SEITA IIZUKA

Orbital catalog and conjunctions (Tracker · Passes · Statistics · GS Planner · public data)

• Space-Track.org — GP element sets (18th Space Defense Squadron / USSPACECOM), CDM Public, TIP predictions, Decay records and SATCAT. Requires an account; OrbitSmith builds the GP/TIP/Decay/SATCAT products through GitHub Actions using Space-Track-compliant daily cadence and safe-window attempts; CDM Public is refreshed up to every 8h at 00:44 / 08:44 / 16:44 UTC via the authenticated Worker. Space-Track's terms govern reuse; cite Space-Track.org and USSPACECOM. Terms
• CelesTrak — GP and Supplemental GP elements by Dr. T.S. Kelso. SupGP sets are fitted to operator-supplied ephemerides for major constellations (Starlink, OneWeb, GNSS, etc.); CelesTrak reports them as generally more accurate than standard GP, and some carry future (forward-fit) epochs. Passes uses CelesTrak's visual group, cached by OrbitSmith's server, as its candidate list.
• GCAT — General Catalog of Artificial Space Objects by Dr. Jonathan McDowell, released under CC BY 4.0 with citation requested by the author. Used for operator / country / launch-date metadata, loaded per object in Tracker popups.
• ESA DISCOS — Space Debris Office environment statistics. Statistics shows a dated point-in-time snapshot (2026-07-31) and an approximate annual reference series through 2025; Timeline reuses the same annual series.
• J.C. McDowell, Space Activities in 2025 (v1.4) — annual orbital launch attempts by country of launch, 2000–2025, for Statistics.
• Amazon Leo mission updates — with Space-Track records and other operator disclosures, the basis of the dated historical constellation estimates in Statistics.
• McCants satellite brightness data — used only by the legacy pass engine inside the OrbitSmith iOS app; the web Passes page does not estimate brightness.

The Tracker catalog (fused active set; the current object count and build time are shown in the Tracker and in the public manifest) is built from Space-Track GP (decay-filtered with 60-day stale-EPOCH retention since 2026-05-01), CelesTrak GP, and CelesTrak SupGP with SupGP prioritized where available (most recent epoch wins). This fusion is an OrbitSmith aggregation; individual object elements are upstream as-is. Some records report an orbital-element epoch that differs from their TLE epoch; OrbitSmith flags but does not reconcile them.

Statistics keeps three scopes separate: the dated ESA DISCOS snapshot (2026-07-31), an approximate annual reference series reconstructed from ESA DISCOS through the latest complete year (2025), and the daily OrbitSmith active fused catalog. Their totals and categories are not directly interchangeable.

The Statistics headline “Fragments, Released Objects & Unidentified” combines ESA payload mission-related objects (PM), payload fragmentation debris (PF), payload debris (PD), rocket mission-related objects (RM), rocket fragmentation debris (RF), rocket debris (RD), and unidentified objects (UI). This is an OrbitSmith aggregate, not a single ESA category; UI objects are not asserted to be debris.

Statistics series notes: OrbitSmith extended stale-EPOCH retention from 30 to 60 days on 1 May 2026; the CDM series changed from messages issued in the prior 24h to events with TCA in the next 24h on the same date. Snapshots from 30 April to 2 May are absent during the Space-Track suspension recovery. Treat these as structural breaks when reading older plots.

Constellation counts after the dated history are OrbitSmith name matches on PAYLOAD records (Amazon Leo satellites carry KUIPER names), not counts of operational satellites. Fragmentation events form a selected, non-exhaustive list with reported or approximate fragment counts. If the Space-Track GP baseline is unavailable, daily object counts fall back to CelesTrak and the page says so.

Earth imagery (Imagery page)

