Sun
Elevated and volatile solar flux raises geomagnetic and radiation risk that propagates into power, communications and the upper atmosphere over the corridor.
Inputs in-distribution, members in agreement, interval tight — the Trust Core stands behind this number.
Stress basis: absolute flux (sfu): 94.5 sfu against a midpoint of 165 sfu.
Ensemble spread
Six heterogeneous members — a GRU, an MLP, a ridge autoregression, Holt smoothing, a seasonal baseline and persistence — each vote on the next value. Their weighted scatter relative to the recent one-step error is the disagreement signal; wide scatter means the members have genuinely split.
Trust signals
Spread of members vs expected one-step error.
Standardised distance of today's 5-feature input row from the training set. Furthest feature: Solar-cycle baseline (+0.76σ).
The combined trust the Core assigns this output.
What drove this prediction
Input×Gradient attribution: the gradient of the neural members' forecast with respect to every input, multiplied by the input, summed over the window — so each named driver carries a sign and a magnitude. The AR member's coefficients fold into the target's own history.
81-day trailing mean of F10.7
Flux one solar rotation ago (recurrent active regions)
The series' own recent values (autoregressive memory)
Yesterday → today difference
27-day trailing mean of F10.7
This stream, audited
7-step path
The ensemble is rolled forward recursively; each horizon carries its own conformal quantile from the calibration window, scaled by the adaptive factor, so the band widens honestly with lead time. Drivers are held at their last value and only the season advances.
Rolling backtest, last 90 steps
Every point is a real one-step forecast made from the data available at the time, against what the feed then reported. Misses outside the band are marked. Adaptive conformal inference nudges α each step toward the 90 % target — currently α=0.060 (band ×1.25 vs static).
Model card
trained 2026-10-02A heterogeneous ensemble: two neural networks trained by backpropagation on the archive, two classical forecasters, and two baselines every member must beat. Weights are set by validation error, never by the test window.
| Member | Kind | Weight | Test MAE* | Vote |
|---|---|---|---|---|
| GRU recurrent net2,185 params | neural · recurrent | 23% | 0.040 | 95.69 |
| MLP feed-forward net1,881 params | neural · feed-forward | 23% | 0.040 | 94.19 |
| AR(9) ridge | statistical | 21% | 0.041 | 94.04 |
| Holt smoothing | statistical | 17% | 0.066 | 91.53 |
| Seasonal naive | baseline | 4% | 0.142 | 111.00 |
| Persistence | baseline | 12% | 0.044 | 93.00 |
* held-out test window, log-transformed units
Space-weather context
Observed, not forecast: the latest NOAA readings the Sun model is anchored to, and the geomagnetic index that turns solar flux into risk on the ground.
Provenance
Governed quantity: Next-day observed 10.7 cm solar radio flux (the official 20:00 UT reading).
archive pulled 2026-10-02 09:52 UTC · live tail merged at request time · all sources publicly available · nothing on this page is simulated