Scans all tracked objects for predicted close approaches using orbital element proximity screening. Estimates miss distance, relative velocity, and collision probability for each conjunction.
Plan collision avoidance maneuvers using Clohessy-Wiltshire relative motion dynamics and Tsiolkovsky rocket equation. Estimates delta-v, fuel mass, and collision probability reduction.
Design a mega-constellation and assess its impact on collision risk, carrying capacity, and orbital sustainability.
Breakdown of cataloged objects by country of origin. Includes active payloads, rocket bodies, and debris attributed to each spacefaring nation.
Major debris-generating events from 1961 to present. Each event injected hundreds to thousands of trackable fragments into orbit, many of which remain today.
Orbital elements come from the 18th Space Defense Squadron GP catalog via the Space-Track.org API. The dataset covers all cataloged objects with perigee below 2000 km (LEO), updated at each fetch cycle. Only objects with valid TLE data are propagated; those that fail SGP4 initialization are excluded from display.
Positions are computed with the SGP4 propagator via satellite.js v4.1.4. SGP4 models J2 oblateness and atmospheric drag through the TLE B* coefficient. Higher-order zonal harmonics (J3, J4) and deep-space lunar/solar perturbations (SDP4 mode) are part of the underlying Vallado formulation but their practical impact on LEO objects over short propagation windows is small.
Vallado, D.A., Crawford, P., Hujsak, R., Kelso, T.S. (2006). “Revisiting Spacetrack Report #3.” AIAA/AAS Astrodynamics Specialist Conference, AIAA 2006-6753.
SGP4 outputs in the TEME frame. Ground-track coordinates are converted to geodetic via GMST. The 3D scene uses an ECI-aligned mapping:
Velocities use the vis-viva equation: v = √(μ(2/r − 1/a)) with μ = 398600.4418 km³/s². Specific energy ε = −μ/(2a), angular momentum h = √(μ a (1−e²)), and J2 nodal precession Ω˙ = −(3/2) n J2 (R⊕/a)² cos(i) / (1−e²)² follow standard formulations.
Lifetimes are approximated with a piecewise-linear function of perigee altitude, calibrated to King-Hele (1987) drag tables and ESA DRAMA lifetime estimates under moderate solar activity (F10.7 ≈ 130 SFU). Key anchor points: ~12 yr at 400 km, ~25 yr at 500 km, ~200 yr at 800 km. Actual lifetimes depend on solar activity, area-to-mass ratio, and orbital eccentricity, none of which are individually modeled here.
King-Hele, D.G. (1987). Satellite Orbits in an Atmosphere: Theory and Applications. Blackie & Son.
The ADR and policy simulations use a kinetic gas approximation for collision rate within 7 altitude bands (200–2000 km):
Fragments per catastrophic collision are set to 120 trackable pieces (>10 cm). This is on the conservative end of NASA Standard Breakup Model estimates; the Iridium-Cosmos collision (2009) produced roughly 2,000 cataloged fragments, but most LEO collisions involve smaller masses and lower energies. For comparison, Johnson et al. (2001) give the general scaling N(>Lc) ≈ 0.1 · M0.75, which yields ~100–400 trackable fragments depending on combined mass.
Rather than using a flat 10 km/s encounter speed, astral-risk computes vrel per altitude band from the actual inclination distribution of cataloged objects. For two objects on circular orbits at inclinations i1 and i2, the relative velocity at their orbital crossing point is:
where vcirc = √(μ/r) is the circular velocity at the band's midpoint altitude. Inclinations are binned into 5-degree buckets and the formula is evaluated for all unique bin pairs, weighted by the number of object pairs in each combination. For objects in the same inclination bin (where Δi ≈ 0), RAAN dispersion still produces crossing geometries; this is modeled as a base component of 15% of vcirc, added in quadrature.
This approach captures the dominant physical effect: retrograde-vs-prograde encounters (e.g., 28 deg vs 98 deg SSO) produce vrel ≈ 14–15 km/s, while co-planar objects in similar inclination bands see vrel ≈ 1–2 km/s. The net effect is that bands dominated by sun-synchronous orbits (~98 deg) interacting with lower-inclination objects show higher collision rates than a flat 10 km/s would predict, while bands with uniform inclination populations show lower rates.
Kessler, D.J. (1991). “Collisional Cascading: The Limits of Population Growth in Low Earth Orbit.” Advances in Space Research, 11(12), 63–66.