• NASA GIBS — Global Imagery Browse Services tiles: MODIS Terra/Aqua and VIIRS SNPP/NOAA-20/NOAA-21 true colour; MODIS NDVI (8-day), snow cover, aerosol optical depth, land surface temperature and cloud fraction; VIIRS Day/Night Band night lights and chlorophyll-a; GHRSST MUR sea surface temperature; IMERG precipitation rate. The default date is yesterday, and some layers step back 1–3 days for processing lag. NASA Earthdata states that data from NASA-led missions are CC0 unless marked otherwise and asks users to cite datasets. Terms
• NASA FIRMS — Fire Information for Resource Management System active-fire detections (VIIRS 375 m day and night, MODIS), requested through the GIBS WMS for the selected date.
• Sentinel-2 Cloudless by EOX — Annual cloud-free Sentinel-2 mosaics for 2022, 2023 and 2024 (10 m class, containing modified Copernicus Sentinel data), served by EOX IT Services. EOX licenses them under CC BY-NC-SA 4.0 for non-commercial use; commercial use needs a separate EOX licence. Terms
• Esri World Imagery — Composite basemap of satellite and aerial imagery credited by Esri to Esri, Vantor, Earthstar Geographics and the GIS User Community; capture dates and resolution vary by place, and the Esri Master License Agreement applies.
• OpenFreeMap · OpenStreetMap · Nominatim — Background maps, labels and fonts on Imagery, the Ground Station Planner and Passes come from OpenFreeMap (© OpenMapTiles, data © OpenStreetMap contributors, ODbL); place searches on Passes and Imagery go from your browser to OSM Nominatim.

Orbit Visualization (3D Globe)

Each catalog object is placed on the 3D globe with two-body Kepler mechanics: the TLE mean anomaly is advanced from the element epoch to page-load time at the mean motion, Kepler's equation is solved by fixed-point iteration, and the point is oriented by inclination, RAAN and argument of perigee. After that, the animation only rotates each point about the polar (Y) axis at its mean-motion rate, so on-screen motion does not follow the true orbit plane. Selecting an object uses the same point; in the Tracker, SGP4 is used only for reentry-watch altitude.

Perigee / Apogee / Period

Computed from TLE mean orbital elements: perigee = a(1 − e) − R⊕, apogee = a(1 + e) − R⊕, period = 1440 / n (minutes), where a is the semi-major axis supplied with the cached record or derived from mean motion (μ = 398,600.4418 km³/s²), e is eccentricity, and n is revolutions per day. R⊕ is a spherical 6,371 km.

Orbital-Plane Speed Difference (Compare Two Orbits)

In the Tracker's Compare Two Orbits panel, OrbitSmith estimates a representative relative velocity from catalog perigee altitude, inclination and RAAN using spherical trigonometry of orbital planes. The crossing angle θ is derived from inclinations and right ascensions: cos(θ) = cos(i₁)cos(i₂) + sin(i₁)sin(i₂)cos(ΔΩ). Then v_rel = √(v₁² + v₂² − 2·v₁·v₂·cos θ), with v₁, v₂ being circular orbital speeds at each object's altitude. The panel also shows the catalog altitude difference.

Retired: Shell Risk Indicator

Earlier Tracker versions used a simplified Kessler-style spatial density model as a relative shell indicator: P = 1 − exp(−ρ · σ · v_rel · Δt), where ρ is an implied shell density, σ is combined cross-section, v_rel is representative relative velocity, and Δt is the selected time window, treating each orbital shell as a uniform population. The current Tracker no longer computes or displays this indicator; it is documented here for readers of older screenshots.

Catalog Distribution and the Retired Congestion Index

The current Tracker panel shows each orbit class's share of the cached catalog using mutually exclusive period bins: LEO < 128 min, MEO 128 to < 720 min, GEO 1,400–1,500 min with apogee < 40,000 km, and HEO/other for everything else. It is descriptive only. Earlier versions showed a spatial flux indicator F = ρ · v_rel · σ per regime, where ρ is object density (count / shell volume), v_rel is a regime-typical relative velocity (10 km/s for LEO, 5 for MEO, 0.5 for GEO, 7 for HEO), and σ = 10 m² is a typical combined cross-section, with the LEO bucket limited to 800–1,000 km and CRITICAL / HIGH / MODERATE / LOW labels relative to LEO's value; that index is no longer computed.

Reentry Current Altitude

For objects on reentry watch, current altitude is computed in-browser via SGP4 propagation (satellite.js) from the cached TLE. Recomputed every 10 seconds, with an arrow comparing the altitude 30 seconds later; the underlying TLE itself is from the cached daily fused catalog, so its epoch may be many hours old.

Reentry Watch Display

Active or unresolved TIP predictions are listed by predicted decay epoch with their uncertainty window (±); objects that already have a confirmed decay record move to Recent. Once a predicted time has passed, an entry reads “inside uncertainty window” or “prediction time passed” instead of a countdown. ▲ WATCH reflects the TIP record's high-interest flag, and RCS size classes are not ground-risk levels. On the globe, the TIP latitude/longitude is marked with a symbol that grows with the window length; it is not an impact footprint, a ground-risk area or an official warning.

