation dose.
Instantaneous/integral flux is particles/(cm²·s·sr); differential channels add a per-keV dimension; daily fluence omits the per-second dimension. X-rays/EUV irradiance are W/m², Mg II is a ratio, magnetic components are nT, solar-wind density is cm⁻³ and F10.7 uses solar flux units (10⁻²²W·m⁻²·Hz⁻¹).
Proton-only dynamic pressure is derived as 1.67262192369×10⁻⁶ × density(cm⁻³) × speed(km/s)², in nPa. It omits the alpha contribution and is labelled derived.
Source products may be available while particular measurements are missing, flagged or old. Counts are products/series/sample slots, not independent corroborating agencies; histories can include explicit gaps.

Risk context

Current R/S/G cards use the published NOAA current scale fields. Missing official values are not inferred as G0/R0/S0. Forecast G categories and R/S probabilities retain NOAA's supplied meanings. Bulletins show issue time and full text; presence in the recent list does not establish that a bulletin is still in effect. Read validity, cancellation and supersession statements.

Impact domains identify affected technologies—such as HF radio, GNSS, power transmission and spacecraft electronics. No local outage likelihood, personal radiation dose or asset-specific operational recommendation is inferred. NASA DONKI event interpretation/CME model arrivals remain separate from NOAA operational forecasts.

Aurora and geospace maps

OVATION is an empirical model, normally driven by upstream solar-wind measurements; NOAA may use Kp fallback input. The grid is geographic longitude/latitude. It is not a geomagnetic-coordinate grid or a direct aurora observation.
All360×181 published integer values are retained. Smooth display uses missing-aware periodic bilinear interpolation and linear texture filtering, with a source-cell display option. Native node inspection/export is unchanged. The fixed0–100% legend, threshold and opacity remain presentation controls; no heat-density sum, new observations or added source resolution is implied.
North, South and World views show the globe without a Mercator polar cutoff. Clicks inspect the nearest accepted source cell. City presets are location references, not forecasts for an individual's horizon.
The day/night mask uses approximate solar geometry at the displayed model's valid time. Clouds, light pollution, local terrain and observing conditions are not incorporated. Aurora visibility is not guaranteed by a model percentage.
The latest aurora field is withheld after its observation freshness ceiling or past-valid-time grace. Up to12 actual captured model frames are retained, normally covering the recent six hours; selected historical frames are explicitly labelled and are not invented past observations.
GloTEC is data-assimilative: its support flag0 indicates empirical background, not a failed value and not an observation. Positive support values indicate increasing/capped assimilated F-region observational support. The optional supported-only layer excludes background-only cells. TECU =10¹⁶electrons/m². hmF2 is km; NmF2 is electrons/m³. No GNSS positioning error is calculated from vertical TEC alone.
Geoelectric magnitude is derived from the supplied Ex/Ey components. Native quality flags and distance to the nearest station are retained without inventing quality categories. Where the payload provides only a date, the file timestamp is explicitly identified as the time basis. Coverage depends on regional conductivity models/surveys.

Asteroid atlas and monitoring

The orbital catalogue is near-Earth asteroids, not all Solar-System small bodies. It is bounded at50,000 records. The map draws a bounded≤120-object sample; tables and paginated exports expose the published catalogue.
Paths are nominal osculating ellipses reconstructed from the source semimajor axis, eccentricity, inclination, ascending node and argument of perihelion in heliocentric IAU76/80 ecliptic J2000. The3D view advances mean anomaly using source mean motion to the selected approximate scene date, only within365.25days of each source epoch. Other markers are withheld. Source epochs remain visible. This two-body visualization omits perturbations/covariance and is not an integrated ephemeris, close-encounter solution or impact corridor. Unsupported frames/unbound trajectories are not fabricated.
The3D scene adds all8planets using JPL Table1 fitted elements/rates. Earth is the Earth–Moon barycenter reference. Positions/orbits use linear AU distance, with enlarged body sizes and illustrative surfaces. Scene UTC uses a fixed+69.184s TT/TDB approximation; it is not precision timing. Earth approach data are separately plotted against time and geocentric miss distance. CAD dates/Julian epochs are TDB; native calendar labels remain TDB. Countdown days use an approximate UTC/Julian comparison, not precision TDB–UTC conversion.
CAD minimum/maximum distances and time uncertainties are provider three-sigma quantities. Diameter uncertainty, when supplied, is one-sigma. Missing sizes remain missing.
MOID is the minimum distance between orbital paths, not a predicted simultaneous encounter. PHA is a source classification, not an impact prediction.
Sentry probabilities are cumulative over their supplied potential-impact windows; Palermo and Torino retain NASA definitions. Uncertain solutions can change with additional observations. Sentry diameters are often approximate estimates from H/albedo. Scout candidates, classification scores and ordinal ratings are not equivalent to Sentry probabilities.
Fireballs are historical sensor reports at peak brightness, with approximate radiated/estimated impact energy and incomplete coverage. Missing locations remain unlocated. They are not future asteroid impact locations; radiated energy (10¹⁰J) and estimated impact energy (kt) remain separate.