Johnson, N.L. et al. (2001). “NASA’s New Breakup Model of EVOLVE 4.0.” Advances in Space Research, 28(9), 1377–1384.
Liou, J.-C., Johnson, N.L. (2006). “Risks in Space from Orbiting Debris.” Science, 311(5759), 340–341.
The ADR model uses several assumed constants that significantly affect projections. These are not derived from a single source but represent reasonable order-of-magnitude estimates based on recent trends:
ESA Space Debris Office (2023). ESA’s Annual Space Environment Report. GEN-DB-LOG-00288-OPS-SD.
This tool uses a custom composite metric inspired by the debris prioritization literature (Pardini & Anselmo, 2020; Liou, 2011). It is not a standardized metric—it is defined here as:
Mass proxy is estimated from radar cross-section size category (SMALL ≈ 10 kg, MEDIUM ≈ 100 kg, LARGE ≈ 1000 kg). These are rough order-of-magnitude estimates; actual masses are not publicly available in the GP catalog. Lifetime and band density are computed as described above. The square root on lifetime prevents extremely long-lived but low-density objects from dominating the index.
The 0–100 normalization and the 25/50/75 thresholds are arbitrary display choices, not physically motivated cutoffs. Different weighting schemes would produce different rankings.
Pardini, C., Anselmo, L. (2020). “Environmental Sustainability of Large Satellite Constellations in Low Earth Orbit.” Acta Astronautica, 170, 27–36.
Liou, J.-C. (2011). “An Active Debris Removal Parametric Study for LEO Environment Remediation.” Advances in Space Research, 47(11), 1865–1876.
Collision energy is estimated for an equal-mass collision using the band-averaged relative velocity (computed from the inclination distribution) in the center-of-mass frame:
NASA classifies collisions as catastrophic when specific energy exceeds 40 J/g (the target is completely fragmented rather than just cratered). This threshold comes from hypervelocity impact experiments.
Fragment estimates use the NASA Standard Breakup Model scaling: N(>10 cm) ≈ 0.1 × Mtotal0.75. This is a simplification—the full SBM depends on collision energy, mass ratio, and whether the event is catastrophic or cratering.
Important caveat: The equal-mass assumption is a worst-case simplification. Most real collisions involve very unequal masses (e.g., a 1 cm debris fragment hitting a satellite), which would produce far less debris.
Johnson, N.L. et al. (2001). “NASA’s New Breakup Model of EVOLVE 4.0.” Advances in Space Research, 28(9), 1377–1384.
McKnight, D.S., Maher, R., Nagl, L. (2021). “Refined Identification of Statistically Most Concerning Derelict Objects in LEO.” Acta Astronautica, 184, 165–172.
The biggest simplification is using 7 discrete altitude bands instead of continuous modeling—this loses detail but keeps simulation fast enough to run in the browser. Collision cross-section is fixed at 10 m² regardless of object size. Mass is estimated from RCS category, which is coarse. There's no solar cycle modeling (decay rates assume moderate activity), and fragment count per collision is constant rather than mass/velocity-dependent. Launch rate and PMD compliance stay constant over the whole projection.
Because of all this, the absolute numbers shouldn't be read as predictions. The tool is more useful for comparing relative outcomes (e.g., 60 removals/yr vs. none) than for forecasting exact population counts.
Expected Annual Loss (EAL) per altitude band is estimated as:
Active satellite fractions and replacement costs are band-specific estimates: 400–600 km is weighted highest (60% active, $250K avg) reflecting Starlink-class constellations. Higher orbits carry higher per-satellite values ($1M–$10M) reflecting science and communications missions. These are order-of-magnitude assumptions—actual satellite values vary by orders of magnitude within each band. The cost-benefit analysis assumes $25M/removal as default (based on published Astroscale/ClearSpace-1 program estimates) and computes ROI = damage_prevented / program_cost.
Astroscale Holdings (2024). “ELSA-d End-of-Life Services Demonstration Mission Results.”
The maximum sustainable population Kmax per altitude band is derived from the equilibrium condition where collision fragment generation equals natural decay:
This approach is inspired by the “orbital carrying capacity” concept developed in the MIT MOCAT framework (Lifson et al., 2022). The simplified analytic form used here ignores launch input and ADR, providing a zeroth-order stability boundary. Band utilization = Ncurrent / Kmax indicates how close each altitude shell is to self-sustaining collision cascading.