Major Debris Sources

The Tracker groups catalog objects typed as debris by the parent name before “ DEB” or “ R/B”, keeps groups of at least 100 objects and lists the 5 largest with mean perigee altitude and inclination. This name-based OrbitSmith aggregation of the current catalog is not an event register and can merge or split events whose names differ; Statistics lists selected events separately.

Visible Pass Prediction

On the web, Passes runs in a browser worker: SGP4 (satellite.js 6.0.2, WGS72) propagates real GP TLE lines from the cached fused catalog for objects in CelesTrak's visual group, or for a short list of key spacecraft if that list is unavailable. A candidate needs at least 10° elevation, the satellite sunlit under a conical Earth-shadow model, and the Sun at or below −6° (civil darkness) for an observer assumed at 0 m on the WGS84 ellipsoid. Boundaries are searched on a 30-second grid and refined, catalogs built more than 25 hours earlier are flagged, and “Easy to spot” means a maximum elevation of at least 30°. Brightness is not numerically estimated.

CDM Risk Level

OrbitSmith groups Pc values into four display bands. The thresholds are anchored to public NASA CARA and ESA collision-avoidance references, but the band names are OrbitSmith-defined and are not official NASA/ESA categories. Specifically: HIGH (Pc ≥ 10⁻⁴) corresponds to the NASA CARA mitigation threshold (NASA Step 2) and ESA's ~1-in-10,000 maneuver threshold. MODERATE (10⁻⁵ ≤ Pc < 10⁻⁴) corresponds to the ESA SCARF dashboard highlight threshold documented in Merz et al. 2017 (SDC7 paper 1017). LOW (10⁻⁷ ≤ Pc < 10⁻⁵) is above NASA's operational-attention floor (NASA Step 2) but below ESA's escalation-candidate threshold. NEGLIGIBLE (Pc < 10⁻⁷) is below NASA's operational-attention threshold. The Pc value itself is upstream from Space-Track; only the band assignment is made by OrbitSmith. OrbitSmith is not an operational tool.

Lunar terrain (Lunar Lab)

• NASA PGDA LOLA MOON_PA gridded dataset — Neumann (2024), doi:10.60903/LOLA_PA: global LOLA elevation in the lunar principal-axes frame, relative to a 1,737.4 km sphere. Lunar Lab uses the 64 pixels/degree grid, stored offline as a 32 cells/degree runtime DEM and drawn with interpolated 64-cell meshes.
• NASA PGDA SLDEM2015 — Barker et al. (2016) merged LOLA and SELENE Terrain Camera DEM for 60°S–60°N, used at 256 cells/degree for the 20 named hero zones (usually 4°×4°; Clavius 3°×3° to stay inside the 60°S limit).
• NASA Scientific Visualization Studio CGI Moon Kit — LROC WAC-derived colour map (2025 version) used as the surface image. NASA SVS notes the kit is optimized for aesthetics rather than science; OrbitSmith adds documented display adjustments. Credit: NASA's Scientific Visualization Studio.

Site centres are approximate reference coordinates reviewed by the project, not yet checked against an authoritative source for every site, and high-latitude tiles keep the east-west stretch of the equirectangular source. Lunar Lab adds local synthetic layers for rover controls, routes, scoring, wheel trails, fine regolith texture, rocks, lighting and mission markers. It is an educational visualization only, not a mission planning, navigation, or safety tool, and is not endorsed by NASA, USGS, or PDS.

Mars terrain (Mars Lab)

• USGS Mars MGS MOLA DEM 463m v2 — MGS MOLA elevation mosaic (128 pixels/degree, heights above the MOLA areoid, 2003 edition; USGS lists it as CC0), resampled offline into static global regional packs; a MOLA-derived normal map provides shading.
• NASA PDS MGS MOLA MEGDR — Upstream Mission Experiment Gridded Data Record for the MOLA elevation product.

Mars Lab uses real-data-derived MOLA elevation, but surface colors (a muted synthetic Mars palette, not measured imagery), rocks, subtle dust/ripple detail, sky/sun/haze presets, rover controls, route planning, scoring, landing-region models, the non-official Ingenuity-inspired rotorcraft and obstacle collision rules are local synthetic layers. It is an educational visualization only, not a mission planning, navigation, or safety tool, and is not endorsed by NASA, USGS, or PDS.