Collection and freshness

The browser reads published snapshots through /agent/api/space-weather-world; it does not trigger provider collection or accept arbitrary provider URLs. A background worker checks due products approximately every two minutes. Most live NOAA products refresh every5–15minutes; daily/history/model products have longer schedules. JPL CAD/Scout are hourly, Sentry six-hourly and the orbital catalogue/fireball archive daily. The Sources view lists each product's actual interval, freshness ceiling, retrieval/check time and error state.

Requests use reviewed HTTPS hosts, protected DNS, response/time bounds, source schemas, persistent pacing/backoff and atomic publication. A shared process-safe JPL broker serializes requests across the router and new worker, reuses cached responses with original retrieval timestamps, supplies a product/contact User-Agent and persists rate-limit/review pauses. Public force/custom-URL/private selectors are rejected by the new dataset.

Exports use source/freshness snapshot guards. If a snapshot changes during pagination, restart the export rather than merging generations. Account changes and replay clear live selections and fence late responses. Hidden map renderers are released.

Source references and credits

NOAA SWPC public data service · NOAA scale definitions · OVATION model · GloTEC · Geoelectric models.
NASA CCMC DONKI · September30,2026 API migration. The new canonical ccmc.gsfc.nasa.gov/DONKI-API/get/… endpoints are used. The former kauai endpoints return a migration notice.
NASA/JPL API documentation and fair use · CAD · SBDB Query · Sentry · Scout · Fireballs.
Solar-cycle/radio context includes NOAA-distributed WDC-SILSO, Royal Observatory of Belgium, Brussels and NRCan/DRAO data. Regional geoelectric products acknowledge NOAA/SWPC, NRCan, USGS and the referenced EPRI/EMTF conductivity work. Basemap credits remain displayed for Esri and OpenStreetMap contributors.

The former predicted_monthly_sunspot_number.json file ends in2022 despite being retrievable today and is not used as a current forecast. The reviewed standalone DONKI WSAEnlilSim path returned404; supported CME-associated model runs and NOAA ENLIL products are used with their own scope. Independent Kyoto Dst redistribution remains a reviewed-source gap; Geospace model Dst is not relabelled Kyoto observation.

3D Solar System and smooth aurora maps

Open Space Weather → Asteroids or Space Weather → Aurora. Use Full screen for the largest map; the same controls work inside the full observatory.

Explore the 3D Solar System

Solar System frames the Sun and all eight planets. Inner planets brings the near-Earth asteroid sample into view. Outer planets and inner orbits span very different distances; use the presets rather than expecting every body to look equally large in one view.
Drag to orbit the camera. Shift/right-drag pans; scroll or pinch zooms. Arrow keys rotate, plus/minus zoom, and 0 resets. Touch uses one finger to rotate and two to pan/zoom. Keyboard focus on the canvas shows its focus outline.
Choose Focus planet or click a visible planet to inspect a textured 3D body. Saturn includes a ring model. Close the information card when you want more room.
Click an asteroid marker or use Layers → Inspect a displayed asteroid for its native NASA/JPL record. Focus selected centers the selected marker when its approximate scene-date position is available.
Time & display chooses a UTC reference date from 2000 through 2049. Play orbits advances the illustrative scene by 8 days/second; Pause orbits stops it and Today returns to the current reference time. Motion is paused by default and stops when the scene is hidden or the account/replay context changes.
Layers control asteroid paths, approximate markers, labels, planet orbit guides and planet bodies. Map view still offers the separate encounter distance/date chart and historical fireball globe. Source tables and exports remain available.