Lifson, M., Ramos, R.P., Wilson, R.S. (2022). “MOCAT-MC: A Monte Carlo Approach to Debris Evolution Modeling.” MIT Astrodynamics, Space Robotics, and Controls Lab. See also: Servadio, S. et al. (2023), “Composition of the Future LEO Environment,” J. Space Safety Eng., 10(4).
The cascade timeline runs the discrete simulation at ADR rates of 0, 20, 40, and 60 objects/year for 100 years and tracks the per-band ratio of collision-generated fragments to natural decay. Phase classification: Stable (fragments < 50% of decay), Approaching (50–100%), Cascade Active (100–200%), Runaway (> 200%). This provides a visual indicator of when each orbital band becomes self-sustaining in debris growth under different remediation scenarios.
The breakup simulation uses the NASA Standard Breakup Model scaling law N(>Lc) = 0.1 × Mtotal0.75 for trackable fragments (>10 cm), with an empirical 8.5× multiplier for lethal non-trackable fragments (>1 cm) based on cross-section-weighted hazard estimates (the true count ratio N(>1cm)/N(>10cm) is substantially higher, on the order of 100–300×, but most sub-cm fragments are non-lethal). Altitude spread is estimated from Δa ≈ 2aΔv/vcirc. Delta collision probability uses the same kinetic gas model as the ADR simulation. This is a first-order estimate; actual fragment distributions depend on collision geometry, material composition, and energy partitioning.
Country scores are a weighted composite: 30% active satellite ratio, 30% inverse mean ECI, 40% estimated 25-year guideline compliance. Important caveat: the compliance estimate is a weak proxy based on natural orbital decay lifetime, not actual deorbit intent or execution. A satellite in a naturally short-lived orbit appears “compliant” even without active disposal, while a satellite with a planned deorbit maneuver in a long-lived orbit appears non-compliant. This metric is purely illustrative and should not be used for regulatory assessment.
3D visualization uses Three.js r128 with InstancedMesh for rendering ~20,000+ objects. Each type (satellite, rocket body, debris) uses a distinct 3D geometry. The Milky Way background is from the ESO/S. Brunier panoramic survey.
Astral Risk is an AI-powered educational tool for exploring the orbital debris environment. It provides an interactive way to test assumptions about whether Low Earth Orbit is approaching a tipping point—and how different policy choices affect the debris trajectory. Unlike validated engineering models like NASA's ORDEM or ESA's MASTER, this tool is designed for accessibility: change the removal rate or toggle a compliance rule and immediately see what happens to collision probability in each altitude band.
The tool connects SGP4 propagation, collision probability modeling, and the NASA Standard Breakup Model with economic risk quantification, carrying capacity, and policy simulation in one interface. One of the most notable findings from the model is how sensitive the cascade timeline is to small changes in post-mission disposal compliance—a shift from 60% to 90% PMD compliance delays the Kessler threshold by decades in some altitude bands.
Objects are not drawn to scale. The dots representing satellites, rocket bodies, and debris in the 3D view are magnified by roughly 10,000–100,000× their actual size relative to Earth. A typical satellite is 1–10 meters across; at true scale it would be completely invisible against a 12,742 km diameter Earth. Debris fragments (<10 cm) would be smaller still. Orbital altitudes are approximately to scale—LEO objects at 400 km are shown at ~6% above Earth’s surface, which is geometrically correct. The exaggerated object size is necessary to make the debris environment visible and interactive.
Drag to rotate, scroll to zoom. Click any object to see its orbit, collision risk, and ECI score. The left panel has filters for type, constellation, country, altitude, inclination, and period. Keys 1–5 switch color modes. Press ? for all shortcuts.
Astral Risk connects debris environment modeling with policy analysis. It propagates the full GP catalog in real time, computes collision risk and environmental criticality per object, and simulates the long-term effects of different mitigation strategies.