Ceres and Vesta terrain (Ceres Lab · Vesta Lab)

• USGS Ceres Dawn FC2 HAMO Global DTM 137m — DLR stereo-photogrammetric DTM published 2016-08-19: about 136.7 m per pixel (60 pixels/degree), about 98% coverage with polar gaps interpolated in the source, stated vertical accuracy about 10 m, heights relative to a 470 km sphere.
• Vesta Dawn Framing Camera HAMO DTM (USGS/DLR) — per the Lab's source notes: 64 source pixels/degree, about 95% coverage with shadowed polar areas interpolated, and a source-product vertical accuracy of ±6 m listed in NASA PDS.
• DLR Dawn Framing Camera global mosaics — broad surface brightness only, rotated into the runtime longitude layout; display colour and close detail add no measurement.

Both Labs resample the DTMs offline into static global packs and lead toward 20 officially named features each. Rocks, wheel trails, lighting, routes, scoring, collision rules and the fictional rover are synthetic.

Mercury terrain (Mercury Lab)

• USGS Mercury MESSENGER Global DEM 665m v2 — Published 2016-10-21: 64 pixels/degree (about 665 m per pixel), heights relative to a 2,439.4 km sphere; interpolated, filtered and smoothed, with accuracy stated as “Best Effort” and no quantified vertical error. Resampled offline to 32 cells/degree (about 1.33 km at the equator).
• Brightness composite — tone from a 1024×512 USGS MESSENGER MDIS low-incidence browse image blended at 72% with high-frequency luminance from the 6132×3066 NASA SVS MESSENGER global mosaic: a visual composite, not calibrated reflectance.

Close detail, rocks, wheel trails, lighting, routes, scoring, collision rules and the fictional rover are synthetic; 20 officially named features serve as approximate destinations.

Enceladus terrain (Enceladus Lab)

• USGS Enceladus Cassini Global DEM 200m (Schenk and McKinnon) — Cassini ISS stereo and shape-model DEM published 2024-08-12: 200 m grid, about 400 m effective horizontal resolution, roughly 25–100 m vertical precision depending on location, heights relative to the Thomas et al. (2016) triaxial ellipsoid. Converted to metres and resampled offline; the Lab renders on an approximate 252.1 km sphere, so displayed heights are not sphere-referenced measurements.
• USGS Enceladus Cassini Global Mosaic 100m (Schenk) — broad brightness; the runtime images are display resamplings, not calibrated reflectance or extra observed detail.

15 drivable destinations and 5 south-polar references. Close detail, fictional ice blocks, routes, lighting and rover behaviour are synthetic.

Titan terrain (Titan Lab · Huygens detail)

• Corlies et al. (2017), Titan topography — Published interpolated global topography: 4 pixels/degree (about 11.24 km at the equator), interpolated from Cassini altimetry, SARtopo and stereo data that cover about 9% of Titan. Every pixel has a value, but most are interpolated rather than measured; smoothing to 16 cells/degree and 64-cell meshes adds no observation, model heights are kept without further datum correction, and the precise vertical datum was not independently checked.
• NASA/JPL-Caltech/Univ. Arizona PIA22770 — Cassini ISS 938 nm near-infrared global mosaic (2018): 9,873 images, 16 pixels/degree (about 2.8 km), photometrically corrected for haze. Used for brightness; it is not natural colour, and the warm tint is artistic.
• Daudon et al. (2020), Huygens landing-site DTM (IPGP) — The Huygens detail view uses a 1.62 × 1.44 km central crop of this DISR-image DTM, sampled every 18 m (grid spacing, not elevation accuracy), excluding the rim, which has known artifacts, and drapes a registered DISR orthomosaic. Shown at 1× horizontally and vertically; rover, stones, colours, sky, lighting, route and battery are synthetic, and travel there is not added to global totals.

Illustrative near relief: because an 11 km source grid cannot resolve rover-scale landforms, the default view adds a deterministic, bounded procedural relief layer on top of the unchanged model. It is not elevation data, is never converted to metres or named-feature shapes, and the Lab can switch back to the source terrain for comparison.

17 model destinations and 4 liquid/polar references. The Huygens Region marker in the global Lab is an approximate study-image anchor, not a surveyed landing position; missions, scoring and collision rules are synthetic.

