Analyse
Attitude state
Is the object stabilised, rotating or tumbling? A classification with a confidence index.
The Estimated attitude state section classifies the measured light curve, and the simulated one next to it, to show whether the model reproduces the observed state.
| Class | Meaning |
|---|---|
| Stabilised | Once the trend of the pass is removed, the variation stays at the noise level |
| Stabilised, with glints | Same, with one or two isolated flashes |
| Regular rotation | A clear period |
| Tumbling / non-periodic | A large variation without a clean period, two incommensurate periods, or a shape that changes along the series |
| Undetermined | Too few points, or a variation slower than the series |
Method
Section titled “Method”The slow trend of the pass (phase and range changing) is removed by a third-degree polynomial in time.
The noise comes from second differences between neighbours (or the given magnitude errors). Isolated flashes — at least 5σ and 0.5 mag brighter than their neighbours, 3 points at most — are set apart when there are one or two.
Lomb–Scargle periodogram, false-alarm probability of the best peak, share of the variance explained by a Fourier series with 2 to 6 harmonics.
A second significant, non-harmonic period in the residuals, or a model fitted on the first half of the series that fails to predict the second half, points to tumbling.
The confidence combines the false-alarm probability, the explained variance and the number of cycles covered.
How good is it?
Section titled “How good is it?”Measured with the synthetic dataset generator on 2 × 150 curves of 10 minutes at a 2 s step:
| True state | Recognised |
|---|---|
| Regular rotation | 86–89 % |
| Stabilised | 80–87 % |
| Tumbling | about 51 % |
| Overall | 71–77 % |
Over 10 minutes a tumbling body often looks like a near-regular rotation: longer series and several nights help.