What the 3D positions mean

Planet positions use NASA/JPL SSD's Table 1 fitted Keplerian elements and rates, in the mean ecliptic/equinox of J2000, valid 1800–2050. This is a lower-accuracy reference model, not a numerical Horizons ephemeris. The Earth marker uses the Earth–Moon barycenter, not an exact Earth geocenter. Source nominal errors are body-dependent; outer-planet distance errors can be hundreds of thousands of kilometres or more.

The scene's UTC reference is converted with a fixed +69.184 seconds modern TT/TDB approximation. It is not an exact historical/future leap-second conversion. This approximation and the fitted orbit model are unsuitable for precise encounter timing, spacecraft navigation or occultation work. For high-precision ephemerides, use NASA/JPL Horizons.

Asteroid paths retain the source semimajor axis, eccentricity, inclination, ascending node and argument of perihelion in heliocentric ecliptic J2000. Approximate scene-date markers advance the source mean anomaly using its published mean motion: M = M₀ + n × (scene JD − source epoch JD). This is two-body visualization; planetary perturbations, close-encounter dynamics and covariance are not propagated. Markers more than 365.25 days from the source epoch, or with missing/unsupported parameters, are withheld. Their valid nominal paths and original source details can still be read. The bounded map sample is at most 120 asteroids; the native catalogue remains separately searchable and exportable.

Positions and orbit paths share a linear astronomical-unit scale. Body sizes are enlarged for visibility, and surface orientation/lighting/stars are illustrative. The 3D scene is not an impact corridor or collision forecast. Use the unchanged NASA/JPL close-approach table for source TDB encounter times, geocentric miss distances and three-sigma bounds. MOID, PHA classification and Sentry probabilities retain their distinct source meanings.

Smooth aurora colors

The NOAA OVATION globe now uses a higher-resolution display raster, missing-aware bilinear interpolation and linear texture filtering. This removes block-like color steps while keeping a fixed 0–100% color scale. Interpolation wraps longitude across the date line and does not fill missing source cells or overshoot neighboring values. Rendering is performed in a bounded worker where supported, with an on-demand fallback.

Layers → Smooth aurora colors (display only) switches between the smooth rendering and source-cell display.
Overlay opacity changes visibility over the basemap. Aurora display threshold hides lower displayed values; it does not increase the published probabilities.
North / South / World, city references, basemap selection, day/night and fullscreen remain available. Higher-quality globe antialiasing and a smoother day/night texture improve visual edges.
A click still reports the exact nearest published grid node, not a newly observed/interpolated probability. Export displayed aurora grid exports the unchanged 360×181 source grid, with 255 reserved for missing values.
Observation time, forecast-valid time and selected historical frame stay separate. Expired current fields are withheld; historical captured frames remain labelled. Smoothing does not add source resolution, a visibility guarantee or a system-failure probability. Clouds, local light pollution and the observer's horizon are not modeled.
The smoothing switch applies to OVATION. GloTEC and regional geoelectric layers keep their own units, source-cell/support rules and interpretation.

Graphics, accessibility and recovery

The scene renders on demand, caps drawing-buffer size and releases hidden graphics. If WebGL 3D is unavailable, the original orbit chart and source tables remain available. A lost 3D context offers Retry 3D. Source failure clears the affected live evidence; reference planet geometry does not imply that asteroid data were successfully fetched. Account/replay changes clear selections and stop animation. No new paid model or provider collection is triggered by the camera controls.

Sources and visual credits

NASA/JPL SSD approximate positions, Table 1, based on E. M. Standish and J. G. Williams (1992). Coefficients and fitted validity interval are retained; the interface limits scene dates to 2000–2049.
NASA/JPL small-body API documentation for the existing orbital catalogue, CAD, Sentry, Scout and historical fireballs.
NOAA SWPC OVATION model.
Planet/Sun/ring textures: Solar System Scope / INOVE, source pack, CC BY 4.0. The source adjusts colors and artistically fills some unmapped terrain. These are illustrative textures, not current planetary observations. Images are hosted locally; credit remains visible on the scene and in object details.
Three.js 0.180.0 and OrbitControls/Line2: MIT license. The pinned bundle and license are included in the release.

The Space Weather methods describe the full measurement, forecast, quality, temporal and coverage rules.