Open the Huygens detail view

Venus terrain (Venus Lab)

• USGS Venus Magellan Global Topography 4641m (GTDR v2) — Magellan radar-altimeter topography, version 2 released 1997-12-18: 4,641 m spacing, heights relative to a 6,051 km sphere (distinct from the commonly quoted mean radius); USGS lists it as public domain. Stored at 16 cells/degree with conservative no-data masks, and driving and drawing share one Catmull-Rom surface at 64 cells/degree. No-data stays unknown and no synthetic relief is added.

Display scale: 250 m per simulation unit horizontally; relief is shown ×2 above the local reference while sphere curvature stays true-scale, so the ground can tilt by up to about 20° far from a region's start. HUD and route slopes come from this display model. The artist view (colour, light, haze and illustrative rocks styled after the Venera 9 and 13 landing-site photographs) is not an optical surface image; rocks elsewhere on Venus are unknown. 20 official features: 18 destinations and 2 references.

Pluto terrain (Pluto Lab)

• USGS Pluto New Horizons LORRI–MVIC Global DEM 300m — Stereo topography of the hemisphere seen at closest approach (published 2017-07-14; Schenk et al. 2018): 300 m per pixel, heights relative to a 1,188.3 km sphere. Stored as strict area averages at 16 cells/degree, so any missing pixel makes a node unknown, and drawn at 64 cells/degree; unknown terrain stops travel.

Relief is shown at 1× (50 m per simulation unit). Ground tint comes from the New Horizons global mosaic, smoothed to about 6 km with an artistic palette and its lighting included — not calibrated colour. Frost, shallow hollows, dust and ice blocks are illustrative, and rover-scale blocks on Pluto are unknown. 21 official features: 18 supported destinations and 3 reference-only.

Phobos shape model (Phobos Lab)

• Gaskell Phobos Shape Model V1.0 (NASA PDS) — Prepared by R. Gaskell in 2006 from 232 Viking Orbiter 1 and Phobos 2 images and archived in the NASA Planetary Data System (cited as Gaskell 2020). Its vertices are a model reconstructed from images, not individually surveyed points. The bundle states 456 landmarks with a formal position uncertainty of 18 m per axis and an RMS residual of 25 m, but includes no coverage or confidence map, so how well the driving region is constrained is unknown. NASA's data catalogue lists the bundle as US public domain (us-pd). data.nasa.gov record
• IAU/USGS Gazetteer of Planetary Nomenclature — Stickney — lists a 9 km crater centred at 1°N, 49°W in planetographic, west-positive coordinates with an unknown control network. The model frame (NAIF PCK00006, planetocentric) is different, so the Lab does not compare the two positions numerically: the name is attached by crater form after one model-wide search for rimmed bowls 4.0–12.4 km across, and the floor centre used for distances (0.0°S 53.5°W) is derived from the shape model, not the catalogue centre.

The drivable region is a 2.5 × 2.5 km square on the Stickney floor (with a 250 m margin that cannot be driven) made of the model's native Q=512 triangles: 19,878 triangles with vertices about 32 m apart — a grid spacing, not a measurement resolution. Kilometres are converted once to metres; nothing is smoothed, filled or flattened. Between vertices the ground is the flat model facet, so relief smaller than the grid is absent and accuracy at rover scale is unknown. Before use, every facet was checked to face the region's reference up, with one surface crossing at every 5 m sample. The region edge stops the rover and there is no global driving; GLOBAL MAP and the display-only far field use the Q=128 model (196,608 triangles).

Relief is shown at 1× (1 simulation unit = 1 m), so on-screen steepness is not exaggerated; the fictional rover is drawn about 3.2 m long. SIM SLOPE is a facet's angle from one fixed reference up (the best-fit ellipsoid normal at the region centre, 7.8° from radial): a geometric measure, not a gravitational slope, because Phobos's effective gravity (self-gravity, rotation and the tide of Mars) is not modelled. GO / CAUTION / NO-GO at 15° and 22° are display categories, not vehicle limits. Driving is kinematic: speeds, ×8 CRUISE and 2 cm sinkage are simulation choices, far faster than a real Phobos vehicle (the Lab cites DLR's figure of about 1 mm/s for the Idefix rover), and bouncing, hopping and soil are not simulated.

Regolith grain, pits, clasts and colour variation (shading only, never height), illustrative boulders, stars, the Sun disc, SURV/LOW lighting, Mars's colour and phase, and the rover are synthetic; the single display tone is brightened for visibility and is not a calibrated colour or albedo. Mars is placed toward the model's +X axis because Phobos rotates synchronously, not from an ephemeris. No MMX images, plans or landing sites are used, and the fictional rover does not represent the Idefix rover or any real hardware; nothing shown is telemetry.

