How far to trust each component
Every component in one table, with what it is, what it was checked against, and the accuracy to expect.
The table
| Component | Method | Checked against | Expect |
|---|---|---|---|
| Galactic cosmic-ray protons | Force-field modulation of the Vos & Potgieter (2015) local interstellar spectrum, with the modulation potential φ on the Usoskin et al. (2017) scale. | AMS-02 proton flux, 2011 to 2013. | Within 10 % above 3 GV. 12 to 35 % high at 1 to 3 GV with the neutron-monitor φ; within 10 % if φ is refit to 670 MV. |
| Galactic cosmic-ray nuclei, Z = 2 to 28 | Force-field modulation of the Badhwar-O'Neill 2014 local interstellar spectra. | HELIOS oxygen at 52 to 83 MeV/n. | About 20 % high. |
| Cosmic rays against distance from the Sun | φ falls linearly with distance, calibrated to a 2.7 %/AU proton gradient at 1.7 GV. | Fine from 0.3 to 5 AU, coarse beyond. | |
| Anomalous cosmic rays | Oxygen fitted to HELIOS, 1975 to 1977; other elements by fixed ratios; intensity as the neutron-monitor rate to the 25th power; radial gradients of 49, 35, 13 and 4 %/AU. | The factor of about 500 between solar minimum and maximum in CREME96. | A factor 2 at solar minimum, 3 near maximum. The shape below 9 MeV/n is assumed. |
| Solar energetic particles | Band-function fits to the fluence of ground-level events; August 1972 from tabulated fluences; peak flux by scaling to the GOES peak above 10 MeV; heavy ions by gradual-event abundances. | GOES fluences above 10 and 30 MeV. | A factor 0.7 to 1.4 on fluence. Peak spectra ignore how the spectrum evolves during an event. |
| Trapped protons and electrons | AP8 and AE8 through IRBEM on their native field epochs, with the South Atlantic Anomaly drifted 0.3° per year; low-orbit protons above 15 MeV doubled. | The flight data compiled by Armstrong & Colborn (2000), ±25 %. | ±25 % above 15 MeV below 2000 km. A factor 2 to 3 below 15 MeV and in medium Earth orbit. |
| Trapped anomalous cosmic rays | One hundred times the interplanetary anomalous oxygen in a shell at L = 2 ± 0.12; nitrogen and neon at half their interplanetary ratios. | SAMPEX, 1992 to 1998. | A factor 3. The dependence on altitude and field strength is assumed. |
| Geomagnetic cutoff | Størmer cutoff, 14.5 GV / L², on IGRF McIlwain L; shifted 0.75° of latitude per Kp step; open above L = 5. | Smart & Shea trajectory-traced grids. | 10 to 20 % in cutoff rigidity. |
| Albedo neutrons | Kole et al. (2015) spectral shape; cutoff and solar-cycle dependence from a cosmic-ray yield integral fitted to OGO-6; averaged over the visible disk; inverse-square from 100 km. | Salyut-6, a balloon flight and the OGO-6 total leakage flux. | Within about 25 % of Salyut-6 and the balloon flight; about a third below the OGO-6 total leakage flux. |
| Albedo photons and charged secondaries | Formulas of Sazonov, Churazov, Mizuno and Abdo as compiled by Cumani et al. (2019). | A factor 2. The charged secondaries are the low-latitude set. | |
| Cosmic X-ray and gamma-ray background | The analytic fit of Gruber et al. (1999). | INTEGRAL and Swift-BAT, which sit 8 to 10 % higher. | About 10 %. |
| The quiet Sun and flare soft X-rays | A 5772 K blackbody scaled to 1361 W/m²; flares as an isothermal continuum with the temperature set by the GOES class. | Infrared and visible only for the quiet Sun. A factor 2 below 10 keV for flares. | |
| Flare hard X-rays | A power law at peak, with the flux at 35 keV and the index from the GOES class. | A factor 3. Measured from 15 to 100 keV and extrapolated to 10 MeV. | |
| Flare gamma-ray lines | The 2.223 MeV line from the GOES class; the 0.511 MeV line from the 2.223 MeV line; de-excitation lines as 2.223 / 0.4, split by Solar Maximum Mission averages. | The INTEGRAL/SPI fluence of 2003 October 28. | A factor 3 to 10 for any one flare. |
| Solar-flare neutrons | The Chupp (1990) emission spectrum of 1982 June 3, times 0.93 for 2003 October 28, with decay in flight. | The fluence above 50 MeV of 1982 June 3. | A factor 2 for these two events. Other flares are unknown. |
| Jovian electrons | 350 E⁻¹·⁶³ exp(−(E − 1)/9.4) electrons m⁻² s⁻¹ sr⁻¹ MeV⁻¹ at 1 AU, with a 13-month magnetic connection computed from the planets' positions. | The ISEE-3 spectral shape. | A factor 2 to 3 on average, with swings of a factor 10. |
| Solar wind and geostationary plasma | ECSS mean solar wind, falling as r⁻²; the ECSS worst-case bi-Maxwellian at geostationary orbit. | Typical values for the wind; a design worst case at geostationary orbit. | |
| Shielding | Geant4 11.4.1 with the QBBC physics list: a spherical aluminum shell under isotropic irradiation, 0.3 to 30 g/cm². | The MSL/RAD cruise dose, and SHIELDOSE-2 for protons. | See the validation results. |
| Dose from the Geant4 fold | Charged-particle fluence times stopping power, with the ICRP 60 quality factor; neutrons and photons as the mean dose in a water sphere of 15 cm radius. | MSL/RAD in cruise and the DOSIS dosimeters on the ISS. | A factor 2 to 3. The mean quality factor is low because target fragments made inside the body are left out. |
| Dose against depth (Python engine only) | SHIELDOSE-2 for electrons, bremsstrahlung and protons. | No heavy ions or neutrons, so deep-shielding cosmic-ray dose is underestimated. |
How to read "Expect"
- A percentage means the model agrees with independent measurements to about that much over the stated range.
- A factor means the truth may lie that many times above or below. A factor 3 is a real spread of nearly an order of magnitude from end to end.
- "Assumed" means the value is ours, chosen to be reasonable, with no measurement behind it.
- An empty "Checked against" means we have not compared the component with anything independent of the data it was fitted to.
The numbers behind this table are on the validation page. The component pages give the reasoning and the sources.
Why two engines agree
The models exist twice: in a Python reference engine and in the browser model that the explorers run. The reference engine is tested against measurements. The browser model is tested against the reference engine, to 5 % in free space and 6 % behind a wall. The data page describes the tables the browser model cannot compute for itself.