When I started this project, I looked at what already exists. ORDEM and MASTER are the gold standard—validated engineering models with decades of flight data calibration. This tool doesn't compete with them on prediction accuracy. What it adds is the policy analysis layer: economic risk quantification, carrying capacity gauges, and multi-scenario cascade modeling, all in a browser anyone can access.
| This tool | NASA ORDEM | ESA MASTER | LeoLabs | STK | |
|---|---|---|---|---|---|
| Free, browser-based | Yes | No | No | Freemium | No |
| Live 3D full catalog | Yes | No | No | Yes | Yes |
| Economic risk ($) | Yes | No | No | No | No |
| Carrying capacity | Yes | No | No | No | No |
| Policy comparison | Yes | No | No | No | Limited |
| ADR cost-benefit | Yes | No | No | No | No |
| Conjunction alerts | Yes | No | No | Yes | Yes |
| Constellation planner | Yes | No | No | No | No |
| Deep space mission tracker | Yes | No | No | No | No |
| Validated model | No | Yes | Yes | Yes | Yes |
This is an educational tool, not an operational model. The collision probability uses a kinetic gas approximation (not conjunction screening), the simulation uses 7 discrete altitude bands instead of continuous modeling, and absolute population numbers carry real uncertainty. It's most useful for comparing relative outcomes—e.g., how different policies or removal rates change the debris trajectory. Full assumptions and citations are in the Methodology panel.
Low Earth Orbit (LEO) is the region of space from roughly 160 km to 2,000 km above Earth’s surface. It is the most heavily used orbital regime: the International Space Station orbits at ~420 km, most Earth observation satellites operate between 500–900 km, and large constellations like Starlink populate the 340–570 km range.
Objects in LEO travel at approximately 7.5–7.8 km/s (about 27,000 km/h). At these velocities, even a 1 cm fragment carries the kinetic energy of a hand grenade. A 10 cm fragment can destroy a spacecraft entirely.
LEO is attractive because it offers low-latency communications, high-resolution imaging, and relatively low launch costs. But these same advantages have made it increasingly congested: as of 2025, the U.S. Space Surveillance Network tracks over 30,000 objects larger than 10 cm in LEO, and statistical models estimate over 1 million fragments larger than 1 cm.
In 1978, NASA scientist Donald Kessler proposed that beyond a critical density of objects in orbit, collisions would produce fragments faster than atmospheric drag can remove them. Each collision creates a cloud of debris, which increases the probability of further collisions, creating a self-sustaining chain reaction known as a collisional cascade.
The Kessler Syndrome does not mean space becomes “impassable” overnight. It describes a long-term runaway process that unfolds over decades to centuries. The danger is that once the cascade begins in a particular altitude band, it becomes effectively irreversible on human timescales—the debris remains in orbit for hundreds of years at altitudes above ~800 km where atmospheric drag is negligible.
Key factors that determine whether a cascade occurs:
Collision avoidance: The ISS performs ~3 avoidance maneuvers per year. Satellite operators must routinely dodge debris, consuming propellant and reducing mission lifetime. SpaceX’s Starlink constellation executes thousands of avoidance maneuvers annually.
Mission risk: Even a 1 cm fragment impact can disable a satellite. The 2009 Iridium–Cosmos collision created over 2,000 trackable fragments, many of which remain in orbit today. China’s 2007 ASAT test on the Fengyun-1C satellite created over 3,500 trackable pieces—the single largest debris-generating event in history.
Economic impact: LEO hosts an estimated $1.5+ trillion in active satellite infrastructure. Collision damage is uninsurable in most cases. A single catastrophic collision can generate enough fragments to threaten dozens of other assets in the same altitude band.
Long-term access: If key altitude bands (particularly 700–1,000 km) enter a cascade, they could become unusable for decades. This would affect weather forecasting, climate monitoring, GPS augmentation, and scientific research.
Post-Mission Disposal (PMD): Satellites should deorbit within 5–25 years of mission end (depending on the regulatory framework). The FCC now requires 5 years for U.S.-licensed satellites. Current global compliance is estimated at ~78%.
Active Debris Removal (ADR): Physically removing defunct objects from orbit using robotic capture, nets, harpoons, or laser ablation. Studies suggest removing 5–10 large objects per year could stabilize the most critical altitude bands. This tool lets you simulate removal rates of 0–150 objects/year.
Debris mitigation: Designing spacecraft to minimize breakup risk—passivation of batteries and fuel tanks, avoidance of mission-related debris, and collision avoidance maneuvering.
Space traffic management: Improved tracking, conjunction screening, and international coordination to reduce collision probability.
Astral Risk uses real orbital data from the U.S. Space Force’s GP catalog (updated regularly via Space-Track.org) and propagates all 20,000+ tracked objects forward in time using the SGP4 algorithm. It then applies several analytical models:
Important caveats: This is an educational tool, not an operational model. It uses simplified physics (7 discrete altitude bands, fixed cross-section, no solar cycle). Absolute numbers should be treated as illustrative. The tool is most useful for comparing relative outcomes—what happens when you change one policy variable while holding others constant.