Mission Designer (guided screening)

5 steps: choose 1 of 7 mission templates and optional mission intent; edit 16 parameters (template values are examples, not requirements); run quick checks for 5 areas; review an unranked list of 8 flight-proven launch vehicles to research (plus 1 archived record); and read the screening summary. Up to 2 comparison concepts can be saved. Inputs stay in the current browser tab (sessionStorage), and nothing is sent to a server.

Quick-check models: power — solar flux 1,361 W/m², 30% cell efficiency, a template incidence angle and 2.5%/year degradation over one circular orbit with a cylindrical-shadow eclipse; link — slant range at 5° elevation, free-space path loss, a reference-station G/T and fixed band losses, against a required C/N₀ of 4.5 dB + 10·log₁₀(data rate); orbit lifetime — a piecewise-exponential density table (200–800 km), C_D = 2.2 and template density, stepped daily down to 120 km (up to 30 years) and compared with design life + 5 years, with a Hohmann de-orbit Δv when needed; mass — fixed subsystem fractions; thermal — a single-node hot/cold balance.

Mission Engineering Workbench (detailed review)

Continues a Designer concept into 13 items: 8 detailed calculations, 4 research and confirmation checks and 1 local report package, covering visibility, ground-station geometry, link budget, mass, power, propulsion, lifetime and disposal, launch compatibility, thermal environment, communications components, environmental verification, programmatic screening and the report. Results stay separate from the quick checks, and unrun or unconfirmed items remain visible.

The link-budget item is deterministic arithmetic on explicit inputs (free-space path loss, received carrier power, C/N₀, Eb/N₀) with 9 named loss categories that must each be included or explicitly excluded; if any is excluded, no complete margin is reported. Planning stations reuse the GS Planner candidate catalog and its stated coordinate uncertainty. Outputs are only as good as the inputs and evidence you provide.

Open the Mission Engineering Workbench

AI consultation data (optional)

Builds 3 local files — a request guide (agent-guide.md), structured data (mission-context.json) and a readable brief (mission-brief.md) — from the current concept, quick checks, confirmed Workbench results and open questions. OrbitSmith does not connect to or send anything to an AI service; you review the content and copy or attach it to an assistant of your choice. Treat any AI answer as unverified.

Open the AI consultation data page

Ground Station Planner

Choose a satellite from the cached catalog (default ISS) or paste your own TLE/OMM, select candidate stations or a custom point, and compare geometric visibility windows above a minimum elevation (default 5°) over a chosen interval (default 1 day), propagated with OrbitSmith's shared SGP4 engine in a browser worker. Candidates come from public provider listings (KSAT, AWS Ground Station, SSC, Leaf Space, Viasat Real-Time Earth, ATLAS): 67 are calculable and 1 retired listing is location-only.

Constellation Designer

Walker-delta layout (T total satellites, P planes, phasing F = 0–2) on circular orbits of one altitude (200–2,000 km) and inclination, up to 64 planes × 64 satellites = 4,096. Two-body motion around a spherical Earth (R = 6,378.1366 km, μ = 398,600.4418 km³/s²) with Earth rotation; no perturbations, drag or station-keeping. A point counts as visible when at least 1 satellite is above the chosen minimum elevation (default 25°). Visibility share and the largest gap use 30-second samples over T+0–24 h, including edge gaps; instantaneous global coverage is a spherical-sample estimate.

Mission Operations Planner

One learning satellite over 24 hours: an ideal circular orbit without precession (200–2,000 km, epoch in UTC 2000–2050), an approximate Sun position, a cylindrical Earth shadow without penumbra, observation targets with off-nadir and daylight limits, and stations with a minimum elevation (catalog candidates reuse GS Planner coordinate estimates). You schedule observations, downlinks and recovery; battery, storage and delivered data follow constant power and data rates. Window boundaries are refined to 0.25 s, but very short grazing passes can be missed.

Solar System

• JPL Solar System Dynamics — Approximate Positions of the Planets — Keplerian elements fitted for 1800–2050, with nominal errors published by JPL; positions for the displayed date are computed in the browser.
• NASA Science — Planets — body facts (diameter, distance, periods, gravity).
• NASA PDS — Gaskell Phobos Shape Model V1.0 — Phobos is drawn from a shared OrbitSmith model that MMX also uses: every fourth row and column of the source's Q=128 grid (Q=32: 6,146 vertices and 12,288 triangles, about 99.9% of the source volume), positions rounded to whole metres and lighting normals averaged from the full Q=128 surface, with no added relief, craters, texture or colour map. A coarser Q=16 sample of the same model is the fallback if the shared file does not load.
• NASA Science — Phobos · Deimos — published axis dimensions for the body facts (Phobos about 27 × 22 × 18 km, Deimos about 15 × 12 × 11 km). Deimos is drawn as a smooth ellipsoid of those axes: a size diagram with no observed shape model, crater map or terrain.
• JPL — Planetary satellite mean elements — MAR099 mean periods (Phobos 0.3187 d, Deimos 1.2625 d), used only to pace the display orbits, not as ephemerides.

Vesta, Ceres and Pluto use simplified educational orbits, and Enceladus and Titan move on reference-period circular orbits with fixed educational phases aligned to the drawn rings — not ephemerides. EXPLORE mode compresses distances and enlarges bodies; DISTANCE mode keeps orbital gaps proportional to distance from the Sun, but bodies stay enlarged. Stars, minor-moon orbits and light effects are illustrations. Textures use NASA/JPL/USGS imagery or NASA-derived CC BY 4.0 maps (Solar System Scope), with a procedural fallback if an image fails. Not suitable for observation, navigation or mission planning.

Mars's moons move on enlarged circular display orbits aligned with Mars's displayed tilt, from arbitrary educational phases at J2000, so their positions are not predictions; sizes and local distances are enlarged in both EXPLORE and DISTANCE views, and the panels give physical dimensions. Phobos's pose (+X toward Mars) is a display choice, not an observed attitude, and its single display tone (the Phobos Lab regolith tone, brightened) is not a calibrated colour or albedo.

The Growing Crowd in Orbit (Timeline)

Uses the same approximate annual ESA DISCOS reference series as Statistics (1957–2025; payload, rocket body, debris and other), selected fragmentation events from ESA DISCOS and published literature, and GCAT context. Small-debris figures (about 1.2 million objects larger than 1 cm and about 140 million larger than 1 mm) are ESA MASTER-8 statistical model estimates, not tracked counts.

Satellite Imagery

Sources, dates and terms are listed under Earth imagery above. Tiles are displayed as delivered by each provider (opacity only), colour legends are reading aids, and OrbitSmith does not reprocess or archive the imagery.

MMX (unofficial mission view)

• NASA/JPL Horizons · Horizons API · data/mmx/manifest.json — One saved and checked snapshot of the trajectory JAXA provided to Horizons (MMX −239), with Earth (399) and Mars (499): Sun-centred positions in km on the ecliptic of J2000 (ICRF), geometric with no light-time correction, from API version 1.2. The current manifest lists trajectory revision 2026-10-20, downloaded 2026-10-22 03:05 UTC and verified by OrbitSmith at 03:09 UTC, covering launch (2026-10-19 19:41:03 UTC) to 2026-12-01 10:00:03 UTC. A 22 October revision that the provider described on 23 October is not yet incorporated.
• JAXA MMX mission page · JAXA launch result (20 Oct 2026) · JAXA press release (21 Oct 2026, Japanese) · JAXA MMX science page — Official pages checked on 2026-10-22 (UTC); later announcements are not included. Confirmed: launch on an H3 rocket from Tanegashima Space Center at 04:41:03 JST on 20 October 2026 (19:41:03 UTC on 19 October), and JAXA's 21 October report that initial critical operations were completed — no completion time was announced, so the page keeps the date only and labels any stand-in map time (noon JST). Planned and subject to change: instrument checks and cruise preparation (no dates announced), Mars arrival in 2027, close study of Phobos with a sample of at least 10 g and observation of Deimos, departure from the Mars system in 2030 and sample delivery to Earth in 2031. No sample has been collected or returned.
• ESA — Phobos (Mars Express HRSC) · ESA — MMX from launch to Phobos — The Phobos photo was taken on 23 July 2008 by the HRSC camera on ESA's Mars Express (ESA/DLR/FU Berlin (G. Neukum)), with brightness enhanced by ESA; it is not an MMX image. The spacecraft drawing is an ESA illustration of the plan (ESA, acknowledgement ATG Europe; published 2026-10-13), not a photo and not MMX's current orientation or location. Resized and cropped copies are offered under the same licence as the originals. CC BY-SA 3.0 IGO
• NASA PDS MGS MOLA · NASA SVS Blue Marble · Solar System Scope — Surface maps: Mars is MOLA elevation with OrbitSmith colour and hillshade, not a natural-colour photograph; Earth is historical NASA Blue Marble imagery; the Sun is an illustrative Solar System Scope map (CC BY 4.0). Orientation and lighting are explanatory, and each provider's terms apply. The size figure uses NASA Science values (Mars about 6,780 km; Phobos about 27 × 22 × 18 km; Deimos about 15 × 12 × 11 km) drawn as ellipses.

Positions exist only inside the snapshot's coverage: between saved samples they are joined by straight lines, gaps are never filled and nothing is extrapolated. For the part shown, the page states the largest difference (relative to Earth) between this interpolation and separately requested in-between values; the error of the original prediction is not known. Times before the tracking cutoff recorded in the snapshot are labelled estimates from tracking, later times predictions. Distance from Earth's centre and the one-way radio time (distance ÷ 299,792.458 km/s) are rounded to about two digits, which says nothing about accuracy; the radio animation is an educational simulation, not a signal or a communication record.

Bodies are drawn enlarged without moving their centres (a True size option and location rings are available), and the 2D view looks straight down on the same data, dropping height from the drawing only. Solid and dashed paths are predicted, not measured tracks; the ◆ marker shows location, not attitude; background stars are decoration. Close to Mars, Phobos is the shared shape model drawn 12–24 px wide at the position given by saved 5-minute Horizons vectors of Phobos (401) relative to Mars (499), satellite model mar099, joined linearly and omitted outside the saved period or where it would overlap Mars; its pose, lighting and size are explanatory. Plans from 2027 on get no position or path, only teaching figures.

Mission archive (Artemis II)

• JPL Horizons — Trajectory from the JPL Horizons post-mission archive (Horizons ID −1024), served through the OrbitSmith API in about 30-minute samples and smoothed for drawing. During the live mission the source was NASA's Orbital Ephemeris Message (OEM) published through AROW.
• NASA Artemis II Press Kit — mission timeline; several milestone times are marked approximate.

Distance and speed readouts are linear interpolations between archived records, or a coarse built-in table if the archive cannot be loaded; they are not spacecraft telemetry. In 3D, positions are scaled so the flyby distance fits the scene, Earth and Moon are enlarged (×3 and ×2.6) and placed from the trajectory geometry, and if position data are unavailable a schematic free-return curve is drawn instead.

Developers and public data

Developers · Public object directory · llms.txt · manifest.json

Developers explains AI-assisted lookups. The public object directory is a family of generated static HTML index pages and per-object JSON records built from one catalog snapshot; its snapshot time and record count are printed on the directory and change with each build, and absence from the directory means absence from that snapshot only. Read-only, keyless endpoints resolve a name or NORAD ID and calculate an SGP4 position (latitude, longitude, WGS84 ellipsoid height) at a requested UTC within ±7 days of the element epoch — a calculation, not an observation, and the limit is not an accuracy guarantee. Requests are rate-limited.

The manifest lists each static dataset — statistics, reentry watch, an aggregate-only CDM summary, the active catalog, GP/TLE and SATCAT shards — with its own publication time, size, SHA-256 hash and limitations. The catalog layer omits TLE lines; GP shards are element-set snapshots, not current positions. No file includes raw CDMs, state vectors or covariance.

Home and category hubs

Home · Orbit · Design · Labs · Explore

Home and the Orbit, Design, Labs and Explore overviews are navigation pages: they summarize and link the tools and perform no calculations of their own.

Articles, Contact and Privacy

Articles · Contact · Privacy Policy

Articles indexes commentary essays by Seita Iizuka published on note (Japanese originals, with note's English translation); they are opinion and analysis, not datasets. The Contact form opens a pre-filled email in your own mail app rather than submitting to a server. The Privacy Policy describes data handling, including which third-party providers your browser contacts.

Software libraries, fonts and decorative textures

Pages load open-source libraries (for example three.js, satellite.js, Chart.js, Leaflet and MapLibre GL JS) and web fonts from public CDNs or OrbitSmith's own assets; they are software, each under its own licence, not data sources. Decorative textures such as the Tracker's Blue Marble Earth and the star fields are illustrations, not current observations.